{"meta":{"query_hash":"122d1dfedb94","filters":{"venue":"Bioinformatics"},"cohort_total":946,"direct_labels_cover":0,"predictions_cover":946,"exported":946,"export_cap":100000,"truncated":false,"label_status":"direct model label, unvalidated","prediction_status":"machine_predicted_unvalidated (Codex and Gemma teacher distillation)","score_status":"score_only:v0-immature-baseline","snapshot":{"source":"OpenAlex, pinned release, all 482 partitions","release":"2026-06-24","frame_built":"2026-07-12"},"permalink":"https://metacan.xera.ac/q/122d1dfedb94","api":"https://metacan.xera.ac/api/v1/cohort?venue=Bioinformatics"},"results":[{"id":"W1793459287","doi":"10.1093/bioinformatics/btv420","title":"CEMAsuite: open source degenerate PCR primer design","year":2015,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Molecular Biology Techniques and Applications","field":"Biochemistry, Genetics and Molecular Biology","cited_by":16,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Division of Chemical, Bioengineering, Environmental, and Transport Systems; Egg Farmers of Canada; Louisiana State University; National Science Foundation","keywords":"Primer (cosmetics); Suite; Computer science; Source code; Computational biology; Sequence (biology); Coding (social sciences); Multiple sequence alignment; Sequence alignment; Biology; Genetics; Programming language; Gene; Peptide sequence; Mathematics; Statistics","score_opus":0.053797628234244284,"score_gpt":0.2991767766964953,"score_spread":0.245379148462251,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1793459287","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0045855097,0.0012739708,0.8552584,0.00031331534,0.000803622,0.0022051532,0.03269234,0.0918114,0.01105627],"genre_scores_gemma":[0.012274556,0.00048402988,0.90705657,0.0005414579,0.00011231896,0.006871236,0.035940792,0.023977745,0.012741366],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9937006,0.00189713,0.0006214325,0.0019220472,0.0013274938,0.00053127785],"domain_scores_gemma":[0.9967552,0.0014863835,0.0003751228,0.00057293626,0.0005868224,0.00022349952],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0058351457,0.004165983,0.0036153775,0.0044966335,0.0017287238,0.0027682856,0.0052550733,0.0023003698,0.105804734],"category_scores_gemma":[0.0115095,0.0039243256,0.0028635715,0.002211276,0.0016937379,0.0018162869,0.0024743162,0.005459417,0.08298575],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.003643106,0.00058293174,0.0022061456,0.005932652,0.0004133567,0.0011915026,0.00088928314,0.004512095,0.2161429,0.02336626,0.5039901,0.2371296],"study_design_scores_gemma":[0.00071705575,0.00064224197,0.0019376263,0.0006417589,0.00021983805,0.0016138272,0.00015396293,0.032707136,0.2404599,0.015773483,0.70462435,0.0005087243],"about_ca_topic_score_codex":0.00090669096,"about_ca_topic_score_gemma":0.002705505,"teacher_disagreement_score":0.105804734,"about_ca_system_score_codex":0.00106526,"about_ca_system_score_gemma":0.0025123237,"threshold_uncertainty_score":0.3539521},"labels":[],"label_agreement":null},{"id":"W1889881724","doi":"10.1093/bioinformatics/btv385","title":"Hyperscape: visualization for complex biological networks","year":2015,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto; Hospital for Sick Children","funders":"Centre National de la Recherche Scientifique; Heart and Stroke Foundation of Canada","keywords":"Visualization; Computer science; Biological network; Computational biology; Artificial intelligence; Biology","score_opus":0.052094076985176846,"score_gpt":0.2850710021989206,"score_spread":0.23297692521374377,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1889881724","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.007750341,0.0010579164,0.6277607,0.0016354138,0.00036366712,0.00038817155,0.05342992,0.29609695,0.011516834],"genre_scores_gemma":[0.10188748,0.0026083135,0.7292573,0.0011131827,0.00024700726,0.0021863438,0.100962624,0.05009397,0.011643725],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9993309,0.00018845155,0.0000546678,0.00012417103,0.00025030118,0.000051568943],"domain_scores_gemma":[0.99706715,0.0018058029,0.00014137742,0.0003837355,0.00039598375,0.00020597265],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016716437,0.0020940048,0.00089197623,0.0033087817,0.0007743019,0.0027861625,0.0025982687,0.0012642829,0.07444757],"category_scores_gemma":[0.0067682634,0.0009692047,0.001397275,0.0022992974,0.0006427241,0.0034765617,0.0035541318,0.0021693541,0.011406799],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00084238464,0.00015533637,0.0026690091,0.002568207,0.00038366934,0.0009441987,0.001066466,0.039798304,0.019283554,0.04823534,0.6724742,0.21157932],"study_design_scores_gemma":[0.00039236483,0.00009885781,0.0032937604,0.00069558836,0.00008964826,0.0009578841,0.00025666648,0.432452,0.024960035,0.14014119,0.39640042,0.0002616404],"about_ca_topic_score_codex":0.0040800227,"about_ca_topic_score_gemma":0.004328659,"teacher_disagreement_score":0.07444757,"about_ca_system_score_codex":0.00070296216,"about_ca_system_score_gemma":0.0010852665,"threshold_uncertainty_score":0.24905193},"labels":[],"label_agreement":null},{"id":"W1891158457","doi":"10.1093/bioinformatics/btv615","title":"funtooNorm: an R package for normalization of DNA methylation data when there are multiple cell or tissue types","year":2015,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Epigenetics and DNA Methylation","field":"Biochemistry, Genetics and Molecular Biology","cited_by":26,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Child and Family Research Institute; University of British Columbia; Douglas Mental Health University Institute; McGill University; Centre Intégré Universitaire de Santé et de Services Sociaux du Saguenay–Lac-Saint-Jean; Université de Sherbrooke; McGill University Health Centre; Canadian Institute for Advanced Research; Jewish General Hospital","funders":"Canadian Institutes of Health Research","keywords":"Normalization (sociology); DNA methylation; DNA; Computer science; Computational biology; Methylation; R package; Biology; Genetics; Programming language; Gene; Gene expression","score_opus":0.07610816683406356,"score_gpt":0.31027202298495143,"score_spread":0.23416385615088786,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1891158457","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0039044397,0.0009243086,0.6965391,0.0007677691,0.0005129409,0.00037975938,0.06566583,0.2263356,0.0049702693],"genre_scores_gemma":[0.029748451,0.00089340564,0.6754365,0.0012570221,0.00025029894,0.003458379,0.092999525,0.18610711,0.009849258],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99423814,0.0021992,0.00040582143,0.0014957213,0.0013688497,0.00029223337],"domain_scores_gemma":[0.9875415,0.007552444,0.0008991876,0.0021411672,0.0015435964,0.00032211698],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008815595,0.0036225235,0.0027295635,0.003021551,0.0011321801,0.0029112895,0.003930769,0.0013905427,0.05586577],"category_scores_gemma":[0.042329,0.0020713678,0.003221671,0.0032279585,0.001559679,0.0023197522,0.0039264266,0.0039904164,0.05159553],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009972617,0.00009320285,0.009729269,0.0034844908,0.0018287963,0.0005030984,0.00066530355,0.01580827,0.020354262,0.013610033,0.7356547,0.19727138],"study_design_scores_gemma":[0.0005622489,0.00027245958,0.0139473975,0.0006270629,0.00087881647,0.0015434803,0.00014000542,0.106405266,0.057687867,0.076255634,0.7410827,0.0005970236],"about_ca_topic_score_codex":0.003206544,"about_ca_topic_score_gemma":0.004595292,"teacher_disagreement_score":0.05586577,"about_ca_system_score_codex":0.0010046762,"about_ca_system_score_gemma":0.0035958823,"threshold_uncertainty_score":0.18688959},"labels":[],"label_agreement":null},{"id":"W1905939051","doi":"10.1093/bioinformatics/btv277","title":"<i>pez</i>: phylogenetics for the environmental sciences","year":2015,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Evolution and Paleontology Studies","field":"Earth and Planetary Sciences","cited_by":192,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"The Scarborough Hospital; McMaster University; University of Toronto; McGill University; Université du Québec à Montréal","funders":"","keywords":"R package; License; Phylogenetic tree; Computer science; MIT License; Open source; The Internet; Data science; Phylogenetics; World Wide Web; Software; Biology; Programming language; Operating system","score_opus":0.059458433312916684,"score_gpt":0.24322307357551956,"score_spread":0.18376464026260286,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1905939051","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0025721246,0.005312019,0.4390107,0.02507814,0.009408714,0.00029700156,0.28921783,0.14239144,0.08671203],"genre_scores_gemma":[0.04531757,0.0060045207,0.35719383,0.0074306866,0.0060020504,0.0018702208,0.35399717,0.16931862,0.05286528],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99595237,0.0014151003,0.00029734205,0.0009782917,0.0011163581,0.00024050254],"domain_scores_gemma":[0.9780115,0.01175083,0.0018371311,0.0038099547,0.0026595793,0.0019310728],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.005829508,0.0018235858,0.002147089,0.003951053,0.0020364618,0.00402851,0.004673551,0.0024835605,0.31243303],"category_scores_gemma":[0.042836502,0.0011333288,0.0012673753,0.008341843,0.0019878668,0.0048733708,0.0052624596,0.0054103197,0.2325898],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000105005674,0.000012587283,0.0007946173,0.0014447268,0.00006533115,0.00007559648,0.00012748715,0.0011965077,0.0013522626,0.02401381,0.9162595,0.054552477],"study_design_scores_gemma":[0.00005554276,0.000022362898,0.0018156122,0.00044555805,0.000040831972,0.0002909228,0.00003052917,0.0037214847,0.0015556815,0.028964562,0.96299356,0.00006331892],"about_ca_topic_score_codex":0.0021609983,"about_ca_topic_score_gemma":0.002850639,"teacher_disagreement_score":0.31243303,"about_ca_system_score_codex":0.0010379604,"about_ca_system_score_gemma":0.0029991644,"threshold_uncertainty_score":0.98073083},"labels":[],"label_agreement":null},{"id":"W1970806484","doi":"10.1093/bioinformatics/btv107","title":"The GPMDB REST interface","year":2015,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Advanced Proteomics Techniques and Applications","field":"Chemistry","cited_by":32,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Manitoba","funders":"University of Manitoba","keywords":"Rest (music); Computer science; Interface (matter); Documentation; File Transfer Protocol; MIT License; Representational state transfer; Service (business); Software; Database; World Wide Web; Programming language; The Internet; Operating system; Web service","score_opus":0.027811404180778206,"score_gpt":0.2976196948278157,"score_spread":0.2698082906470375,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1970806484","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0014555122,0.00042936136,0.09696678,0.00069412746,0.000372164,0.0003102572,0.16462307,0.7038143,0.031334434],"genre_scores_gemma":[0.020485297,0.0006037749,0.06302245,0.0028851226,0.00026690363,0.0014909998,0.6767924,0.1934657,0.04098736],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9978163,0.00035942096,0.00019516888,0.00041468462,0.0009135093,0.00030089426],"domain_scores_gemma":[0.9968162,0.00074694323,0.0001472322,0.0012073563,0.00076410716,0.00031821575],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003235645,0.0031795518,0.0023422719,0.002773614,0.0012136995,0.0048866156,0.007891678,0.0026564442,0.23970816],"category_scores_gemma":[0.009443753,0.0019983049,0.0023785955,0.0027453967,0.0006390361,0.004509724,0.0073056445,0.0039416985,0.27493194],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00051857234,0.000051545638,0.00049631915,0.00036099306,0.00005025425,0.00013443065,0.000084721796,0.00046922613,0.0014981937,0.003150453,0.9657172,0.027468137],"study_design_scores_gemma":[0.00041409692,0.000049730552,0.001938755,0.00031458115,0.00005596216,0.0003759855,0.00006938549,0.015974302,0.0103397155,0.016875274,0.953435,0.00015721236],"about_ca_topic_score_codex":0.008478731,"about_ca_topic_score_gemma":0.005675814,"teacher_disagreement_score":0.23970816,"about_ca_system_score_codex":0.0019856242,"about_ca_system_score_gemma":0.0018659224,"threshold_uncertainty_score":0.80190367},"labels":[],"label_agreement":null},{"id":"W1971987634","doi":"10.1093/bioinformatics/bts219","title":"SEQuel: improving the accuracy of genome assemblies","year":2012,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":70,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Center for Research Resources; Natural Sciences and Engineering Research Council of Canada; University of California, San Diego; National Institutes of Health; National Science Foundation","keywords":"Contig; De Bruijn graph; Sequence assembly; De Bruijn sequence; Indel; Hybrid genome assembly; Substitution (logic); Computer science; Genome; Reference genome; Computational biology; Error detection and correction; Algorithm; Graph; Biology; Genetics; Theoretical computer science; Gene; Mathematics; Combinatorics; Programming language","score_opus":0.019702669294599684,"score_gpt":0.24813360594177075,"score_spread":0.22843093664717107,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1971987634","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0863956,0.0014067645,0.8300637,0.0007814904,0.0005627399,0.00023295698,0.009150884,0.06550537,0.0059004566],"genre_scores_gemma":[0.13710332,0.0007238865,0.8299888,0.00039983404,0.00015306455,0.00021378646,0.017598195,0.010319982,0.0034991028],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99282324,0.0023659805,0.0007436702,0.001724123,0.0019312982,0.0004117198],"domain_scores_gemma":[0.9700613,0.00982157,0.0019319659,0.009626286,0.008084626,0.0004741904],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0107642785,0.0021897764,0.0019544638,0.0024940406,0.0014521651,0.0023196142,0.003039063,0.0016540209,0.00810098],"category_scores_gemma":[0.048856426,0.0013347378,0.001807549,0.002603091,0.0007625017,0.00354872,0.0035025296,0.0026398594,0.008283638],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00147119,0.00022089,0.021463245,0.0023213893,0.0004906633,0.0010464095,0.0013292425,0.16442503,0.23522322,0.013933036,0.08656899,0.4715067],"study_design_scores_gemma":[0.00018284444,0.00048292644,0.009511195,0.0003853511,0.00035574427,0.0012232198,0.00036972068,0.49450636,0.38392314,0.009308298,0.09947323,0.00027787677],"about_ca_topic_score_codex":0.0048073493,"about_ca_topic_score_gemma":0.0051674736,"teacher_disagreement_score":0.0107642785,"about_ca_system_score_codex":0.0012683156,"about_ca_system_score_gemma":0.0019993873,"threshold_uncertainty_score":0.05692768},"labels":[],"label_agreement":null},{"id":"W1976496804","doi":"10.1093/bioinformatics/btv163","title":"Learning chromatin states with factorized information criteria","year":2015,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":15,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Institute of Genetics; Ministry of Education, Culture, Sports, Science and Technology; Japan Society for the Promotion of Science; University of Tokyo; Waseda University; Research Organization of Information and Systems","keywords":"Chromatin; Computer science; Computational biology; Nucleosome; Epigenome; Epigenetics; Genome; Bayesian probability; ChIP-sequencing; Epigenomics; Biology; Genetics; Artificial intelligence; DNA methylation; DNA; Gene","score_opus":0.008461581363955289,"score_gpt":0.22402182272729004,"score_spread":0.21556024136333476,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1976496804","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.015222076,0.00038845473,0.98269504,0.00017440914,0.000017936698,0.00007222467,0.00026107935,0.00073183054,0.00043702117],"genre_scores_gemma":[0.54489213,0.00039709485,0.4500105,0.0002994685,0.00019447232,0.0004081724,0.0024060386,0.00024938164,0.0011426939],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99727184,0.0010760842,0.00016597759,0.0008157005,0.00048442493,0.00018592011],"domain_scores_gemma":[0.9861562,0.011148166,0.00072760193,0.00052190095,0.0012210911,0.00022498565],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003967296,0.0014668278,0.0019362341,0.0033849892,0.0007070933,0.0014955677,0.0019818812,0.0018716779,0.0020609195],"category_scores_gemma":[0.017032595,0.0007241795,0.0015909236,0.0017234301,0.0015016405,0.001772459,0.0012768132,0.0015746467,0.00045708974],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00025302538,0.0001278341,0.005533235,0.00023877929,0.00016544572,0.00017203567,0.00016335252,0.82311106,0.0023148316,0.0147077525,0.002556537,0.15065613],"study_design_scores_gemma":[0.000012449897,0.000023568211,0.00026718114,0.000015804488,0.000009546012,0.000023581812,0.0000062694794,0.9866736,0.000573075,0.012090694,0.00029344167,0.0000107797705],"about_ca_topic_score_codex":0.008827306,"about_ca_topic_score_gemma":0.004804189,"teacher_disagreement_score":0.008827306,"about_ca_system_score_codex":0.0016521928,"about_ca_system_score_gemma":0.0015195735,"threshold_uncertainty_score":0.020981312},"labels":[],"label_agreement":null},{"id":"W1978377575","doi":"10.1093/bioinformatics/btp280","title":"ISMB/ECCB 2009 Stockholm","year":2009,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetics, Bioinformatics, and Biomedical Research","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Computer science; Library science","score_opus":0.016323508100717782,"score_gpt":0.28009558421644304,"score_spread":0.2637720761157253,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1978377575","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0044520074,0.018135943,0.012157684,0.02083894,0.026756583,0.00038332707,0.029902916,0.007678684,0.8796938],"genre_scores_gemma":[0.014767942,0.008626182,0.012771457,0.003122689,0.0028695632,0.0003928245,0.034590855,0.005111224,0.9177473],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99723476,0.0002875959,0.00016047736,0.0007658845,0.0012414153,0.0003098094],"domain_scores_gemma":[0.99733704,0.000176016,0.00010937068,0.000622941,0.0011455485,0.0006090484],"candidate_categories":["insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.002750311,0.0017175226,0.0011829783,0.0023707796,0.0017127303,0.008800018,0.002302865,0.003073914,0.37075344],"category_scores_gemma":[0.0032348374,0.00063885463,0.00097016274,0.0018987925,0.00076747406,0.002705317,0.0048408615,0.003271432,0.40523937],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00008603687,0.000051114876,0.00012266629,0.00013559712,0.000009326802,0.000076768454,0.000016794602,0.00038714008,0.0006746652,0.009153636,0.927706,0.06158024],"study_design_scores_gemma":[0.000015744803,0.000013456433,0.00036467632,0.00009690028,0.0000043772425,0.000073816824,0.000013762268,0.00024179381,0.0004999157,0.0022267133,0.99643755,0.000011313194],"about_ca_topic_score_codex":0.0050593717,"about_ca_topic_score_gemma":0.0036859803,"teacher_disagreement_score":0.6292466,"about_ca_system_score_codex":0.0029977404,"about_ca_system_score_gemma":0.003856149,"threshold_uncertainty_score":0.89754385},"labels":[],"label_agreement":null},{"id":"W1981576666","doi":"10.1093/bioinformatics/btm223","title":"Efficient parameter estimation for RNA secondary structure prediction","year":2007,"lang":"en","type":"article","venue":"Bioinformatics","topic":"RNA and protein synthesis mechanisms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":248,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"National Institutes of Health; National Institute of General Medical Sciences; Natural Sciences and Engineering Research Council of Canada; Mitacs","keywords":"Computer science; Energy minimization; Energy (signal processing); Algorithm; Constraint (computer-aided design); Minification; Estimation theory; Mathematical optimization; Mathematics; Statistics","score_opus":0.008428418146540894,"score_gpt":0.23505822218269098,"score_spread":0.2266298040361501,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1981576666","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.016362172,0.00011366509,0.98077637,0.000119677716,0.000011720272,0.000044977434,0.00019593023,0.0017902795,0.00058508833],"genre_scores_gemma":[0.278198,0.00016481557,0.71824294,0.0001261823,0.000032601085,0.00031876742,0.0015201437,0.0005644409,0.00083208235],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99928576,0.0003026524,0.000034201094,0.00011687957,0.00021744939,0.00004318573],"domain_scores_gemma":[0.9959311,0.0028839977,0.0002833606,0.00043062863,0.00040117226,0.000069755115],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013216385,0.0010673775,0.00079002103,0.0009049873,0.00048113806,0.00057753106,0.001620174,0.0011394428,0.0026239043],"category_scores_gemma":[0.009361004,0.0005715022,0.0005526946,0.0011446495,0.0006998569,0.0012235454,0.0009783198,0.0013533167,0.0011142598],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00010545405,0.000052331445,0.0014167528,0.0000920299,0.00003980153,0.00006453351,0.000042430616,0.86961997,0.008159317,0.0052534337,0.0024759676,0.112678014],"study_design_scores_gemma":[0.00000913538,0.000006901524,0.00011995785,0.0000037369553,0.0000022782306,0.000013404355,0.0000026442813,0.99304706,0.0024674803,0.004002233,0.00031892047,0.000006233949],"about_ca_topic_score_codex":0.0057974155,"about_ca_topic_score_gemma":0.0056012836,"teacher_disagreement_score":0.0057974155,"about_ca_system_score_codex":0.00077887304,"about_ca_system_score_gemma":0.0017274265,"threshold_uncertainty_score":0.0115273595},"labels":[],"label_agreement":null},{"id":"W1993204833","doi":"10.1093/bioinformatics/btr342","title":"genBlastG: using BLAST searches to build homologous gene models","year":2011,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":121,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Michael Smith Health Research BC; Simon Fraser University; University of British Columbia; BC Cancer Agency","funders":"","keywords":"Homologous chromosome; Computer science; Computational biology; Homologous recombination; Gene; Genetics; Biology","score_opus":0.0691356794700421,"score_gpt":0.2522758472029456,"score_spread":0.1831401677329035,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1993204833","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.055955265,0.0076435474,0.5519604,0.002397483,0.00085657113,0.0015971845,0.10929732,0.25044033,0.019851819],"genre_scores_gemma":[0.044843648,0.003709867,0.7820605,0.00047661355,0.00011796726,0.0013031848,0.14511587,0.019287467,0.0030848426],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9985435,0.00045501796,0.0001407283,0.00043349373,0.00032527724,0.00010201574],"domain_scores_gemma":[0.9989753,0.0004813969,0.00012203229,0.00018600374,0.00013426074,0.000101080564],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0032226802,0.0037769563,0.0031191022,0.0041358685,0.0019065875,0.0027191278,0.003671736,0.0020311493,0.018132376],"category_scores_gemma":[0.006065586,0.0017702188,0.0012897071,0.0052089775,0.0008242771,0.0026031036,0.0025203365,0.0029296207,0.024020929],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.005575092,0.0005430016,0.007489875,0.009466103,0.00091072347,0.0020676055,0.0009943687,0.010166441,0.16780113,0.015580944,0.53139293,0.24801184],"study_design_scores_gemma":[0.002565395,0.0011967298,0.010186334,0.0013426649,0.0008184071,0.0074912086,0.0007129752,0.12837796,0.18245631,0.053717937,0.6105329,0.00060123275],"about_ca_topic_score_codex":0.00094961334,"about_ca_topic_score_gemma":0.0010709199,"teacher_disagreement_score":0.018132376,"about_ca_system_score_codex":0.0007993006,"about_ca_system_score_gemma":0.0013073618,"threshold_uncertainty_score":0.060658872},"labels":[],"label_agreement":null},{"id":"W1993730176","doi":"10.1093/bioinformatics/18.12.1633","title":"Statistical analysis of high-density oligonucleotidearrays: a multiplicative noise model","year":2002,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":86,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Centre hospitalier de l'Université Laval","funders":"National Institute of Allergy and Infectious Diseases; Center for AIDS Research, University of Washington","keywords":"Multiplicative function; Computer science; Statistical analysis; Noise (video); Multiplicative noise; Statistical model; Oligonucleotide; Algorithm; Computational biology; Statistical physics; Statistics; Mathematics; Artificial intelligence; Biology; Genetics; Telecommunications; Physics; Gene","score_opus":0.02114399827825491,"score_gpt":0.25410128395502685,"score_spread":0.23295728567677193,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1993730176","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.008400087,0.0006344747,0.9893009,0.00034849183,0.00008951807,0.00009674136,0.00026935202,0.00031308556,0.0005472568],"genre_scores_gemma":[0.4896801,0.0045671086,0.48396578,0.00179781,0.0016467671,0.0028039368,0.0044296873,0.0006680235,0.010440811],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.98371017,0.008087419,0.0006162764,0.0037569876,0.003251583,0.00057761726],"domain_scores_gemma":[0.94495636,0.046645485,0.0026724776,0.0031729226,0.002168178,0.00038470305],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.019523662,0.0017864702,0.002285021,0.0015999121,0.00062481797,0.0019265357,0.0038252722,0.0025203535,0.0022583532],"category_scores_gemma":[0.04891649,0.0008114695,0.002649795,0.0021655662,0.0032229589,0.0025358736,0.0019121987,0.0037809538,0.0012484079],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00068751985,0.0003884461,0.017635548,0.0014564667,0.0014004733,0.00079641014,0.00070891995,0.6202148,0.0139809875,0.1839935,0.0094716,0.14926526],"study_design_scores_gemma":[0.000030532825,0.00020054904,0.0035852094,0.00004594141,0.00010440169,0.00020133147,0.00003228235,0.8946437,0.0018993278,0.09602911,0.0031799795,0.000047619582],"about_ca_topic_score_codex":0.0021401779,"about_ca_topic_score_gemma":0.0014154648,"teacher_disagreement_score":0.019523662,"about_ca_system_score_codex":0.0013456246,"about_ca_system_score_gemma":0.001354479,"threshold_uncertainty_score":0.10325223},"labels":[],"label_agreement":null},{"id":"W2000519643","doi":"10.1093/bioinformatics/btm024","title":"A simulation test bed for hypotheses of genome evolution","year":2007,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genome Rearrangement Algorithms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":45,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Dalhousie University","funders":"Australian Research Council; Genome Atlantic","keywords":"Computer science; Component (thermodynamics); Genetic algorithm; Genome; Software; Horizontal gene transfer; Data mining; Biology; Evolutionary biology; Gene; Genetics","score_opus":0.017961718638179174,"score_gpt":0.26076676186649106,"score_spread":0.2428050432283119,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2000519643","genre_codex":"empirical","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.68802184,0.00092350063,0.28070572,0.0026717458,0.00025681767,0.0003448764,0.004169624,0.0019389581,0.020966865],"genre_scores_gemma":[0.9034744,0.00030819606,0.08913213,0.00024150246,0.00007688745,0.00062881765,0.0033540234,0.00020454846,0.0025794765],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9992248,0.0004833324,0.000035999845,0.00009953887,0.000089056644,0.00006721951],"domain_scores_gemma":[0.9795202,0.018426126,0.00036249642,0.00048226514,0.0008004806,0.0004085333],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002540612,0.0008424942,0.0007980475,0.0011771051,0.0009153547,0.0012447597,0.0018764006,0.0016421746,0.009913633],"category_scores_gemma":[0.01754119,0.000384219,0.0008447539,0.00097448425,0.0010721122,0.001080112,0.0012355731,0.0011329234,0.00067898526],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001269594,0.00009060375,0.0032279356,0.0000784232,0.00003144612,0.00013622965,0.000047767913,0.97891325,0.00026383935,0.011892853,0.0013868755,0.0038038732],"study_design_scores_gemma":[0.000026728496,0.000018943783,0.00014346786,0.0000057600105,0.000004086403,0.000015096446,0.000013288268,0.9942821,0.00016759934,0.0050211255,0.00029852006,0.000003252392],"about_ca_topic_score_codex":0.0059065903,"about_ca_topic_score_gemma":0.002919365,"teacher_disagreement_score":0.009913633,"about_ca_system_score_codex":0.0011341004,"about_ca_system_score_gemma":0.0010388899,"threshold_uncertainty_score":0.033164382},"labels":[],"label_agreement":null},{"id":"W2002295829","doi":"10.1093/bioinformatics/btv003","title":"Clonality inference in multiple tumor samples using phylogeny","year":2015,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Cancer Genomics and Diagnostics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":284,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University; BC Cancer Agency","funders":"BC Cancer Agency; Natural Sciences and Engineering Research Council of Canada; Genome Canada","keywords":"Phylogenetics; Biology; Phylogenetic tree; Inference; In silico; Computational biology; Evolutionary biology; Computer science; Genetics; Gene; Artificial intelligence","score_opus":0.06875968615801234,"score_gpt":0.2941810204304562,"score_spread":0.22542133427244385,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2002295829","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.20670119,0.0006238822,0.7879759,0.00042643942,0.000027846034,0.000081604776,0.001972797,0.0009981206,0.0011922198],"genre_scores_gemma":[0.7478815,0.0002533022,0.24728604,0.00019982988,0.000052111736,0.00011917262,0.0035235272,0.00019805822,0.00048644733],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9984394,0.0006675389,0.000089869,0.00048351494,0.0002322982,0.000087404995],"domain_scores_gemma":[0.9914785,0.0061612125,0.00093825144,0.000675465,0.00044359057,0.0003030144],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0029133095,0.00076429965,0.0012473898,0.0017060214,0.00078848633,0.0014696941,0.001491611,0.0013959676,0.0017623334],"category_scores_gemma":[0.015422852,0.00070402713,0.0010146347,0.0018843963,0.00091450865,0.0014046436,0.0015550496,0.001775848,0.00041692352],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00051936303,0.00009550799,0.06872291,0.00037713812,0.0003245736,0.00076711253,0.00027640277,0.81392,0.02618374,0.012341991,0.0026403114,0.073831014],"study_design_scores_gemma":[0.000030100651,0.000039493876,0.0041738236,0.000028601478,0.000045174405,0.00032470838,0.00003391402,0.9674817,0.004653782,0.02184294,0.0013271654,0.000018698602],"about_ca_topic_score_codex":0.0024348823,"about_ca_topic_score_gemma":0.0033712029,"teacher_disagreement_score":0.0029133095,"about_ca_system_score_codex":0.00079637655,"about_ca_system_score_gemma":0.001039568,"threshold_uncertainty_score":0.015407264},"labels":[],"label_agreement":null},{"id":"W2012328875","doi":"10.1093/bioinformatics/btu714","title":"Characterization of structural variants with single molecule and hybrid sequencing approaches","year":2014,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":63,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Human Genome Research Institute; Natural Sciences and Engineering Research Council of Canada; Burroughs Wellcome Fund; National Institutes of Health; National Science Foundation","keywords":"DNA sequencing; Structural variation; Genome; Computational biology; Illumina dye sequencing; Biology; Sequence assembly; Software; Hybrid genome assembly; Computer science; Reference genome; Genetics; DNA; Gene","score_opus":0.017511604946140517,"score_gpt":0.18655111901378327,"score_spread":0.16903951406764275,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2012328875","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.53248644,0.001629562,0.44666985,0.00017446604,0.00010023435,0.0001625725,0.01153266,0.00311822,0.0041259993],"genre_scores_gemma":[0.6281963,0.0005300878,0.35668817,0.00018675637,0.00003747973,0.00016738515,0.011967627,0.00057872775,0.0016474931],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9994823,0.00008836277,0.000031081527,0.00022076236,0.0001513354,0.000026178663],"domain_scores_gemma":[0.9985238,0.00077409594,0.00028058927,0.00017676236,0.00018097798,0.00006394891],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00087767653,0.0004443232,0.0004359191,0.001482976,0.00022402963,0.0007834433,0.0005676508,0.0007262245,0.001932708],"category_scores_gemma":[0.002424071,0.00028068613,0.0006676602,0.0012114259,0.00030936926,0.00044876154,0.00041257092,0.00057209603,0.0010462985],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00070509594,0.00012181188,0.029943397,0.00053893455,0.00045020503,0.00028905075,0.00020946705,0.011190406,0.89021045,0.0033597818,0.0014460064,0.06153537],"study_design_scores_gemma":[0.00005396462,0.0005310683,0.06343657,0.00005433867,0.00029441115,0.0013057117,0.00016717675,0.15850753,0.7575223,0.0056526186,0.012365892,0.00010841673],"about_ca_topic_score_codex":0.0006344208,"about_ca_topic_score_gemma":0.0012940617,"teacher_disagreement_score":0.001932708,"about_ca_system_score_codex":0.0003344402,"about_ca_system_score_gemma":0.000157551,"threshold_uncertainty_score":0.0064655542},"labels":[],"label_agreement":null},{"id":"W2016928406","doi":"10.1093/bioinformatics/bth451","title":"Discovering patterns to extract protein–protein interactions from full texts","year":2004,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Biomedical Text Mining and Ontologies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":238,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; Recall rate; Scientific literature; Matching (statistics); Information retrieval; Natural language processing; Artificial intelligence; Data mining; Biology","score_opus":0.015499909559941955,"score_gpt":0.267165727324153,"score_spread":0.25166581776421104,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2016928406","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.14015535,0.0022224074,0.82563525,0.001515233,0.00011106002,0.0008665443,0.012340663,0.01188825,0.0052652713],"genre_scores_gemma":[0.14158413,0.0009543115,0.8376223,0.00018710073,0.00011106727,0.00047838074,0.016636407,0.00033595064,0.0020904369],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9983241,0.00029749412,0.00032063996,0.00056244235,0.00041349788,0.000081860446],"domain_scores_gemma":[0.99323684,0.0042275465,0.0008915794,0.0005086994,0.0009319963,0.00020336945],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016072782,0.001359032,0.00095645466,0.008935859,0.0007376109,0.0016514097,0.0013687036,0.001199044,0.0033064066],"category_scores_gemma":[0.008883455,0.0005236529,0.0010605433,0.0058569727,0.0006311505,0.0036376158,0.0014113077,0.0009292141,0.0035135765],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006136655,0.00067552633,0.023975987,0.002244829,0.00028940203,0.0014783839,0.0008317591,0.005176969,0.06321051,0.0062115504,0.015484469,0.879807],"study_design_scores_gemma":[0.00042050757,0.00086425524,0.051733825,0.0006936747,0.0010279241,0.010380289,0.002551163,0.5369634,0.2139628,0.09066459,0.090519115,0.00021845313],"about_ca_topic_score_codex":0.0009805065,"about_ca_topic_score_gemma":0.0015919708,"teacher_disagreement_score":0.008935859,"about_ca_system_score_codex":0.00044651856,"about_ca_system_score_gemma":0.0010888096,"threshold_uncertainty_score":0.011061072},"labels":[],"label_agreement":null},{"id":"W2022441134","doi":"10.1093/bioinformatics/btg182","title":"Class prediction and discovery using gene microarray andproteomics mass spectroscopy data: curses, caveats, cautions","year":2003,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":332,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"National Research Council Canada; National Research Council Institute for Biodiagnostics","funders":"","keywords":"Computer science; Curse of dimensionality; Artificial intelligence; Pattern recognition (psychology); Feature (linguistics); Relevance (law); Data mining; Microarray analysis techniques; Identification (biology); Outlier; Machine learning; Biology; Gene","score_opus":0.02672686303689213,"score_gpt":0.270721456698394,"score_spread":0.24399459366150184,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2022441134","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":"methods","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":"methods","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.04042498,0.009708071,0.59968996,0.31817144,0.010563032,0.0013198437,0.0034614073,0.0050137946,0.011647548],"genre_scores_gemma":[0.26424077,0.004080911,0.6048182,0.10468014,0.009547694,0.003185086,0.0012120375,0.0012235169,0.007011552],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.8853942,0.0689757,0.013073975,0.005416232,0.02624118,0.0008988306],"domain_scores_gemma":[0.4902711,0.38457993,0.020043481,0.061289754,0.0419264,0.0018893305],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.17960419,0.0015315367,0.0024460743,0.00530771,0.0044263876,0.006545733,0.007214189,0.0036864884,0.0023637277],"category_scores_gemma":[0.48958647,0.0009583092,0.0018631137,0.0054881056,0.009792486,0.00741513,0.004712669,0.0101850005,0.0020403962],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00087631715,0.00024764874,0.038886126,0.0033071628,0.0012259092,0.001900594,0.008468962,0.010044381,0.0048637814,0.11623336,0.38190645,0.43203938],"study_design_scores_gemma":[0.00033691135,0.0003437484,0.029062388,0.0048657553,0.0005395048,0.0044059125,0.0042308113,0.07430884,0.01426827,0.67509896,0.1919379,0.00060108304],"about_ca_topic_score_codex":0.011780829,"about_ca_topic_score_gemma":0.020101856,"teacher_disagreement_score":0.8203958,"about_ca_system_score_codex":0.002325916,"about_ca_system_score_gemma":0.0041055074,"threshold_uncertainty_score":0.94984907},"labels":[],"label_agreement":null},{"id":"W2029921227","doi":"10.1093/bioinformatics/btl648","title":"DataBiNS: a BioMoby-based data-mining workflow for biological pathways and non-synonymous SNPs","year":2007,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Biomedical Text Mining and Ontologies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"St. Paul's Hospital; University of British Columbia","funders":"Genome Canada","keywords":"Workflow; Computer science; Identifier; Web service; Workflow management system; World Wide Web; XML; Data mining; Information retrieval; Database; Programming language","score_opus":0.07184932710687342,"score_gpt":0.306166776981287,"score_spread":0.2343174498744136,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2029921227","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.018247329,0.0009802737,0.61933535,0.0018198857,0.00026356635,0.0010191095,0.15233162,0.19976021,0.0062426245],"genre_scores_gemma":[0.066642165,0.0013846625,0.69625115,0.0006867591,0.00010774782,0.0014527381,0.21681412,0.0108252205,0.0058353986],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99879104,0.0002029247,0.00023728328,0.0003541747,0.0003352958,0.00007928092],"domain_scores_gemma":[0.99553865,0.0018095409,0.00056764524,0.0009010235,0.0007453342,0.00043777341],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0037567788,0.0014368992,0.0012511188,0.0043064565,0.0013258215,0.0023779687,0.0021041606,0.0010454615,0.013000688],"category_scores_gemma":[0.009133165,0.0007340323,0.001435945,0.004279893,0.00070503564,0.0018350597,0.0024447225,0.0009688497,0.009238868],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0042732283,0.00051886536,0.034634203,0.006583897,0.0007621548,0.002430073,0.0018073447,0.015319772,0.073056825,0.028411774,0.3579709,0.474231],"study_design_scores_gemma":[0.0009124899,0.00024718867,0.026773378,0.0010749538,0.0002762195,0.0022308063,0.0007812782,0.11431911,0.104229696,0.068328485,0.6804254,0.00040103344],"about_ca_topic_score_codex":0.0069573587,"about_ca_topic_score_gemma":0.008724658,"teacher_disagreement_score":0.013000688,"about_ca_system_score_codex":0.0011405379,"about_ca_system_score_gemma":0.0036781435,"threshold_uncertainty_score":0.04349166},"labels":[],"label_agreement":null},{"id":"W2040481795","doi":"10.1093/bioinformatics/btu673","title":"FuncPatch: a web server for the fast Bayesian inference of conserved functional patches in protein 3D structures","year":2014,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":26,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University","funders":"","keywords":"Inference; Computer science; Web server; Bayesian probability; Computational biology; Bayesian inference; Artificial intelligence; Biology; The Internet; World Wide Web","score_opus":0.018036198505034032,"score_gpt":0.22551523590222008,"score_spread":0.20747903739718604,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2040481795","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.006659608,0.00053753663,0.5070888,0.00035780895,0.00024088893,0.00027701168,0.017954553,0.46255463,0.00432921],"genre_scores_gemma":[0.14876805,0.0012882355,0.6334089,0.0008703829,0.00027330816,0.0022212975,0.09966738,0.10336038,0.010142087],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9990609,0.00016953089,0.000072691604,0.00020302096,0.00037596762,0.000117887765],"domain_scores_gemma":[0.99623674,0.0018915626,0.000320556,0.0006136491,0.00067675114,0.0002608265],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0025129882,0.0032585165,0.0020768533,0.0026533322,0.001120729,0.0019713007,0.008448143,0.0024655776,0.05531099],"category_scores_gemma":[0.011055517,0.0019448346,0.0023524784,0.0019941004,0.0009788236,0.0033482907,0.0025102417,0.003032352,0.028301852],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0020190466,0.0004080405,0.008955275,0.0026943493,0.0007953331,0.0014710705,0.00042478406,0.112225644,0.01699667,0.023302518,0.5710448,0.25966248],"study_design_scores_gemma":[0.00077593984,0.00014404181,0.003936784,0.00020712409,0.000109879045,0.0010968908,0.00008143727,0.8297209,0.02366202,0.03621874,0.103712805,0.0003334034],"about_ca_topic_score_codex":0.0080103185,"about_ca_topic_score_gemma":0.0064324196,"teacher_disagreement_score":0.05531099,"about_ca_system_score_codex":0.0014678632,"about_ca_system_score_gemma":0.0027399098,"threshold_uncertainty_score":0.18503374},"labels":[],"label_agreement":null},{"id":"W2042280132","doi":"10.1093/bioinformatics/18.11.1538","title":"AcePrimer: automation of PCR primer design based on genestructure","year":2002,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetics, Aging, and Longevity in Model Organisms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"BC Cancer Agency","funders":"BC Cancer Agency; Genome Canada","keywords":"Primer (cosmetics); Perl; Caenorhabditis elegans; Genetics; Biology; Computational biology; In silico PCR; genomic DNA; Variants of PCR; Computer science; Polymerase chain reaction; Gene; Programming language; Multiplex polymerase chain reaction","score_opus":0.019623065265681378,"score_gpt":0.2214974692156349,"score_spread":0.2018744039499535,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2042280132","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.004319506,0.00043500419,0.9249524,0.00017947107,0.00026003493,0.0007229712,0.006475396,0.059911996,0.0027432335],"genre_scores_gemma":[0.0126140695,0.0003162341,0.95815057,0.00026837445,0.00006639712,0.0023184167,0.011082346,0.011218939,0.0039646886],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99645895,0.0008405056,0.00051101,0.0009062959,0.0010238297,0.00025939266],"domain_scores_gemma":[0.9946068,0.0031261544,0.00046864667,0.00075108104,0.0007917336,0.00025546056],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005204848,0.0027483318,0.0024659904,0.0032800133,0.0010025129,0.0016558495,0.0026943805,0.0011511901,0.03966989],"category_scores_gemma":[0.010499658,0.0024751637,0.0017245987,0.0015500387,0.00087784167,0.0015858782,0.0015172411,0.0041440725,0.0420069],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0024012395,0.0004275709,0.0016164312,0.0027769895,0.00021843516,0.0006947789,0.00052925304,0.004154238,0.5223934,0.006321704,0.1310146,0.32745132],"study_design_scores_gemma":[0.0005817033,0.0004604817,0.0028133749,0.00030228661,0.00016791474,0.0013319639,0.00006722655,0.037812542,0.68753755,0.008490715,0.26012927,0.0003049688],"about_ca_topic_score_codex":0.00057468866,"about_ca_topic_score_gemma":0.0013649409,"teacher_disagreement_score":0.03966989,"about_ca_system_score_codex":0.0006950728,"about_ca_system_score_gemma":0.0015052655,"threshold_uncertainty_score":0.13270903},"labels":[],"label_agreement":null},{"id":"W2042555832","doi":"10.1093/bioinformatics/bth247","title":"An interactive triple-helical collagen builder","year":2004,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Collagen: Extraction and Characterization","field":"Materials Science","cited_by":104,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto; University of Alberta","funders":"","keywords":"Documentation; Computer science; Parsing; Terminal (telecommunication); Programming language; Algorithm","score_opus":0.014075450458584472,"score_gpt":0.2734084892093066,"score_spread":0.2593330387507221,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2042555832","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.011772505,0.0003385565,0.53731436,0.00030087397,0.00019138359,0.00035724335,0.025423877,0.40951934,0.014781793],"genre_scores_gemma":[0.08601627,0.0006104716,0.7433706,0.0005265289,0.00014305246,0.0015422712,0.072150856,0.0523015,0.043338444],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9996382,0.000036331494,0.000024754905,0.00008997905,0.00016785164,0.00004283032],"domain_scores_gemma":[0.999316,0.00030406413,0.000048971215,0.00009154072,0.00014930329,0.00009017834],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000746354,0.0013273446,0.0006130139,0.0008182774,0.00051906955,0.0007498054,0.0017525945,0.00062956836,0.09433736],"category_scores_gemma":[0.0016607976,0.0006079334,0.0006958545,0.00060269726,0.00022368792,0.0006164158,0.0012885309,0.0009546193,0.034642585],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001649689,0.00019366747,0.0034754272,0.0013012174,0.00011871883,0.0010980311,0.0003585405,0.009382927,0.13878104,0.004245401,0.3992154,0.44017985],"study_design_scores_gemma":[0.0005514621,0.00023118933,0.007352513,0.00025404623,0.00011593919,0.001839211,0.000108030414,0.22059695,0.21502359,0.008026425,0.5457191,0.00018150032],"about_ca_topic_score_codex":0.000995221,"about_ca_topic_score_gemma":0.0019546964,"teacher_disagreement_score":0.09433736,"about_ca_system_score_codex":0.0003632277,"about_ca_system_score_gemma":0.0005194096,"threshold_uncertainty_score":0.3155899},"labels":[],"label_agreement":null},{"id":"W2048796447","doi":"10.1093/bioinformatics/btl043","title":"Comparison of <i>P-RnaPredict</i> and <i>mfold</i>—algorithms for RNA secondary structure prediction","year":2006,"lang":"en","type":"article","venue":"Bioinformatics","topic":"RNA and protein synthesis mechanisms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":55,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"Natural Sciences and Engineering Research Council of Canada; Simon Fraser University","keywords":"RNA; Computer science; Algorithm; Computational biology; Biology; Genetics; Gene","score_opus":0.01121468157371556,"score_gpt":0.2478205323499416,"score_spread":0.23660585077622606,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2048796447","genre_codex":"empirical","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.49782518,0.005708861,0.41979843,0.0020489988,0.00075772667,0.0004064649,0.002660415,0.044118714,0.026675256],"genre_scores_gemma":[0.48645648,0.001322863,0.49872586,0.00037265022,0.00012494897,0.00030844624,0.006625057,0.0019450067,0.004118673],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99903095,0.00032578348,0.000075147036,0.00019570987,0.00027309443,0.00009936735],"domain_scores_gemma":[0.9963511,0.0021896227,0.00015762584,0.00036793845,0.00073154387,0.00020219726],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0025291534,0.0011939455,0.0008109493,0.001180155,0.00072550646,0.0008238523,0.0016130705,0.00088819885,0.002155134],"category_scores_gemma":[0.006147233,0.00035062322,0.0006021188,0.0013485581,0.00046165095,0.0013702739,0.00065059605,0.0009057714,0.0013568741],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0041044075,0.0007401371,0.016294248,0.0007189606,0.00043079362,0.00016131022,0.00017928494,0.37609416,0.014238572,0.008805523,0.028647346,0.5495852],"study_design_scores_gemma":[0.00016238246,0.00025224776,0.0020305002,0.000019615318,0.00003519774,0.00008617146,0.000043047097,0.9833023,0.009059507,0.002142005,0.0028508347,0.000016171662],"about_ca_topic_score_codex":0.005537861,"about_ca_topic_score_gemma":0.0042637032,"teacher_disagreement_score":0.005537861,"about_ca_system_score_codex":0.0010152783,"about_ca_system_score_gemma":0.0016585324,"threshold_uncertainty_score":0.01337564},"labels":[],"label_agreement":null},{"id":"W2059736685","doi":"10.1093/bioinformatics/btq393","title":"Inferring cancer subnetwork markers using density-constrained biclustering","year":2010,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Cancer Genomics and Diagnostics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":54,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University; University of British Columbia","funders":"Natural Sciences and Engineering Research Council of Canada; Mitacs","keywords":"Subnetwork; Computer science; Cancer; Computational biology; Software; Breast cancer; Biclustering; Artificial intelligence; Biology; Cluster analysis; Genetics","score_opus":0.011435322409209905,"score_gpt":0.2502772924467948,"score_spread":0.2388419700375849,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2059736685","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.13198611,0.00043459606,0.86031127,0.00036548055,0.000027208636,0.00019733023,0.0022606477,0.0034530777,0.0009642597],"genre_scores_gemma":[0.6050417,0.0002736279,0.38357973,0.0001864436,0.000043419794,0.0003528239,0.00921328,0.00033647177,0.00097246637],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9984316,0.0007459377,0.00011492116,0.00033915232,0.00026933287,0.000098942466],"domain_scores_gemma":[0.99233174,0.0053651948,0.0005547395,0.00076446653,0.00076668605,0.00021720058],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002679307,0.0016879282,0.0015786391,0.002707281,0.000732548,0.0010742162,0.0016985113,0.0010237059,0.0019995312],"category_scores_gemma":[0.016241461,0.0007324336,0.0011927321,0.0023522289,0.00077302236,0.00092096225,0.0019845255,0.0012362154,0.0009970726],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00043680661,0.00015634928,0.041609827,0.0003007379,0.000374334,0.00030041256,0.00021444318,0.8417875,0.005705365,0.0030884838,0.0045493557,0.10147643],"study_design_scores_gemma":[0.000019234429,0.000020295272,0.0025384498,0.000014934499,0.000029133138,0.000059994,0.000023491993,0.98999,0.0014829086,0.0054456973,0.0003630669,0.0000128299125],"about_ca_topic_score_codex":0.0087011,"about_ca_topic_score_gemma":0.00997099,"teacher_disagreement_score":0.0087011,"about_ca_system_score_codex":0.0006511391,"about_ca_system_score_gemma":0.0012087224,"threshold_uncertainty_score":0.017300904},"labels":[],"label_agreement":null},{"id":"W2076448238","doi":"10.1093/bioinformatics/btu759","title":"Local statistics allow quantification of cell-to-cell variability from high-throughput microscope images","year":2014,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Cell Image Analysis Techniques","field":"Biochemistry, Genetics and Molecular Biology","cited_by":18,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"Canadian Institutes of Health Research","keywords":"Cell; Protein subcellular localization prediction; Microscope; Subcellular localization; Microscopy; Biology; Computational biology; Cell biology; Cytoplasm; Pathology; Genetics; Gene; Medicine","score_opus":0.005301778245642983,"score_gpt":0.23982888409499534,"score_spread":0.23452710584935235,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2076448238","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.079113156,0.00041377152,0.9094958,0.00013430556,0.000028853296,0.00007308713,0.0021493912,0.007438695,0.0011529014],"genre_scores_gemma":[0.6309918,0.00035622655,0.3593347,0.00017774709,0.00009305685,0.00041644557,0.0047231656,0.0027394095,0.001167518],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99718446,0.00060750515,0.00018399961,0.00087087427,0.0010022135,0.0001508663],"domain_scores_gemma":[0.9789876,0.011473232,0.0035709452,0.0031066097,0.0024029955,0.0004586416],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0033356275,0.00067774596,0.0011855684,0.0024113206,0.00056232844,0.00252401,0.0014695359,0.0008410577,0.0022017458],"category_scores_gemma":[0.016147561,0.0005160287,0.00093224243,0.002610596,0.0009741984,0.0019369628,0.0014953833,0.0015197861,0.0011427886],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010354258,0.00029989713,0.07217441,0.0013101316,0.0010088849,0.0005697189,0.0006365536,0.2336953,0.3963389,0.022362981,0.013027256,0.25754052],"study_design_scores_gemma":[0.000040298306,0.0001301153,0.058714062,0.000058034977,0.00014502168,0.0004288683,0.00012428992,0.7710879,0.14189844,0.022235544,0.0050013354,0.00013607374],"about_ca_topic_score_codex":0.0021316186,"about_ca_topic_score_gemma":0.0033023423,"teacher_disagreement_score":0.0033356275,"about_ca_system_score_codex":0.0017174411,"about_ca_system_score_gemma":0.0009145374,"threshold_uncertainty_score":0.01764071},"labels":[],"label_agreement":null},{"id":"W2095681424","doi":"10.1093/bioinformatics/btu276","title":"Detecting independent and recurrent copy number aberrations using interval graphs","year":2014,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Cancer Genomics and Diagnostics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":29,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Human Genome Research Institute; Natural Sciences and Engineering Research Council of Canada; Burroughs Wellcome Fund; National Institutes of Health; National Science Foundation","keywords":"Heuristics; Computer science; Algorithm","score_opus":0.014584606739911712,"score_gpt":0.2643972812608552,"score_spread":0.2498126745209435,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2095681424","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.12954926,0.00073290104,0.85865486,0.00048564098,0.00002891948,0.0001917048,0.0028619913,0.0040109465,0.0034838354],"genre_scores_gemma":[0.49237463,0.00047844683,0.49797246,0.00018849246,0.00007119566,0.00016893407,0.006827377,0.00042579148,0.0014926639],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99911803,0.00019365265,0.000044674962,0.00029816912,0.00026350224,0.00008204861],"domain_scores_gemma":[0.9957658,0.0029930698,0.0004888624,0.00029960656,0.00031080292,0.00014194126],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010489613,0.00075321965,0.00072929997,0.0025710207,0.00051943655,0.00114933,0.0015707279,0.00075883453,0.0023801292],"category_scores_gemma":[0.0061120167,0.00040243322,0.00096221303,0.0020804042,0.0005697292,0.0013755609,0.0010468224,0.0010108155,0.0005906251],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00044826657,0.0001984466,0.026030641,0.00046120252,0.00016803664,0.00055768265,0.00022880897,0.67371297,0.019756954,0.019051425,0.007970156,0.25141543],"study_design_scores_gemma":[0.000036692352,0.0000552678,0.0037793654,0.000024285306,0.00004320122,0.00035072624,0.00007473006,0.95051825,0.0059323935,0.036625937,0.0025371914,0.000021989654],"about_ca_topic_score_codex":0.0043718345,"about_ca_topic_score_gemma":0.0051349,"teacher_disagreement_score":0.0043718345,"about_ca_system_score_codex":0.000817074,"about_ca_system_score_gemma":0.0008559622,"threshold_uncertainty_score":0.008692741},"labels":[],"label_agreement":null},{"id":"W2095885244","doi":"10.1093/bioinformatics/bth439","title":"Primer Design and Marker Clustering for Multiplex SNP-IT Primer Extension Genotyping Assay using Statistical Modeling","year":2004,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; McGill Genome Centre","funders":"","keywords":"Primer (cosmetics); Genotyping; Cluster analysis; Primer extension; Multiplex; SNP genotyping; SNP; Computer science; Statistical model; Computational biology; Biology; Genetics; Single-nucleotide polymorphism; Genotype; Artificial intelligence; Chemistry; Gene","score_opus":0.06619816459178136,"score_gpt":0.3131831912085179,"score_spread":0.24698502661673655,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2095885244","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0030372175,0.00010571426,0.99421,0.000051328178,0.00003395602,0.00026835015,0.00029447326,0.0017805747,0.00021833695],"genre_scores_gemma":[0.01054827,0.00008347796,0.98689854,0.00007081,0.000013869142,0.00097387214,0.0006623861,0.00031466776,0.00043416073],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9927987,0.003249382,0.000765529,0.0017150625,0.0012079949,0.00026331653],"domain_scores_gemma":[0.99191225,0.004513641,0.00086862966,0.0013064794,0.0012438287,0.00015514035],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0072218804,0.0018756569,0.0017034521,0.0018971169,0.0008687104,0.0011605899,0.0021784783,0.0018126044,0.0061013373],"category_scores_gemma":[0.017386252,0.001801382,0.0017390845,0.002277375,0.0008345088,0.0010635894,0.0010681129,0.0025729327,0.0062618623],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0016073311,0.00051833084,0.0064582624,0.0024972728,0.00041851233,0.00058530935,0.00053268526,0.11384624,0.5259467,0.03768683,0.01591011,0.2939925],"study_design_scores_gemma":[0.00018063813,0.00065306143,0.0023368879,0.00013510347,0.00026412134,0.0010634194,0.000058886057,0.49163345,0.43183523,0.024709076,0.046886913,0.00024323058],"about_ca_topic_score_codex":0.00060038356,"about_ca_topic_score_gemma":0.0010175931,"teacher_disagreement_score":0.0072218804,"about_ca_system_score_codex":0.00095065526,"about_ca_system_score_gemma":0.0021557626,"threshold_uncertainty_score":0.038193405},"labels":[],"label_agreement":null},{"id":"W2096066492","doi":"10.1093/bioinformatics/btu485","title":"Monte Carlo algorithms for Brownian phylogenetic models","year":2014,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Natural Sciences and Engineering Research Council of Canada; Agence Nationale de la Recherche","keywords":"Monte Carlo method; Markov chain Monte Carlo; Discretization; Algorithm; Statistical physics; Computer science; Brownian motion; Hybrid Monte Carlo; Mathematics; Mathematical optimization; Physics; Statistics; Mathematical analysis","score_opus":0.021551032029976158,"score_gpt":0.2363251618297108,"score_spread":0.21477412979973465,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2096066492","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0010080973,0.0003029653,0.99690175,0.00012703834,0.000034373625,0.000054872664,0.00008399974,0.00029423184,0.0011926156],"genre_scores_gemma":[0.064901896,0.0009982744,0.92701197,0.00028030385,0.00020882068,0.0008577493,0.0008876051,0.0005807178,0.004272628],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9973705,0.00141704,0.00015460807,0.00035337746,0.0005329331,0.00017159934],"domain_scores_gemma":[0.9855208,0.011860539,0.00055359805,0.00072061433,0.0010038567,0.0003406097],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0049980264,0.0015129923,0.0019369483,0.0019852037,0.0012265928,0.0025925178,0.0034259588,0.002893429,0.010662561],"category_scores_gemma":[0.025444984,0.001232314,0.0018170617,0.002364227,0.0021266048,0.003531992,0.0030203492,0.0041742604,0.0030984064],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000059153663,0.00005115262,0.0007767549,0.00017406153,0.000064160675,0.00007805052,0.00011953984,0.6839139,0.0003770983,0.27197558,0.003301169,0.0391094],"study_design_scores_gemma":[0.000020472458,0.000007011771,0.000043396816,0.000027345272,0.000007313233,0.00002384035,0.0000068563486,0.86091983,0.00012872039,0.13655046,0.0022538702,0.000010970247],"about_ca_topic_score_codex":0.0060057985,"about_ca_topic_score_gemma":0.0056294804,"teacher_disagreement_score":0.010662561,"about_ca_system_score_codex":0.002098161,"about_ca_system_score_gemma":0.0026079188,"threshold_uncertainty_score":0.035669804},"labels":[],"label_agreement":null},{"id":"W2096265658","doi":"10.1093/bioinformatics/btm552","title":"Identification of linked regions using high-density SNP genotype data in linkage analysis","year":2007,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":18,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Natural Sciences and Engineering Research Council of Canada; University of Hong Kong; City University of Hong Kong","keywords":"Identification (biology); Linkage (software); SNP; Genotype; Genetics; Computational biology; Biology; SNP array; Genetic linkage; Computer science; Single-nucleotide polymorphism; Gene","score_opus":0.04586778108750514,"score_gpt":0.3173606241963624,"score_spread":0.27149284310885724,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2096265658","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.08285716,0.00015389382,0.9069224,0.00020349897,0.000015353167,0.00015270161,0.001184505,0.007291896,0.0012185975],"genre_scores_gemma":[0.2506275,0.00015544424,0.7466126,0.00006716786,0.000016878776,0.00021757759,0.0012729596,0.00034518252,0.00068470975],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.998073,0.001152449,0.00012919075,0.0002840735,0.00031058773,0.000050698138],"domain_scores_gemma":[0.98724514,0.010546653,0.0006572262,0.00078036566,0.0005735663,0.00019709633],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004233187,0.0005703777,0.0008055504,0.0020090137,0.000514452,0.0011846974,0.0012423558,0.00077610026,0.0036559678],"category_scores_gemma":[0.016158776,0.00042704117,0.0005844131,0.001981718,0.00046964866,0.0008915361,0.0014492217,0.00062362093,0.0014137386],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001354027,0.0005335308,0.11455263,0.0009028964,0.00040515882,0.0019161112,0.0011460116,0.21583039,0.032479938,0.011902967,0.010301802,0.6086745],"study_design_scores_gemma":[0.00029166008,0.0002214706,0.025123825,0.00012188257,0.0002546907,0.0018685362,0.0002031104,0.88506305,0.05028309,0.026093563,0.0103459535,0.00012922469],"about_ca_topic_score_codex":0.0015306256,"about_ca_topic_score_gemma":0.0018886351,"teacher_disagreement_score":0.004233187,"about_ca_system_score_codex":0.0002810787,"about_ca_system_score_gemma":0.00079113606,"threshold_uncertainty_score":0.022387445},"labels":[],"label_agreement":null},{"id":"W2096349828","doi":"10.1093/bioinformatics/bth033","title":"GelScape: a web-based server for interactively annotating, manipulating, comparing and archiving 1D and 2D gel images","year":2004,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Advanced Proteomics Techniques and Applications","field":"Chemistry","cited_by":20,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Computer science; Annotation; Java applet; Java; Software; Automatic image annotation; Image warping; Web server; Image retrieval; Image (mathematics); Web page; World Wide Web; Database; The Internet; Operating system; Artificial intelligence","score_opus":0.022021752804720185,"score_gpt":0.28336869555760646,"score_spread":0.2613469427528863,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2096349828","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.009410535,0.0025691881,0.38894454,0.0007817936,0.0007676315,0.00085298193,0.11847324,0.45671716,0.021483012],"genre_scores_gemma":[0.027391847,0.002688519,0.49932042,0.00072465267,0.000382341,0.002153896,0.36579624,0.042869147,0.058672946],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99948466,0.000064473374,0.00005453831,0.00010798577,0.0002393928,0.000048935497],"domain_scores_gemma":[0.9988231,0.00024446147,0.00012626787,0.00024164544,0.00032820334,0.00023639624],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001202149,0.0023496612,0.0013016093,0.0035491516,0.0010412337,0.0022574714,0.0026774984,0.001262633,0.061101113],"category_scores_gemma":[0.0019241845,0.0014942032,0.0007622365,0.0015271759,0.0004085534,0.0016887947,0.00184369,0.0016581891,0.07629312],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000945999,0.00018565397,0.0007868269,0.001792869,0.00014647329,0.0005143783,0.00021709352,0.0010877261,0.19364779,0.0018518652,0.62122643,0.17759699],"study_design_scores_gemma":[0.00031041433,0.00013670199,0.0054991567,0.00030666872,0.000101847116,0.0022679858,0.00012762567,0.015873367,0.18818463,0.006188369,0.78067607,0.0003271729],"about_ca_topic_score_codex":0.0016427969,"about_ca_topic_score_gemma":0.0027772896,"teacher_disagreement_score":0.061101113,"about_ca_system_score_codex":0.0005276154,"about_ca_system_score_gemma":0.0011169044,"threshold_uncertainty_score":0.20440358},"labels":[],"label_agreement":null},{"id":"W2096482937","doi":"10.1093/bioinformatics/btv321","title":"ABC: a tool to identify SNVs causing allele-specific transcription factor binding from ChIP-Seq experiments","year":2015,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":25,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Princess Margaret Cancer Centre; University Health Network; University of Toronto; Ontario Institute for Cancer Research","funders":"National Cancer Institute; Canadian Institutes of Health Research","keywords":"Perl; Computational biology; Computer science; Biology; SNP; False positive paradox; Chromatin; Genetics; Transcription factor; Single-nucleotide polymorphism; Gene; Artificial intelligence; Programming language","score_opus":0.039393317429218665,"score_gpt":0.2816686085563228,"score_spread":0.24227529112710414,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2096482937","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.017510338,0.00092732545,0.51457304,0.0006878196,0.00051825505,0.00078606134,0.07363368,0.38490427,0.0064591384],"genre_scores_gemma":[0.07846132,0.0005874793,0.76214194,0.0016872669,0.00016846499,0.0050659734,0.08568411,0.059672084,0.006531318],"study_design_codex":"not_applicable","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9982284,0.00042288928,0.00020124303,0.0005049094,0.0005011245,0.0001415379],"domain_scores_gemma":[0.99209577,0.006049869,0.00045816542,0.00058083475,0.0006073423,0.00020813558],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0038782894,0.0028644812,0.002528374,0.0037928736,0.0014949327,0.0022316745,0.003713161,0.0020557237,0.06055212],"category_scores_gemma":[0.014727298,0.0017246922,0.0021572812,0.0023017533,0.00082320103,0.0013473084,0.002762221,0.0020327256,0.02090928],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0020803781,0.00037738073,0.022055736,0.0052879746,0.0016402665,0.0014836646,0.0005742857,0.019033365,0.038609263,0.009948601,0.7248269,0.17408217],"study_design_scores_gemma":[0.002315281,0.00041655684,0.021641951,0.00068522774,0.00096115936,0.0028988316,0.00029612624,0.5496525,0.069102034,0.03790439,0.3134528,0.00067312754],"about_ca_topic_score_codex":0.003813809,"about_ca_topic_score_gemma":0.007824335,"teacher_disagreement_score":0.06055212,"about_ca_system_score_codex":0.0009971097,"about_ca_system_score_gemma":0.0022754506,"threshold_uncertainty_score":0.20256704},"labels":[],"label_agreement":null},{"id":"W2096529831","doi":"10.1093/bioinformatics/btv159","title":"A novel motif-discovery algorithm to identify co-regulatory motifs in large transcription factor and microRNA co-regulatory networks in human","year":2015,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":32,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Computational biology; Gene regulatory network; microRNA; Biological network; Biology; Motif (music); Gene; Transcription factor; Robustness (evolution); DNA binding site; Regulation of gene expression; Systems biology; Computer science; Genetics; Gene expression; Promoter","score_opus":0.017958116469330213,"score_gpt":0.2728153177653513,"score_spread":0.2548572012960211,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2096529831","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.064096704,0.0008014738,0.9181573,0.00016536412,0.000044961715,0.00021625507,0.0024476147,0.013011307,0.0010590497],"genre_scores_gemma":[0.17168751,0.00016158947,0.821242,0.00011785125,0.000024249668,0.00028103564,0.0046467255,0.00049467455,0.0013444374],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9994437,0.00009580019,0.000044894092,0.00027594995,0.00009678849,0.000042808388],"domain_scores_gemma":[0.99906784,0.00055981806,0.00010838119,0.000093830095,0.000121008416,0.000049179467],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009732484,0.0009725129,0.0009010419,0.003006582,0.00065618556,0.000682711,0.0014253489,0.0011019916,0.0032826758],"category_scores_gemma":[0.0036274234,0.0005215371,0.0009793885,0.002064173,0.0003775348,0.0010090354,0.0010426388,0.00061002315,0.00085091754],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011230262,0.00037884264,0.025089713,0.0009208455,0.0005748386,0.00084048486,0.00036004986,0.15636459,0.03697023,0.009201773,0.020866156,0.7473093],"study_design_scores_gemma":[0.00013263125,0.00008758193,0.002347168,0.000025622348,0.00005341409,0.00070284685,0.000054436638,0.9758395,0.006517658,0.009349153,0.0048650997,0.000024810335],"about_ca_topic_score_codex":0.002742286,"about_ca_topic_score_gemma":0.0047403607,"teacher_disagreement_score":0.0032826758,"about_ca_system_score_codex":0.00054777233,"about_ca_system_score_gemma":0.0010660702,"threshold_uncertainty_score":0.010981619},"labels":[],"label_agreement":null},{"id":"W2096589119","doi":"10.1093/bioinformatics/btt452","title":"The BioPAX Validator","year":2013,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"National Cancer Institute; National Human Genome Research Institute; National Institutes of Health","keywords":"Validator; Computer science; Programming language; Service (business); Consistency (knowledge bases); Java; World Wide Web; Information retrieval; Software engineering; Artificial intelligence","score_opus":0.007486780389627012,"score_gpt":0.2108091162860064,"score_spread":0.20332233589637938,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2096589119","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0024346982,0.00076998427,0.21509677,0.0020888187,0.0010164578,0.0008848876,0.13231955,0.59813005,0.047258724],"genre_scores_gemma":[0.021877114,0.0012127202,0.20782232,0.0029118399,0.0004822266,0.0026743999,0.40128595,0.28919965,0.072533734],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99324846,0.0014526853,0.0013404505,0.0014240758,0.0020372544,0.0004970335],"domain_scores_gemma":[0.9770861,0.0068677715,0.0020464102,0.006244688,0.0070800767,0.00067483936],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.011492315,0.0034389484,0.0017038066,0.005170757,0.002428414,0.0073127197,0.0046203732,0.002376182,0.17894235],"category_scores_gemma":[0.030074254,0.0027810612,0.0023729808,0.005186708,0.001061128,0.008184634,0.0054877703,0.0040739337,0.19067752],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010397726,0.000096148455,0.0016336727,0.0014766149,0.00008220014,0.00039523817,0.00054684427,0.0006241925,0.0033522116,0.012703991,0.8862817,0.09176731],"study_design_scores_gemma":[0.00014600123,0.00005325278,0.001283643,0.00048635434,0.000043765795,0.00039212237,0.00009177016,0.0032634323,0.011761786,0.00824857,0.9741013,0.00012810594],"about_ca_topic_score_codex":0.003932341,"about_ca_topic_score_gemma":0.0020879654,"teacher_disagreement_score":0.17894235,"about_ca_system_score_codex":0.002373158,"about_ca_system_score_gemma":0.0048249457,"threshold_uncertainty_score":0.59862185},"labels":[],"label_agreement":null},{"id":"W2097185378","doi":"10.1093/bioinformatics/btr392","title":"MetATT: a web-based metabolomics tool for analyzing time-series and two-factor datasets","year":2011,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Metabolomics and Mass Spectrometry Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":73,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"National Institute for Nanotechnology; University of Alberta","funders":"Genome Alberta; Alberta Innovates","keywords":"Principal component analysis; Computer science; Metabolomics; Visualization; Data mining; Multivariate statistics; Time series; Session (web analytics); Identification (biology); Series (stratigraphy); Bioinformatics; Machine learning; Artificial intelligence; Biology","score_opus":0.018311101628872624,"score_gpt":0.24747668004217865,"score_spread":0.22916557841330604,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2097185378","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0053720917,0.0004616311,0.4469121,0.00040493673,0.0002065385,0.00038462144,0.12526129,0.41872406,0.0022727288],"genre_scores_gemma":[0.049135827,0.0007966605,0.69960105,0.0005780659,0.00023805322,0.0035749876,0.20222919,0.03876116,0.005084953],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99908805,0.00019219721,0.0001468039,0.00020616173,0.00030691596,0.000059816524],"domain_scores_gemma":[0.99498063,0.003177405,0.00044969792,0.0006695676,0.00047110725,0.0002515718],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0034216766,0.002679613,0.0017305294,0.0057226284,0.0007533765,0.0024765132,0.0026033944,0.0014308834,0.04925218],"category_scores_gemma":[0.009573015,0.0013596299,0.0025642237,0.005301173,0.00033301735,0.0027363147,0.0028600064,0.0020676493,0.014617942],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0031042076,0.00056961674,0.0155967865,0.0059653437,0.002064705,0.0021780608,0.00089456013,0.015234786,0.051830847,0.010877606,0.5125134,0.37917006],"study_design_scores_gemma":[0.0015042534,0.00056368974,0.033995032,0.0008925262,0.0008604945,0.0026607618,0.0004548668,0.2617574,0.07123938,0.08455405,0.5406123,0.00090529333],"about_ca_topic_score_codex":0.0020273244,"about_ca_topic_score_gemma":0.0033319294,"teacher_disagreement_score":0.04925218,"about_ca_system_score_codex":0.00054213044,"about_ca_system_score_gemma":0.001417107,"threshold_uncertainty_score":0.16476494},"labels":[],"label_agreement":null},{"id":"W2097392849","doi":"10.1093/bioinformatics/btt624","title":"Flexible analysis of RNA-seq data using mixed effects models","year":2013,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":95,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"NIHR Cambridge Biomedical Research Centre; Biotechnology and Biological Sciences Research Council; Cancer Research UK","keywords":"Computer science; Expression (computer science); Bayesian probability; RNA-Seq; Data mining; Selection (genetic algorithm); Computational biology; Gene expression; Biology; Artificial intelligence; Gene; Genetics; Transcriptome","score_opus":0.05719972929131283,"score_gpt":0.2979505318125822,"score_spread":0.24075080252126935,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2097392849","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0025711386,0.000088864654,0.9943329,0.00007719376,0.00002597055,0.000051950883,0.0005317447,0.0022119663,0.00010827091],"genre_scores_gemma":[0.05391226,0.00021146121,0.9402007,0.00025377885,0.00007697822,0.0007492598,0.0022666776,0.0015833839,0.00074543676],"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9947548,0.0027783045,0.0002754727,0.001269596,0.0007407325,0.00018117909],"domain_scores_gemma":[0.98669785,0.0101186875,0.00076200755,0.00127791,0.00079711474,0.00034650732],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.011892257,0.0020726267,0.0021036642,0.0016165646,0.0010247588,0.002451507,0.0048500267,0.0016810098,0.0040427134],"category_scores_gemma":[0.018185461,0.0017505942,0.0055152783,0.0020763534,0.0014784927,0.0018562309,0.0027690951,0.003939656,0.0016180613],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010437,0.00025173003,0.011743368,0.0013696012,0.002208026,0.0007263448,0.0006111235,0.7406136,0.058144435,0.038533114,0.007922627,0.1368323],"study_design_scores_gemma":[0.00006080465,0.00006048384,0.0012065727,0.000033462704,0.00011390884,0.00008168499,0.000022855342,0.9497485,0.007196638,0.037429955,0.0039822734,0.00006286703],"about_ca_topic_score_codex":0.003408879,"about_ca_topic_score_gemma":0.006225067,"teacher_disagreement_score":0.011892257,"about_ca_system_score_codex":0.0013111244,"about_ca_system_score_gemma":0.001911723,"threshold_uncertainty_score":0.06289303},"labels":[],"label_agreement":null},{"id":"W2097508475","doi":"10.1093/bioinformatics/btt138","title":"Cytoscape App Store","year":2013,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":192,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"National Institute of General Medical Sciences; National Institutes of Health","keywords":"App store; Computer science; World Wide Web; Software; Open source; Data science; Visualization; Information retrieval; Data mining; Operating system","score_opus":0.006885813618096346,"score_gpt":0.20567563890478122,"score_spread":0.19878982528668487,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2097508475","genre_codex":"dataset","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0018828083,0.0018675148,0.026256282,0.0017056155,0.0017950964,0.00075890886,0.6073371,0.30756983,0.05082681],"genre_scores_gemma":[0.018385712,0.0024771036,0.056353208,0.003877184,0.001029083,0.00436853,0.75425565,0.09084249,0.06841103],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9992729,0.00008827251,0.00007027148,0.00015634498,0.0003258378,0.000086383894],"domain_scores_gemma":[0.99559057,0.001926467,0.00025292896,0.0005793926,0.0011584287,0.00049215055],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0012945868,0.0018937332,0.0019576552,0.0049869902,0.0015571779,0.002624651,0.004260334,0.0021036512,0.3213534],"category_scores_gemma":[0.010185774,0.001770149,0.0011135791,0.0033544367,0.00036580145,0.0035501532,0.002592737,0.0027979237,0.21323282],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00011970063,0.000023929944,0.0002814437,0.0012658138,0.000053862146,0.00018296992,0.00007718304,0.00016532381,0.0009303044,0.001264814,0.9778463,0.017788425],"study_design_scores_gemma":[0.00025219796,0.00003194943,0.0012236547,0.00040391122,0.00008054216,0.0005197805,0.000050907827,0.0018580208,0.0054002586,0.008071763,0.981967,0.0001400265],"about_ca_topic_score_codex":0.004724558,"about_ca_topic_score_gemma":0.009234549,"teacher_disagreement_score":0.3213534,"about_ca_system_score_codex":0.00057549495,"about_ca_system_score_gemma":0.0022856144,"threshold_uncertainty_score":0.968007},"labels":[],"label_agreement":null},{"id":"W2097706952","doi":"10.1093/bioinformatics/bth903","title":"Analysis of domain correlations in yeast protein complexes","year":2004,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":14,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Lunenfeld-Tanenbaum Research Institute; Mount Sinai Hospital","funders":"Canadian Institutes of Health Research; Genome Canada","keywords":"Computational biology; Biology; Saccharomyces cerevisiae; Ribosome; Protein domain; RNA-binding protein; Domain (mathematical analysis); RNA; Protein–protein interaction; Genetics; Yeast; Gene","score_opus":0.008864431876002487,"score_gpt":0.2266656028947092,"score_spread":0.2178011710187067,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2097706952","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9749951,0.0008411814,0.018163987,0.00006789809,0.000007294839,0.000041792435,0.0045206724,0.00053217326,0.00082981645],"genre_scores_gemma":[0.9704065,0.00034897382,0.016100615,0.000019259622,0.000008413657,0.00006619351,0.01279899,0.0000578981,0.00019304098],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9993824,0.00014516427,0.00006049397,0.00018585466,0.00016242224,0.00006371474],"domain_scores_gemma":[0.9972941,0.0014401498,0.0005878241,0.00022303488,0.0002935182,0.00016132781],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007490227,0.00055115955,0.00045789065,0.0039210725,0.00053261296,0.00072035077,0.00044859562,0.00033402053,0.0011809696],"category_scores_gemma":[0.004433938,0.00021930232,0.00055039336,0.0036897685,0.00036653577,0.0006567426,0.0006372131,0.000346113,0.00026303763],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0020965345,0.0004180244,0.49919432,0.0016782772,0.0012191313,0.0015413137,0.0006035802,0.2671367,0.10994957,0.012716718,0.0064471103,0.096998826],"study_design_scores_gemma":[0.00005832099,0.00014683977,0.27080524,0.000055076256,0.00024551852,0.0012102526,0.00023944612,0.67426175,0.03399633,0.013241791,0.0056846957,0.000054687738],"about_ca_topic_score_codex":0.0025798585,"about_ca_topic_score_gemma":0.0024475919,"teacher_disagreement_score":0.0039210725,"about_ca_system_score_codex":0.0008114316,"about_ca_system_score_gemma":0.00045501045,"threshold_uncertainty_score":0.0058873296},"labels":[],"label_agreement":null},{"id":"W2097828850","doi":"10.1093/bioinformatics/btp010","title":"Profiling model T-cell metagenomes with short reads","year":2009,"lang":"en","type":"article","venue":"Bioinformatics","topic":"T-cell and B-cell Immunology","field":"Immunology and Microbiology","cited_by":42,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canada's Michael Smith Genome Sciences Centre; BC Cancer Agency","funders":"Genome British Columbia; Michael Smith Health Research BC; Genome Canada","keywords":"T-cell receptor; Computational biology; Biology; Profiling (computer programming); Genetics; T cell; Algorithm; Computer science; Immune system","score_opus":0.017203179056754572,"score_gpt":0.22348327644139065,"score_spread":0.2062800973846361,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2097828850","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7615795,0.00031470103,0.23240532,0.00018610545,0.00004964334,0.000152697,0.0028066735,0.0011911561,0.0013142665],"genre_scores_gemma":[0.83776903,0.00028230133,0.1544007,0.00018914789,0.00002321967,0.0003760932,0.00565339,0.00023101381,0.0010750581],"study_design_codex":"simulation_or_modeling","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9993869,0.00017684312,0.000036068683,0.00021254696,0.00013147181,0.000056018132],"domain_scores_gemma":[0.99830735,0.00090914324,0.0001889557,0.00021610042,0.00028331167,0.00009507807],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010843013,0.0006525749,0.00060708774,0.0004544571,0.00042777354,0.0008551796,0.0007307345,0.0009620709,0.0010228242],"category_scores_gemma":[0.003239438,0.0004531907,0.00083731697,0.0006883361,0.0004927818,0.00096813554,0.0005346653,0.0007778213,0.00054154865],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000581397,0.00014436261,0.02299063,0.00024773687,0.00010716604,0.00020487951,0.00021639609,0.86246616,0.097846106,0.0023025516,0.00040641753,0.01248617],"study_design_scores_gemma":[0.000025038626,0.00018640474,0.003572737,0.0000107139795,0.00002311824,0.00007387747,0.00007171112,0.96636105,0.025688589,0.0030215313,0.00094383425,0.000021373417],"about_ca_topic_score_codex":0.002757478,"about_ca_topic_score_gemma":0.0024069713,"teacher_disagreement_score":0.002757478,"about_ca_system_score_codex":0.0008724694,"about_ca_system_score_gemma":0.0005984611,"threshold_uncertainty_score":0.006330192},"labels":[],"label_agreement":null},{"id":"W2098429327","doi":"10.1093/bioinformatics/btn524","title":"The Burkholderia Genome Database: facilitating flexible queries and comparative analyses","year":2008,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Cystic Fibrosis Research Advances","field":"Medicine","cited_by":252,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"Canadian Institutes of Health Research; Cystic Fibrosis Foundation Therapeutics; Michael Smith Health Research BC; Cystic Fibrosis Foundation","keywords":"Computer science; Burkholderia; Genome; Database; Computational biology; World Wide Web; Biology; Genetics; Gene; Bacteria","score_opus":0.180599334328155,"score_gpt":0.4012624785675375,"score_spread":0.2206631442393825,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2098429327","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.02834185,0.0072975615,0.4159136,0.00693017,0.001164632,0.001143277,0.28420904,0.22511303,0.029886879],"genre_scores_gemma":[0.05986868,0.0034064306,0.4657665,0.0008484235,0.00044992918,0.0008163264,0.4431426,0.020068875,0.0056323055],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99792045,0.00065390655,0.00024909127,0.0004427917,0.0005507889,0.00018297098],"domain_scores_gemma":[0.9956462,0.0017983253,0.00041123666,0.00086139195,0.00069535733,0.0005873631],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003972836,0.0015397304,0.0018531448,0.0035276753,0.0011912266,0.003027992,0.003674282,0.0017969933,0.03196827],"category_scores_gemma":[0.0150981685,0.0009491548,0.00075461884,0.007381297,0.00058591843,0.0035206303,0.0035000609,0.0023201515,0.039307956],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.002495052,0.0002621104,0.0060269674,0.0038079417,0.00017510216,0.0010819084,0.00079697167,0.003571694,0.07528336,0.009808947,0.56315535,0.33353466],"study_design_scores_gemma":[0.0009001832,0.00021449049,0.012050556,0.0009621391,0.00016621807,0.0015093832,0.00043045898,0.03292491,0.03559279,0.02384571,0.8911102,0.0002930787],"about_ca_topic_score_codex":0.0022291797,"about_ca_topic_score_gemma":0.003183352,"teacher_disagreement_score":0.03196827,"about_ca_system_score_codex":0.0006802169,"about_ca_system_score_gemma":0.0015425899,"threshold_uncertainty_score":0.1069445},"labels":[],"label_agreement":null},{"id":"W2098566726","doi":"10.1093/bioinformatics/bth124","title":"Evaluation and optimization of clustering in gene expression data analysis","year":2004,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":49,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"National Research Council Canada","funders":"","keywords":"Cluster analysis; Computer science; Partition (number theory); Identification (biology); Data mining; Stability (learning theory); Cluster (spacecraft); Expression (computer science); Gene; Computational biology; Consensus clustering; Artificial intelligence; Pattern recognition (psychology); Machine learning; Biology; Genetics; Correlation clustering; Mathematics; CURE data clustering algorithm","score_opus":0.0391304351557178,"score_gpt":0.3075528175625202,"score_spread":0.2684223824068024,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2098566726","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.041800003,0.0008051412,0.95494115,0.00026064104,0.00003056748,0.0002102206,0.00027321983,0.0011159648,0.0005631367],"genre_scores_gemma":[0.34913006,0.0003517692,0.6475092,0.00010493754,0.00006667076,0.00037562277,0.0012922959,0.0004846297,0.0006848402],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99013025,0.005061425,0.0008618169,0.0014643064,0.0022207196,0.00026154157],"domain_scores_gemma":[0.9717856,0.017746048,0.0023288222,0.0018106314,0.0058929734,0.00043584994],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.014988339,0.0015430474,0.002186423,0.0045636804,0.001356128,0.0022828074,0.002095489,0.0017114067,0.00088298024],"category_scores_gemma":[0.04244475,0.00073613174,0.001424901,0.0040648817,0.0020856755,0.0018306908,0.001955653,0.0013668551,0.00055622857],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00076665945,0.0002222712,0.022350162,0.00091183477,0.00051853206,0.00013418392,0.00051294523,0.6540633,0.014386939,0.008863596,0.0033167352,0.2939529],"study_design_scores_gemma":[0.000026317439,0.00010639368,0.003559785,0.00003326674,0.000047951045,0.00007582245,0.00006335281,0.97935337,0.008881023,0.0069679995,0.00085589744,0.000028799137],"about_ca_topic_score_codex":0.0050018476,"about_ca_topic_score_gemma":0.00331139,"teacher_disagreement_score":0.014988339,"about_ca_system_score_codex":0.0023649458,"about_ca_system_score_gemma":0.0021183235,"threshold_uncertainty_score":0.079266846},"labels":[],"label_agreement":null},{"id":"W2099377894","doi":"10.1093/bioinformatics/bts053","title":"JointSNVMix: a probabilistic model for accurate detection of somatic mutations in normal/tumour paired next-generation sequencing data","year":2012,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Cancer Genomics and Diagnostics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":175,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canada's Michael Smith Genome Sciences Centre; University of British Columbia; BC Cancer Agency","funders":"Canadian Institutes of Health Research; Michael Smith Health Research BC","keywords":"Somatic cell; DNA sequencing; Computational biology; Probabilistic logic; Computer science; Genetics; Biology; Artificial intelligence; Gene","score_opus":0.1153016910058751,"score_gpt":0.28669141719358626,"score_spread":0.17138972618771114,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2099377894","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.009931353,0.0003087054,0.98517376,0.00034359022,0.00005327258,0.000113477814,0.0016025555,0.0017765176,0.0006967659],"genre_scores_gemma":[0.38735226,0.0010135728,0.59111214,0.0007225213,0.00029663753,0.0013823695,0.008895435,0.0011545428,0.008070576],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9976042,0.0009518173,0.00011549564,0.0007058656,0.0004682385,0.00015453108],"domain_scores_gemma":[0.99443674,0.0041325283,0.00046500526,0.00040700997,0.00036487335,0.00019373512],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005441641,0.0012026104,0.0016466171,0.0015536094,0.0006492714,0.0023582748,0.0033604933,0.0020956954,0.0049798116],"category_scores_gemma":[0.014988678,0.0013035869,0.002471124,0.0013478717,0.0016348035,0.0018360504,0.0021930994,0.0025643993,0.0016340992],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005646052,0.00009869419,0.008508849,0.0003140076,0.00032577955,0.00027940652,0.00022773808,0.845834,0.0040657255,0.050832197,0.00710437,0.081844665],"study_design_scores_gemma":[0.000023265986,0.000023822904,0.0005418039,0.000014446689,0.00002962855,0.000076293145,0.000008209947,0.97069764,0.00068287685,0.025651557,0.0022277876,0.00002253716],"about_ca_topic_score_codex":0.009592495,"about_ca_topic_score_gemma":0.010895213,"teacher_disagreement_score":0.009592495,"about_ca_system_score_codex":0.0015501914,"about_ca_system_score_gemma":0.0018834083,"threshold_uncertainty_score":0.028778493},"labels":[],"label_agreement":null},{"id":"W2099432911","doi":"10.1093/bioinformatics/btm487","title":"MDQC: a new quality assessment method for microarrays based on quality control reports","year":2007,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":44,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Vancouver Coastal Health Research Institute; University of British Columbia; Vancouver Coastal Health; Providence Health Care","funders":"","keywords":"Mahalanobis distance; Computer science; Quality (philosophy); Data mining; Bioconductor; Outlier; Quality assurance; Data quality; Process (computing); Multivariate statistics; Control (management); Quality Score; Artificial intelligence; Metric (unit); Machine learning; Engineering","score_opus":0.044256724069191226,"score_gpt":0.394229669492128,"score_spread":0.3499729454229368,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2099432911","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00286047,0.00044805015,0.98488927,0.0003568735,0.00012374454,0.0002487511,0.0016173867,0.008807944,0.00064759265],"genre_scores_gemma":[0.03658847,0.0003609452,0.95550275,0.0002287995,0.00014339552,0.001001539,0.003708217,0.0015206531,0.00094526407],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.97685915,0.0061676805,0.0025635648,0.0031449797,0.010778302,0.0004862699],"domain_scores_gemma":[0.92004067,0.034714263,0.011031813,0.010566722,0.022473237,0.0011732989],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.024908287,0.0020046884,0.0020131078,0.008198143,0.0015922285,0.00665419,0.0042703445,0.0017733122,0.0048295613],"category_scores_gemma":[0.10125644,0.0012136145,0.0023675594,0.0070792497,0.0017923018,0.003711556,0.003152705,0.0037407242,0.0018578131],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011450744,0.00026385408,0.025242249,0.0024232767,0.00076595275,0.00047291743,0.0010578335,0.090287045,0.038641375,0.055889256,0.05799294,0.7258182],"study_design_scores_gemma":[0.00025505637,0.00032931345,0.011137127,0.0003617891,0.00031726406,0.00082584214,0.0002768109,0.76603884,0.06274054,0.060363725,0.09680304,0.0005505609],"about_ca_topic_score_codex":0.005592709,"about_ca_topic_score_gemma":0.0041587395,"teacher_disagreement_score":0.024908287,"about_ca_system_score_codex":0.0035102088,"about_ca_system_score_gemma":0.0054925424,"threshold_uncertainty_score":0.13172919},"labels":[],"label_agreement":null},{"id":"W2099999454","doi":"10.1093/bioinformatics/btv143","title":"Probing the binding affinity of amyloids to reduce toxicity of oligomers in diabetes","year":2015,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Alzheimer's disease research and treatments","field":"Medicine","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"McGill University","funders":"Canadian Institutes of Health Research; McGill University","keywords":"Amylin; Amyloid (mycology); Amyloid fibril; Chemistry; Membrane; Toxicity; Amyloid disease; Fibril; Biophysics; Biochemistry; Insulin; Biology; Amyloid β; Medicine; Islet","score_opus":0.056556154752666836,"score_gpt":0.32062946861408237,"score_spread":0.2640733138614155,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2099999454","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6492817,0.0006012219,0.3383885,0.0010587209,0.000043910593,0.00011749158,0.0012999917,0.0025368463,0.006671563],"genre_scores_gemma":[0.83327526,0.0002822075,0.16351312,0.00026362945,0.000015079046,0.0001599596,0.00077620795,0.000240041,0.0014743767],"study_design_codex":"simulation_or_modeling","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9998099,0.00007079604,0.000008467163,0.000035261346,0.00004913027,0.000026456795],"domain_scores_gemma":[0.9988337,0.000767813,0.0001321358,0.00006205613,0.0001448008,0.000059409642],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008720991,0.00054345804,0.00059619214,0.00045141886,0.0004404442,0.00041698827,0.0010606869,0.0007910406,0.0026126204],"category_scores_gemma":[0.002710799,0.00030190137,0.00050436356,0.00044205898,0.00034008778,0.0005905033,0.00045281995,0.00070061395,0.00036368662],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00014465314,0.00023597799,0.0092556635,0.00018460723,0.00006127721,0.00007901982,0.000044273205,0.9520201,0.017313754,0.0047950116,0.0014147997,0.014450946],"study_design_scores_gemma":[0.000030233698,0.000050067574,0.0008110514,0.000006219851,0.000013896038,0.00002654017,0.000008934046,0.9887417,0.0067335065,0.0031951882,0.00037421982,0.00000849994],"about_ca_topic_score_codex":0.0070939357,"about_ca_topic_score_gemma":0.006464421,"teacher_disagreement_score":0.0070939357,"about_ca_system_score_codex":0.0008159237,"about_ca_system_score_gemma":0.00123927,"threshold_uncertainty_score":0.01410526},"labels":[],"label_agreement":null},{"id":"W2100033550","doi":"10.1093/bioinformatics/16.12.1105","title":"Saturated BLAST: an automated multiple intermediate sequence search used to detect distant homology","year":2000,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Protein Structure and Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":65,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"National Institute of General Medical Sciences; National Institutes of Health","keywords":"Perl; Computer science; False positive paradox; Sequence (biology); Sequence database; Cluster analysis; Parsing; Data mining; Multiple sequence alignment; Sequence alignment; Artificial intelligence; Programming language; Biology; Peptide sequence; Gene; Genetics","score_opus":0.014202928386738112,"score_gpt":0.279582879899132,"score_spread":0.2653799515123939,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2100033550","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.055199705,0.0022011518,0.81822747,0.00042522652,0.0002161712,0.00096721464,0.010381488,0.10513719,0.0072443658],"genre_scores_gemma":[0.09663751,0.0007969025,0.8731001,0.00022887386,0.000068751135,0.00092997437,0.022074146,0.00350489,0.0026588067],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9985966,0.0004942541,0.00015237468,0.00030800674,0.00034156506,0.00010727198],"domain_scores_gemma":[0.99863225,0.00067106367,0.00021748197,0.00014908862,0.00021437337,0.00011567389],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0020435308,0.0017870333,0.0018180921,0.0033858276,0.001238976,0.0012213145,0.0023883632,0.0009569643,0.0070970533],"category_scores_gemma":[0.0042641247,0.00096906105,0.0010122036,0.0032190199,0.00073540147,0.0017143339,0.0016148138,0.0010390616,0.006317297],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0059906985,0.0006931312,0.01320849,0.0041887835,0.0008990076,0.0032595238,0.001184627,0.014811359,0.25940982,0.012738501,0.16448307,0.5191331],"study_design_scores_gemma":[0.0013565937,0.00203533,0.020279028,0.0006794434,0.00076014956,0.010883491,0.0006521761,0.3936334,0.28199056,0.02635558,0.26087698,0.0004972422],"about_ca_topic_score_codex":0.0010368535,"about_ca_topic_score_gemma":0.0012169826,"teacher_disagreement_score":0.0070970533,"about_ca_system_score_codex":0.00048702423,"about_ca_system_score_gemma":0.0010534774,"threshold_uncertainty_score":0.02374202},"labels":[],"label_agreement":null},{"id":"W2100148038","doi":"10.1093/bioinformatics/btt151","title":"FishingCNV: a graphical software package for detecting rare copy number variations in exome-sequencing data","year":2013,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomic variations and chromosomal abnormalities","field":"Biochemistry, Genetics and Molecular Biology","cited_by":51,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University and Génome Québec Innovation Centre","funders":"Canadian Institutes of Health Research","keywords":"Computer science; Software; Scripting language; Java; Copy-number variation; Graphical user interface; Pipeline (software); Exome sequencing; Exome; Data mining; Programming language; Mutation; Biology; Genetics; Gene; Genome","score_opus":0.025449746610317375,"score_gpt":0.25206958009952374,"score_spread":0.22661983348920636,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2100148038","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0038709748,0.00035069833,0.42172682,0.00032888804,0.00019406635,0.00048801652,0.116822496,0.45294607,0.0032720135],"genre_scores_gemma":[0.03132195,0.000494174,0.65030175,0.000843739,0.00011146382,0.0040948233,0.18011686,0.12333319,0.009382011],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99818367,0.00044124655,0.00021742779,0.0005134497,0.00050990784,0.00013427185],"domain_scores_gemma":[0.9956066,0.0030616883,0.00035491792,0.00036946632,0.00047353897,0.00013378554],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004302676,0.002906239,0.0021204443,0.0061644823,0.00077490497,0.0016839058,0.0036244171,0.0010551754,0.09989358],"category_scores_gemma":[0.011555747,0.0017150734,0.0022488667,0.003370403,0.00056919025,0.0018823328,0.0027753857,0.0019948778,0.025743373],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010318689,0.00013822924,0.008519055,0.004360974,0.0011694988,0.00075837236,0.0005751807,0.0074513853,0.018985953,0.0070303595,0.75095564,0.19902343],"study_design_scores_gemma":[0.0013913432,0.0003070814,0.030385876,0.00088373973,0.00070361956,0.002113411,0.0001925504,0.17751911,0.051626824,0.04463706,0.68959934,0.0006400096],"about_ca_topic_score_codex":0.004606333,"about_ca_topic_score_gemma":0.0050391653,"teacher_disagreement_score":0.09989358,"about_ca_system_score_codex":0.0009447563,"about_ca_system_score_gemma":0.002007266,"threshold_uncertainty_score":0.33417726},"labels":[],"label_agreement":null},{"id":"W2100266218","doi":"10.1093/bioinformatics/btu849","title":"An automatic tool to analyze and cluster macromolecular conformations based on self-organizing maps","year":2014,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Neural Networks and Applications","field":"Computer Science","cited_by":39,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canadian Nautical Research Society","funders":"","keywords":"Computer science; Cluster (spacecraft); Data mining; Artificial intelligence; Programming language","score_opus":0.006141447108255647,"score_gpt":0.2239100227889064,"score_spread":0.21776857568065075,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2100266218","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.005922677,0.00016911136,0.7476628,0.00020627058,0.00011082452,0.00030681555,0.013463916,0.22944187,0.002715802],"genre_scores_gemma":[0.04666815,0.0002598664,0.9177896,0.0001682382,0.000058619655,0.0012721487,0.017325498,0.012007549,0.004450442],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9995073,0.00005471,0.00004087863,0.00012277439,0.00022698303,0.000047309375],"domain_scores_gemma":[0.99908173,0.00035555364,0.00010721493,0.0001336279,0.00023476615,0.000087125154],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009639238,0.001979662,0.0011648635,0.0038823849,0.0009875506,0.001799746,0.0026692639,0.0009987968,0.022398613],"category_scores_gemma":[0.0034922548,0.0010246983,0.0016484838,0.0023862305,0.00053462305,0.0015566841,0.00217398,0.0013302827,0.009929393],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005615108,0.00038467927,0.006137539,0.0018737385,0.00053902925,0.00083270523,0.0005957305,0.036927115,0.025862014,0.015026526,0.28733706,0.6239224],"study_design_scores_gemma":[0.0002564344,0.00010093338,0.0049975323,0.00020544002,0.000098282464,0.00077631394,0.00017492128,0.79771376,0.04313335,0.04553824,0.106804,0.00020075183],"about_ca_topic_score_codex":0.002899042,"about_ca_topic_score_gemma":0.0043413597,"teacher_disagreement_score":0.022398613,"about_ca_system_score_codex":0.0006982709,"about_ca_system_score_gemma":0.0016936961,"threshold_uncertainty_score":0.07493085},"labels":[],"label_agreement":null},{"id":"W2100922775","doi":"10.1093/bioinformatics/btv157","title":"Test set bias affects reproducibility of gene signatures","year":2015,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":123,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Princess Margaret Cancer Centre; University of Toronto; University Health Network; Montreal Clinical Research Institute","funders":"National Institute of General Medical Sciences; National Cancer Institute; National Center for Advancing Translational Sciences; BC Cancer Agency","keywords":"Normalization (sociology); Computer science; Data set; Data mining; Set (abstract data type); Test set; Reproducibility; Sample size determination; Population; Test data; Statistics; Pattern recognition (psychology); Artificial intelligence; Mathematics; Medicine","score_opus":0.0516341996318329,"score_gpt":0.29076790735133456,"score_spread":0.23913370771950165,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2100922775","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.46172822,0.00338188,0.49509585,0.0052362727,0.0018935128,0.0008295311,0.007354883,0.010199733,0.014280083],"genre_scores_gemma":[0.9439228,0.00017440219,0.04390522,0.0013774899,0.00023516102,0.0005967413,0.0065294257,0.0016004313,0.0016582933],"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9034208,0.05378648,0.0074113943,0.017789653,0.015662596,0.0019291152],"domain_scores_gemma":[0.5654068,0.33472288,0.012898583,0.065363936,0.019888418,0.0017193567],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.08116712,0.0013678279,0.0017207445,0.0019881923,0.0023859695,0.0048703584,0.003172259,0.0019430491,0.00330984],"category_scores_gemma":[0.32254252,0.0006377353,0.0016756768,0.0030817927,0.005795214,0.0028914008,0.0040257787,0.0032606032,0.002678828],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0061414987,0.00065886404,0.46025243,0.0021135558,0.0036105819,0.001086824,0.0023195308,0.06442279,0.06282516,0.01599567,0.03212433,0.34844878],"study_design_scores_gemma":[0.0006794132,0.0021825982,0.29590747,0.00050769246,0.0013051771,0.0029903466,0.0012940274,0.29541135,0.24356675,0.12173128,0.034012355,0.00041156478],"about_ca_topic_score_codex":0.0021528096,"about_ca_topic_score_gemma":0.0021096026,"teacher_disagreement_score":0.9188329,"about_ca_system_score_codex":0.0019871085,"about_ca_system_score_gemma":0.002312413,"threshold_uncertainty_score":0.42925787},"labels":[],"label_agreement":null},{"id":"W2101347991","doi":"10.1093/bioinformatics/btm585","title":"Choosing BLAST options for better detection of orthologs as reciprocal best hits","year":2007,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":535,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Wilfrid Laurier University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Reciprocal; Genome; Matching (statistics); Computational biology; Computer science; Gene; Biology; Genetics; Mathematics; Statistics","score_opus":0.014556145587829139,"score_gpt":0.2589700547250416,"score_spread":0.24441390913721245,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2101347991","genre_codex":"empirical","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6028555,0.0041005053,0.34752783,0.0015087809,0.00042034284,0.0010396201,0.009453053,0.028080095,0.0050142054],"genre_scores_gemma":[0.34634456,0.0006443936,0.6373175,0.00058926706,0.000061287166,0.0009814526,0.009995043,0.0034375694,0.0006289431],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.98879015,0.005271143,0.002373447,0.0015807578,0.0014076947,0.00057685096],"domain_scores_gemma":[0.97613424,0.017236998,0.0016481188,0.0018050243,0.0024020842,0.00077357976],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.015401445,0.0026244372,0.003067904,0.0054933657,0.0016392088,0.002498676,0.0022414608,0.0030021716,0.0078343395],"category_scores_gemma":[0.047464684,0.0010566476,0.0015222643,0.004567757,0.0007438138,0.003936914,0.0020757872,0.0022181103,0.0039715283],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.01265645,0.0010416722,0.044397626,0.0073031276,0.00070836913,0.0032170266,0.0016130041,0.006637349,0.77406496,0.004615853,0.013807267,0.12993722],"study_design_scores_gemma":[0.0012796861,0.0034194563,0.06599035,0.0016390997,0.0012141358,0.008095636,0.0024344337,0.12571627,0.7232352,0.016090523,0.04997454,0.0009107382],"about_ca_topic_score_codex":0.00032761865,"about_ca_topic_score_gemma":0.00056974933,"teacher_disagreement_score":0.015401445,"about_ca_system_score_codex":0.00043445983,"about_ca_system_score_gemma":0.00069301215,"threshold_uncertainty_score":0.081451535},"labels":[],"label_agreement":null},{"id":"W2101456936","doi":"10.1093/bioinformatics/btn645","title":"Predicting the binding preference of transcription factors to individual DNA <i>k</i>-mers","year":2008,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":28,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"Howard Hughes Medical Institute","keywords":"Inference; Computational biology; Biology; Transcription factor; DNA; DNA sequencing; DNA binding site; Preference; Gene; Genetics; Mechanism (biology); Computer science; Gene expression; Artificial intelligence; Mathematics; Promoter","score_opus":0.028418969096521675,"score_gpt":0.21702661677774326,"score_spread":0.1886076476812216,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2101456936","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9542553,0.00045723066,0.03834361,0.00009892334,0.000009459404,0.000035172845,0.00544989,0.00080103497,0.00054943236],"genre_scores_gemma":[0.93546975,0.00017072671,0.04940554,0.000048305585,0.000013900297,0.00004057173,0.014343785,0.00007130658,0.00043615038],"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9996402,0.000060002058,0.000033294917,0.00017590921,0.000049594717,0.00004104643],"domain_scores_gemma":[0.99861896,0.00085869315,0.00017266288,0.00009448513,0.00015264077,0.000102582395],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008648748,0.00061783363,0.00082095864,0.0012691105,0.00030907235,0.0004778404,0.00056109886,0.00073978823,0.0010439045],"category_scores_gemma":[0.0025597275,0.00020437727,0.0006782561,0.0008607789,0.0002368664,0.0004970682,0.00024335904,0.0005527104,0.0008559077],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00422148,0.00062742626,0.41863763,0.00112015,0.0005690409,0.00034736088,0.00018041524,0.23727708,0.2054892,0.00094337243,0.005889475,0.124697335],"study_design_scores_gemma":[0.000088124994,0.00027235996,0.102581814,0.000025433934,0.00013510391,0.00061156024,0.000105615945,0.83789605,0.05348423,0.0025465474,0.0022046887,0.00004846706],"about_ca_topic_score_codex":0.0038537357,"about_ca_topic_score_gemma":0.006498854,"teacher_disagreement_score":0.0038537357,"about_ca_system_score_codex":0.00045952422,"about_ca_system_score_gemma":0.0004251056,"threshold_uncertainty_score":0.0076625943},"labels":[],"label_agreement":null},{"id":"W2101497394","doi":"10.1093/bioinformatics/bts226","title":"Towards 3D structure prediction of large RNA molecules: an integer programming framework to insert local 3D motifs in RNA secondary structure","year":2012,"lang":"en","type":"article","venue":"Bioinformatics","topic":"RNA and protein synthesis mechanisms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":45,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Université de Montréal; McGill University","funders":"National Institute of General Medical Sciences; Nepal Academy of Science and Technology","keywords":"RNA; Insert (composites); Nucleic acid structure; Nucleic acid secondary structure; Protein secondary structure; Computational biology; Computer science; Integer programming; Structural motif; Algorithm; Biology; Genetics; Engineering; Structural engineering; Gene; Biochemistry","score_opus":0.008293551987385947,"score_gpt":0.24386110810428036,"score_spread":0.2355675561168944,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2101497394","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.014188365,0.00008595142,0.9823314,0.00012932626,0.000018061866,0.000040545285,0.00015832136,0.0019440295,0.0011039545],"genre_scores_gemma":[0.10439471,0.00013183171,0.8931358,0.00010400191,0.000034997447,0.00023404963,0.0005817026,0.00045281617,0.0009301247],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99972886,0.000090183945,0.0000121000685,0.000054961776,0.00008260281,0.000031244977],"domain_scores_gemma":[0.9993113,0.00045162425,0.000066152934,0.00003150361,0.000087049484,0.00005233382],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008554144,0.0010796285,0.0006832435,0.0006909779,0.00040720552,0.00093138526,0.0012922247,0.0010466439,0.0027049875],"category_scores_gemma":[0.0020115832,0.0004986811,0.0010450756,0.00052269216,0.0007362026,0.0005687084,0.00087193877,0.001556119,0.00085403427],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006569177,0.00007150883,0.000749388,0.0000958549,0.000019727313,0.00008276175,0.000041790532,0.95627755,0.0030504735,0.0074793613,0.0012520443,0.030813893],"study_design_scores_gemma":[0.0000034439076,0.0000066121042,0.00001755071,0.0000031578402,0.0000011492164,0.000004641649,0.0000031531454,0.99791604,0.00040461717,0.001404373,0.00023348405,0.0000016754536],"about_ca_topic_score_codex":0.0031620236,"about_ca_topic_score_gemma":0.0032994966,"teacher_disagreement_score":0.0031620236,"about_ca_system_score_codex":0.00049288187,"about_ca_system_score_gemma":0.0010536676,"threshold_uncertainty_score":0.009049058},"labels":[],"label_agreement":null},{"id":"W2101763464","doi":"10.1093/bioinformatics/bts072","title":"The identification of short linear motif-mediated interfaces within the human interactome","year":2012,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":72,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Lunenfeld-Tanenbaum Research Institute; University of Toronto; Mount Sinai Hospital","funders":"","keywords":"Interactome; Motif (music); Identification (biology); Computer science; Computational biology; Biology; Genetics; Gene; Ecology","score_opus":0.014387946493191764,"score_gpt":0.2640914391990048,"score_spread":0.24970349270581305,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2101763464","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6467756,0.0069551463,0.2816185,0.0011444794,0.00008905121,0.0003092562,0.04784883,0.007056708,0.008202343],"genre_scores_gemma":[0.6897176,0.0014334377,0.24321215,0.00037511505,0.00010082315,0.00028517807,0.062285516,0.000678206,0.0019120169],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99933416,0.00011365547,0.000052460793,0.00025909577,0.00017766104,0.00006298502],"domain_scores_gemma":[0.9984072,0.0007856345,0.00036432163,0.00016880389,0.0001536477,0.00012040905],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007499597,0.00052192673,0.00050862075,0.002404606,0.0004216209,0.0009255485,0.00046548343,0.0005220459,0.0034305903],"category_scores_gemma":[0.003390884,0.00018106936,0.0005737766,0.0022335302,0.00037815954,0.00083041645,0.0012529261,0.00069932884,0.0022127314],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0026387929,0.00037533234,0.15726109,0.0036426836,0.0005316458,0.0028541489,0.0010807824,0.014073994,0.4988341,0.010255359,0.031873547,0.2765786],"study_design_scores_gemma":[0.00025073666,0.00069672323,0.30667052,0.00071870664,0.00079358794,0.011603337,0.0009475295,0.21739142,0.3176272,0.039123915,0.103933364,0.00024291743],"about_ca_topic_score_codex":0.00061684835,"about_ca_topic_score_gemma":0.0011581619,"teacher_disagreement_score":0.0034305903,"about_ca_system_score_codex":0.00032270775,"about_ca_system_score_gemma":0.0004162032,"threshold_uncertainty_score":0.011476517},"labels":[],"label_agreement":null},{"id":"W2101793150","doi":"10.1093/bioinformatics/btn208","title":"Cytoscape ESP: simple search of complex biological networks","year":2008,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"U.S. National Library of Medicine; National Institute on Drug Abuse; National Institutes of Health; Genome Canada; Ontario Genomics; Israel Science Foundation; National Institute of General Medical Sciences; Ontario Genomics Institute","keywords":"Plug-in; Computer science; Biological network; Visualization; Software; Simple (philosophy); Syntax; Data mining; Information retrieval; Theoretical computer science; Programming language; Computational biology; Artificial intelligence; Biology","score_opus":0.03854118255699446,"score_gpt":0.25912986841015606,"score_spread":0.2205886858531616,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2101793150","genre_codex":"software","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.022035975,0.0016756116,0.39825734,0.0010263079,0.00052539544,0.00073163427,0.13489226,0.42437944,0.016476016],"genre_scores_gemma":[0.13441673,0.0016556536,0.565494,0.0011593321,0.00025994438,0.0033878693,0.23946321,0.037763983,0.016399251],"study_design_codex":"not_applicable","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9993788,0.0001022141,0.00005744391,0.00015688343,0.00026249388,0.00004219767],"domain_scores_gemma":[0.99847347,0.0008424057,0.00009543504,0.0002402387,0.00020180548,0.00014669147],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010683632,0.0016672243,0.0010965844,0.003802789,0.00086950744,0.0017110459,0.0023582422,0.0011065255,0.044704486],"category_scores_gemma":[0.0053428886,0.001076953,0.0011422471,0.0028162596,0.000337747,0.0020639866,0.0024142277,0.0014321983,0.014390462],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009452159,0.0001973637,0.004017863,0.0047384864,0.0008330916,0.0013075303,0.00038039,0.0076801,0.04935768,0.015469748,0.713376,0.2016966],"study_design_scores_gemma":[0.0012300819,0.0002574081,0.008732221,0.00049868424,0.00039212598,0.0032647327,0.00023382109,0.17763476,0.124032035,0.08348363,0.599827,0.0004135156],"about_ca_topic_score_codex":0.0019702418,"about_ca_topic_score_gemma":0.0049035903,"teacher_disagreement_score":0.044704486,"about_ca_system_score_codex":0.00040331678,"about_ca_system_score_gemma":0.0013844221,"threshold_uncertainty_score":0.14955145},"labels":[],"label_agreement":null},{"id":"W2101963711","doi":"10.1093/bioinformatics/18.suppl_1.s303","title":"Multiple genome rearrangement: a general approach via the evolutionary genome graph","year":2002,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genome Rearrangement Algorithms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of New Brunswick","funders":"","keywords":"Genome; Phylogenetic tree; Graph; Set (abstract data type); Computer science; Theoretical computer science; Biology; Computational biology; Evolutionary biology; Gene; Genetics","score_opus":0.015668192281693593,"score_gpt":0.2004549549426579,"score_spread":0.1847867626609643,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2101963711","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0039517535,0.00033351555,0.9933113,0.00038546932,0.000020570707,0.000056660992,0.00007573176,0.0002549369,0.0016100804],"genre_scores_gemma":[0.09437185,0.0011591429,0.90006256,0.00023124831,0.00009136697,0.00017358093,0.0005955194,0.00020231502,0.0031124938],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9985714,0.0005533349,0.000065801294,0.00039945095,0.00033063104,0.00007933745],"domain_scores_gemma":[0.99808073,0.0010658118,0.00012440138,0.0004057577,0.00022275084,0.00010050015],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014006697,0.0008058835,0.0010198321,0.0029466166,0.0009776349,0.0021877976,0.0031978257,0.0017549829,0.0033029465],"category_scores_gemma":[0.0050147716,0.00048723276,0.001556814,0.003906608,0.0021267405,0.0056075817,0.0023484812,0.0020674171,0.0007582707],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007318662,0.00007408492,0.0011565005,0.00039904576,0.000084159554,0.0002890915,0.00046290405,0.18399142,0.0036416303,0.6280483,0.0036153723,0.17816429],"study_design_scores_gemma":[0.000018294653,0.000034381374,0.00033069434,0.000037555128,0.00003621012,0.00035822034,0.0001299977,0.39882565,0.001425322,0.58433634,0.014438613,0.000028653905],"about_ca_topic_score_codex":0.0032949473,"about_ca_topic_score_gemma":0.0031368795,"teacher_disagreement_score":0.0033029465,"about_ca_system_score_codex":0.0016237629,"about_ca_system_score_gemma":0.001148261,"threshold_uncertainty_score":0.011781275},"labels":[],"label_agreement":null},{"id":"W2102039892","doi":"10.1093/bioinformatics/btn163","title":"Biomolecular network motif counting and discovery by color coding","year":2008,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":218,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Color-coding; Computer science; Combinatorics; Bounded function; Mathematics; Biology; Theoretical computer science","score_opus":0.007867447708500175,"score_gpt":0.2044838500234672,"score_spread":0.19661640231496702,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2102039892","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.100728,0.00032891333,0.89367735,0.00022648134,0.000033772514,0.0000747377,0.00048998784,0.0028545912,0.0015861705],"genre_scores_gemma":[0.41013566,0.00024097308,0.5860555,0.000121221296,0.000024772182,0.00013283547,0.0013741427,0.0002036192,0.0017112519],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9992816,0.00015211132,0.000033886347,0.00017290801,0.00025371488,0.00010584462],"domain_scores_gemma":[0.9967648,0.0014230151,0.00043398183,0.00047577496,0.00073599676,0.00016646627],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006511149,0.00045666297,0.0005968158,0.0028671012,0.0004914482,0.0010604433,0.0016415039,0.00080777716,0.0011660075],"category_scores_gemma":[0.006019038,0.00026099168,0.00049244886,0.0029466853,0.0006181087,0.0014227383,0.0010279288,0.00076425704,0.00038593574],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00045351865,0.00016785161,0.024563033,0.00028502298,0.00012255178,0.000417783,0.0003082221,0.21726803,0.06333224,0.05597301,0.010110794,0.62699807],"study_design_scores_gemma":[0.000010380438,0.00001840404,0.0013124143,0.0000071783497,0.000010888413,0.00018648339,0.000030500523,0.96638995,0.012526449,0.017748967,0.0017448673,0.000013411996],"about_ca_topic_score_codex":0.003977025,"about_ca_topic_score_gemma":0.003702747,"teacher_disagreement_score":0.003977025,"about_ca_system_score_codex":0.0011325998,"about_ca_system_score_gemma":0.0008048674,"threshold_uncertainty_score":0.008217573},"labels":[],"label_agreement":null},{"id":"W2102057828","doi":"10.1093/bioinformatics/btm032","title":"NetMatch: a Cytoscape plugin for searching biological networks","year":2007,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":83,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"Ministero dell’Istruzione, dell’Università e della Ricerca","keywords":"Plug-in; Computer science; Biological network; Software; Data mining; Computational biology; Programming language; Biology","score_opus":0.021569911012977377,"score_gpt":0.26628448893634754,"score_spread":0.24471457792337017,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2102057828","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0075935796,0.0032102594,0.44104046,0.001508089,0.0011129691,0.0005798215,0.14658912,0.38223752,0.016128164],"genre_scores_gemma":[0.061336175,0.0036480296,0.5351207,0.0017788134,0.00052192534,0.0026689197,0.31851628,0.050839864,0.025569223],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99908483,0.00013850704,0.000056238452,0.00025785793,0.00039968608,0.000062869454],"domain_scores_gemma":[0.9984366,0.0007370516,0.00014781738,0.00024083589,0.00020424432,0.00023341345],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010812929,0.0016091535,0.0013703124,0.0043938193,0.0012500117,0.0016917315,0.004116156,0.0018215273,0.061548144],"category_scores_gemma":[0.0051821955,0.0012327378,0.0010514135,0.003082808,0.00042716454,0.002832414,0.0025168303,0.0027090698,0.028214658],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00076450274,0.00015466283,0.0021456874,0.005110759,0.0005301453,0.0009243032,0.00031439873,0.005621275,0.049917553,0.01382819,0.7310092,0.18967932],"study_design_scores_gemma":[0.00040398235,0.00010636917,0.0036208378,0.00039203776,0.00022585821,0.0017444924,0.00014048352,0.05611268,0.07813604,0.058173526,0.8006466,0.00029713288],"about_ca_topic_score_codex":0.0027947673,"about_ca_topic_score_gemma":0.0056051626,"teacher_disagreement_score":0.061548144,"about_ca_system_score_codex":0.0005844939,"about_ca_system_score_gemma":0.0013757428,"threshold_uncertainty_score":0.20589906},"labels":[],"label_agreement":null},{"id":"W2102089979","doi":"10.1093/bioinformatics/btu453","title":"Effects of small particle numbers on long-term behaviour in discrete biochemical systems","year":2014,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Microtubule and mitosis dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":37,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"Deutsche Forschungsgemeinschaft","keywords":"Markup language; Computer science; Term (time); Theoretical computer science; Extension (predicate logic); Systems biology; Computational biology; Biology; XML; Programming language; World Wide Web; Physics","score_opus":0.005476390262512516,"score_gpt":0.22310761737710239,"score_spread":0.21763122711458988,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2102089979","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.80315864,0.0013320341,0.18593642,0.0011391476,0.00015266436,0.00008953747,0.00034086316,0.00072738976,0.00712328],"genre_scores_gemma":[0.9894778,0.00027365572,0.009005169,0.000061468185,0.00003444569,0.00006359673,0.00011206391,0.0001008368,0.00087092095],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9993181,0.00020425278,0.000049083395,0.0001461428,0.0001980662,0.00008428345],"domain_scores_gemma":[0.9888602,0.007737532,0.0015738674,0.00061580946,0.0004532677,0.0007593738],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001542034,0.0004965556,0.0006380557,0.00096320955,0.00086993875,0.0013199584,0.00096169196,0.0010999506,0.0027099783],"category_scores_gemma":[0.011550645,0.00040926988,0.0007643685,0.00034141334,0.0020916646,0.0019080552,0.001272473,0.0010566657,0.00026865082],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00038005543,0.00031649697,0.017201388,0.0007971084,0.00017305743,0.0016599237,0.00083685224,0.70162475,0.0683828,0.1901146,0.0022269955,0.016285965],"study_design_scores_gemma":[0.000041295432,0.000095443254,0.0035609296,0.000029267376,0.00005056665,0.00016127406,0.000061051,0.94200253,0.007587313,0.045426056,0.0009451586,0.00003912257],"about_ca_topic_score_codex":0.0020794065,"about_ca_topic_score_gemma":0.0012491292,"teacher_disagreement_score":0.0027099783,"about_ca_system_score_codex":0.001305771,"about_ca_system_score_gemma":0.00064699916,"threshold_uncertainty_score":0.009474099},"labels":[],"label_agreement":null},{"id":"W2102610257","doi":"10.1093/bioinformatics/bts272","title":"BreakFusion: targeted assembly-based identification of gene fusions in whole transcriptome paired-end sequencing data","year":2012,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":59,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canada's Michael Smith Genome Sciences Centre","funders":"National Cancer Institute; National Human Genome Research Institute; BC Cancer Agency","keywords":"Scalability; Computational biology; Sequence assembly; Computer science; Transcriptome; Software; Genome; Identification (biology); DNA sequencing; R package; Data mining; Gene; Biology; Genetics; Gene expression; Database; Computational science","score_opus":0.0369776133322311,"score_gpt":0.2615952537368206,"score_spread":0.22461764040458948,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2102610257","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.039549798,0.0002892661,0.8339009,0.00019587,0.00016016905,0.00021142978,0.019592568,0.1050431,0.0010568536],"genre_scores_gemma":[0.1407376,0.00024097245,0.7726243,0.00022853205,0.00007314989,0.00079916586,0.067051835,0.016822759,0.0014217444],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9982008,0.00033833127,0.0001504663,0.00071614387,0.0004616196,0.00013275917],"domain_scores_gemma":[0.9971661,0.0013999064,0.00042008684,0.0005501144,0.00029276544,0.00017100552],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004329645,0.001855867,0.0014525923,0.0022897338,0.0012128061,0.0019874743,0.0019775557,0.0011328949,0.009347187],"category_scores_gemma":[0.007890431,0.0009839723,0.0021924847,0.0020124058,0.0007211304,0.0015119833,0.0023012958,0.0022523701,0.005126289],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.004256949,0.00037335732,0.022014042,0.0033562253,0.002042218,0.0019371359,0.0011049433,0.03282444,0.5233736,0.009406521,0.13696122,0.26234925],"study_design_scores_gemma":[0.00067359774,0.0006627486,0.024604805,0.00018941639,0.0007002975,0.002241373,0.00027432456,0.43054274,0.46294352,0.017493837,0.0592993,0.00037400043],"about_ca_topic_score_codex":0.0008915238,"about_ca_topic_score_gemma":0.0017437697,"teacher_disagreement_score":0.009347187,"about_ca_system_score_codex":0.0005986688,"about_ca_system_score_gemma":0.001298431,"threshold_uncertainty_score":0.03126949},"labels":[],"label_agreement":null},{"id":"W2102803266","doi":"10.1093/bioinformatics/19.2.234","title":"A naive Bayes model to predict coupling between seventransmembrane domain receptors and G-proteins","year":2003,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Machine Learning in Bioinformatics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":47,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"AstraZeneca (Canada)","funders":"","keywords":"G protein-coupled receptor; Computational biology; Protein sequencing; Bayesian probability; Transmembrane domain; Bayes' theorem; Computer science; Coupling (piping); Biology; Artificial intelligence; Bioinformatics; Machine learning; Peptide sequence; Genetics; Receptor; Gene","score_opus":0.009458093025253722,"score_gpt":0.24074601573063606,"score_spread":0.23128792270538234,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2102803266","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.270357,0.0028742172,0.7118545,0.002874797,0.00037450198,0.0005159711,0.0020954802,0.0013018512,0.007751707],"genre_scores_gemma":[0.882888,0.0007767483,0.104032345,0.00093188643,0.00033647494,0.0005696394,0.0023495418,0.00008361853,0.008031766],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99862146,0.0005872672,0.00008395179,0.0003246258,0.00019818747,0.00018448984],"domain_scores_gemma":[0.9937191,0.0049331104,0.00033663082,0.00008873266,0.0007327381,0.00018963676],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004783319,0.0013846693,0.0021045133,0.0013436474,0.0009617602,0.0015160671,0.0021990759,0.0020633463,0.0043734247],"category_scores_gemma":[0.007034574,0.0008077479,0.0013811574,0.0008549115,0.0008648632,0.0011386881,0.00076678296,0.0017638325,0.0012161318],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00068092626,0.00019428396,0.011086517,0.00016406081,0.0001660172,0.00024313071,0.00015490007,0.9168395,0.0010506439,0.00840806,0.0032850506,0.057726808],"study_design_scores_gemma":[0.00003142745,0.000038173504,0.0004144406,0.000022934255,0.000020184974,0.000026425854,0.000010005407,0.9942257,0.00018474497,0.004778822,0.00023555825,0.000011565483],"about_ca_topic_score_codex":0.020886926,"about_ca_topic_score_gemma":0.013578355,"teacher_disagreement_score":0.020886926,"about_ca_system_score_codex":0.0018161106,"about_ca_system_score_gemma":0.0017920322,"threshold_uncertainty_score":0.04153073},"labels":[],"label_agreement":null},{"id":"W2103006829","doi":"10.1093/bioinformatics/btq122","title":"LOTUS, a new domain associated with small RNA pathways in the germline","year":2010,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Chromosomal and Genetic Variations","field":"Agricultural and Biological Sciences","cited_by":79,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Multidisciplinary University Research Initiative; Institute of Cancer Research; Université Pierre et Marie Curie; Institut National Du Cancer; Centre National de la Recherche Scientifique","keywords":"Piwi-interacting RNA; Biology; Germline; Argonaute; Genetics; Germ plasm; Lotus japonicus; Lotus; RasiRNA; RNA interference; RNA-binding protein; Germ cell; RNA; Computational biology; Cell biology; Gene; Embryo; Mutant; Botany","score_opus":0.02410917086021114,"score_gpt":0.19432815445272086,"score_spread":0.17021898359250973,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2103006829","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9561071,0.008293518,0.02823254,0.00017860759,0.00010817901,0.000060396334,0.0009742547,0.00082009425,0.005225409],"genre_scores_gemma":[0.9658847,0.0016070051,0.02084792,0.00019391901,0.000049059963,0.000054859942,0.0024879433,0.00007876541,0.008795897],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99995446,0.0000030762683,0.000002747464,0.00001660484,0.000013385097,0.000009733818],"domain_scores_gemma":[0.999928,0.00001002773,0.00002100455,0.000008921367,0.000007719495,0.000024289024],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00005032168,0.00023015622,0.00016979937,0.0003702983,0.00031753696,0.00036085982,0.00017031729,0.00027222754,0.0010114204],"category_scores_gemma":[0.000094209114,0.000114790986,0.0002494417,0.0002474345,0.00021713949,0.00050584,0.00041806174,0.00026868028,0.0005642376],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00018725611,0.000018236418,0.0037294754,0.00012942345,0.000014055101,0.000678595,0.0000664034,0.00017418586,0.9784203,0.0014783774,0.00033711965,0.0147664705],"study_design_scores_gemma":[0.00004856233,0.00042814048,0.06779205,0.00005277095,0.00010466639,0.01665419,0.0002004839,0.0060148067,0.7706086,0.0030347218,0.13498981,0.00007114041],"about_ca_topic_score_codex":0.00028562086,"about_ca_topic_score_gemma":0.0005179563,"teacher_disagreement_score":0.0010114204,"about_ca_system_score_codex":0.00022909633,"about_ca_system_score_gemma":0.0001875249,"threshold_uncertainty_score":0.0033835769},"labels":[],"label_agreement":null},{"id":"W2103145565","doi":"10.1093/bioinformatics/btt308","title":"Identification of deleterious synonymous variants in human genomes","year":2013,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Rare Diseases","field":"Biochemistry, Genetics and Molecular Biology","cited_by":91,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"SickKids Foundation; Hospital for Sick Children; University of Toronto","funders":"","keywords":"Biology; Identification (biology); Genome; In silico; Genetics; Computational biology; Genomics; Prioritization; Mutation; Human genome; 1000 Genomes Project; Single-nucleotide polymorphism; Gene; Genotype","score_opus":0.006778232057257463,"score_gpt":0.2260188908531121,"score_spread":0.21924065879585464,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2103145565","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.90489304,0.004974918,0.05566125,0.0009655077,0.00009023876,0.00012008604,0.02917125,0.0025849624,0.001538637],"genre_scores_gemma":[0.8398003,0.0017689997,0.1120529,0.00070434797,0.00010664511,0.000085718,0.04470212,0.00029547434,0.00048362243],"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9986094,0.0003025892,0.00016942454,0.0005281554,0.0003333316,0.000057133177],"domain_scores_gemma":[0.9959519,0.0023016369,0.00091157504,0.0002878311,0.00035695222,0.00019000615],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016207288,0.0005664091,0.000690738,0.0024772473,0.0007413508,0.0008816356,0.00060816726,0.0008479458,0.0013412423],"category_scores_gemma":[0.007592747,0.00028664767,0.0006747864,0.0032329746,0.0004595744,0.0004524616,0.00075180596,0.0004950558,0.0006068639],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013995711,0.00022558405,0.64255023,0.0028262676,0.0010588915,0.0027558072,0.00076623785,0.04425906,0.093313925,0.005624616,0.021878114,0.1833416],"study_design_scores_gemma":[0.00037938534,0.0006330381,0.5931715,0.00065638893,0.0011435013,0.013398376,0.0009234838,0.23107256,0.08210448,0.040741775,0.035555698,0.00021983593],"about_ca_topic_score_codex":0.0030015134,"about_ca_topic_score_gemma":0.006537651,"teacher_disagreement_score":0.0030015134,"about_ca_system_score_codex":0.00033785685,"about_ca_system_score_gemma":0.0006953487,"threshold_uncertainty_score":0.008571327},"labels":[],"label_agreement":null},{"id":"W2103173634","doi":"10.1093/bioinformatics/btl497","title":"Cell++—simulating biochemical pathways","year":2006,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene Regulatory Network Analysis","field":"Biochemistry, Genetics and Molecular Biology","cited_by":55,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Hospital for Sick Children; University of Toronto","funders":"Natural Sciences and Engineering Research Council of Canada; Hospital for Sick Children","keywords":"Computer science; Computational biology; Biology","score_opus":0.007429909915520843,"score_gpt":0.20236138685553787,"score_spread":0.19493147694001703,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2103173634","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.17508622,0.0015282244,0.735832,0.0014888962,0.0003977765,0.00038378264,0.008424358,0.01771848,0.05914032],"genre_scores_gemma":[0.53930473,0.0019111022,0.43324238,0.00043570655,0.000108863074,0.0012089451,0.006278451,0.0025821493,0.014927686],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99978715,0.000066476365,0.000011825817,0.000040323222,0.00007010512,0.00002412473],"domain_scores_gemma":[0.9993068,0.00040910696,0.000045658267,0.00006045975,0.000090610694,0.000087325636],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004904519,0.0008231241,0.0007822176,0.00044128217,0.0006017321,0.0009858828,0.0018001306,0.0010726874,0.008449621],"category_scores_gemma":[0.002051375,0.00041574475,0.00069606316,0.00083033624,0.00069398363,0.0008974154,0.0011346764,0.0012397954,0.0013794078],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000926889,0.000038202717,0.0011704119,0.00017791508,0.00003722026,0.000077865036,0.000056880315,0.95493835,0.0025569757,0.026288064,0.0039691594,0.0105962325],"study_design_scores_gemma":[0.000044947254,0.000011152632,0.00010193947,0.000008792757,0.0000074982268,0.00002887084,0.000007501074,0.9837647,0.0017864892,0.006027421,0.008202151,0.000008615061],"about_ca_topic_score_codex":0.0061753476,"about_ca_topic_score_gemma":0.0038093785,"teacher_disagreement_score":0.008449621,"about_ca_system_score_codex":0.0006370315,"about_ca_system_score_gemma":0.0012499784,"threshold_uncertainty_score":0.028266788},"labels":[],"label_agreement":null},{"id":"W2103365019","doi":"10.1093/bioinformatics/btq271","title":"SLIMS—a user-friendly sample operations and inventory management system for genotyping labs","year":2010,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Biomedical Text Mining and Ontologies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":25,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"Canadian Institutes of Health Research","keywords":"Computer science; User Friendly; Database; Java; Personalization; Interface (matter); Sample (material); Documentation; Source code; World Wide Web; User interface; Operating system","score_opus":0.013319885421856857,"score_gpt":0.2546115274215406,"score_spread":0.24129164199968373,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2103365019","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0028123453,0.000224964,0.15026543,0.0005288326,0.00019657014,0.0006991437,0.020750964,0.8176435,0.0068781893],"genre_scores_gemma":[0.07556572,0.0009098286,0.5662824,0.0045996164,0.00093249243,0.0060199434,0.17678635,0.13326387,0.03563968],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99669516,0.0006331492,0.00038186897,0.00069955154,0.0013252412,0.00026499262],"domain_scores_gemma":[0.9916945,0.0026431275,0.001151566,0.0015511585,0.0018069813,0.0011525922],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0048386385,0.002319237,0.0016114194,0.003906658,0.0009219825,0.0029849045,0.0051768464,0.0011283669,0.08722075],"category_scores_gemma":[0.012059731,0.0017377125,0.0010433681,0.0018292319,0.00068699033,0.0034196672,0.0051254234,0.0023235315,0.07698797],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014623064,0.00019772224,0.00467504,0.0009594897,0.00015367301,0.00044315332,0.00045193234,0.00165571,0.014263457,0.0027710109,0.7206998,0.2522668],"study_design_scores_gemma":[0.0010609882,0.00035894423,0.010841049,0.00058084814,0.00016521255,0.0010447925,0.00021555526,0.046463225,0.0471887,0.009287062,0.88229835,0.0004953804],"about_ca_topic_score_codex":0.0021217694,"about_ca_topic_score_gemma":0.0016352324,"teacher_disagreement_score":0.08722075,"about_ca_system_score_codex":0.0010414583,"about_ca_system_score_gemma":0.0021344735,"threshold_uncertainty_score":0.2917825},"labels":[],"label_agreement":null},{"id":"W2103991913","doi":"10.1093/bioinformatics/btg030","title":"<i>K</i>-ary clustering with optimal leaf ordering for gene expression data","year":2003,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":118,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Wilfrid Laurier University","funders":"Leverhulme Trust; Florida State University; Natural Sciences and Engineering Research Council of Canada; Burroughs Wellcome Fund","keywords":"Cluster analysis; Hierarchical clustering; Computer science; Tree (set theory); Binary tree; Pairwise comparison; Hierarchical clustering of networks; Single-linkage clustering; Robustness (evolution); CURE data clustering algorithm; Data mining; Algorithm; Nearest-neighbor chain algorithm; Correlation clustering; Canopy clustering algorithm; Tree structure; Mathematics; Artificial intelligence; Gene; Biology; Combinatorics; Genetics","score_opus":0.055061850156336696,"score_gpt":0.306350724116407,"score_spread":0.2512888739600703,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2103991913","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0062975967,0.0001145863,0.9838688,0.00017130915,0.000030390158,0.00011409299,0.00042088248,0.008386512,0.0005958102],"genre_scores_gemma":[0.024293406,0.000060835966,0.97310793,0.000059419115,0.0000144547985,0.00015367572,0.0011783591,0.00056860235,0.0005633194],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99793774,0.00044617016,0.00018854489,0.00050060335,0.0007726861,0.00015428054],"domain_scores_gemma":[0.99694866,0.0009310616,0.00032539913,0.0009291266,0.00068641175,0.00017933661],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0021153945,0.0010722531,0.0012031817,0.0024068283,0.0018573427,0.0016160012,0.0027395776,0.0018598372,0.004141501],"category_scores_gemma":[0.0102513395,0.0008082603,0.0012197417,0.0041939137,0.0012350102,0.0019335304,0.0021857354,0.0019722062,0.0054497398],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006282751,0.000357411,0.004625145,0.0006794899,0.00012865187,0.0002718981,0.00086226827,0.18105958,0.04641232,0.021830961,0.036616053,0.7065279],"study_design_scores_gemma":[0.00008747509,0.00008242814,0.0013178393,0.000039893373,0.00002166478,0.00021629201,0.000089672656,0.9379001,0.019475104,0.030690582,0.010027176,0.000051761876],"about_ca_topic_score_codex":0.0061216233,"about_ca_topic_score_gemma":0.008840216,"teacher_disagreement_score":0.0061216233,"about_ca_system_score_codex":0.0015504106,"about_ca_system_score_gemma":0.0024545384,"threshold_uncertainty_score":0.013854742},"labels":[],"label_agreement":null},{"id":"W2104040212","doi":"10.1093/bioinformatics/btg139","title":"GeneComber: combining outputs of gene prediction programs for improved results","year":2003,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":34,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia; University of British Columbia Hospital","funders":"","keywords":"Computer science; Interface (matter); Perl; Software; Frame (networking); User interface; Relational database management system; Web application; Database; Information retrieval; Relational database; Data mining; Interoperability; Application programming interface; Programming language; World Wide Web; Operating system","score_opus":0.013806031331283376,"score_gpt":0.22770177500039665,"score_spread":0.21389574366911326,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2104040212","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.090257816,0.0045167087,0.60273755,0.0022601103,0.0017869944,0.00040685048,0.020646563,0.26667494,0.010712425],"genre_scores_gemma":[0.2622645,0.0017722871,0.64482594,0.0010725114,0.00065989903,0.0006303274,0.04982045,0.03033504,0.0086190505],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99777514,0.00044628553,0.00015196415,0.0006428881,0.000805205,0.00017837576],"domain_scores_gemma":[0.996723,0.0014749058,0.00024787986,0.0006116881,0.00075272005,0.00018982204],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003155293,0.0026243124,0.0013945829,0.0016476725,0.0007952391,0.0020465332,0.002529299,0.0011946782,0.022712115],"category_scores_gemma":[0.0059756152,0.0006241061,0.0013197882,0.0024991196,0.0008039391,0.001774992,0.001796651,0.0013877585,0.01319409],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0051102177,0.00055194215,0.017669462,0.0034372965,0.00048639218,0.002113464,0.00046484842,0.011961031,0.28950855,0.0044954484,0.14802928,0.5161721],"study_design_scores_gemma":[0.00059374137,0.0008263578,0.015000706,0.00044720538,0.00045613668,0.0014280668,0.00019940622,0.21461199,0.60324657,0.0088542495,0.15398179,0.00035390846],"about_ca_topic_score_codex":0.0009401741,"about_ca_topic_score_gemma":0.0012523744,"teacher_disagreement_score":0.022712115,"about_ca_system_score_codex":0.0005123011,"about_ca_system_score_gemma":0.00068859686,"threshold_uncertainty_score":0.07597959},"labels":[],"label_agreement":null},{"id":"W2104342875","doi":"10.1093/bioinformatics/btm114","title":"A fast and flexible approach to oligonucleotide probe design for genomes and gene families","year":2007,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":43,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto; University Health Network","funders":"Canadian Institutes of Health Research; National Natural Science Foundation of China","keywords":"Genome; Oligonucleotide; Flexibility (engineering); Computer science; Selection (genetic algorithm); Software; Computational biology; Biology; Gene; DNA sequencing; Gene family; Metagenomics; Data mining; Genetics; Artificial intelligence","score_opus":0.026061594489059897,"score_gpt":0.25945074660245515,"score_spread":0.23338915211339525,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2104342875","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.006866876,0.00033281863,0.96625024,0.0001280696,0.00009014123,0.00033344576,0.0013951224,0.021855213,0.0027479623],"genre_scores_gemma":[0.019155858,0.00021705175,0.9718032,0.00022806869,0.000017060742,0.001080952,0.0024630148,0.0014182867,0.0036164897],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9982948,0.00039909207,0.000121101846,0.00053820293,0.00053811935,0.00010862967],"domain_scores_gemma":[0.998384,0.0008007594,0.00013612752,0.00026703512,0.00033568218,0.00007645462],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0020641575,0.0019504405,0.00091887097,0.0014231475,0.00066380174,0.00090706814,0.0017464567,0.0013172062,0.0175609],"category_scores_gemma":[0.00370012,0.0016545698,0.0011294596,0.001241724,0.00071001914,0.0010849213,0.0013406619,0.002184969,0.010054813],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00077818043,0.00014553563,0.0007658668,0.0011241484,0.000112803755,0.00041225558,0.00031990782,0.011077333,0.6409951,0.009238936,0.02660008,0.3084299],"study_design_scores_gemma":[0.00030818052,0.00071685144,0.0012777431,0.00016527735,0.00009743258,0.0013957737,0.00008895745,0.13681325,0.6835935,0.013194671,0.16214602,0.0002022957],"about_ca_topic_score_codex":0.00074805215,"about_ca_topic_score_gemma":0.0013267979,"teacher_disagreement_score":0.0175609,"about_ca_system_score_codex":0.00068663224,"about_ca_system_score_gemma":0.0008047881,"threshold_uncertainty_score":0.058747053},"labels":[],"label_agreement":null},{"id":"W2104398544","doi":"10.1093/bioinformatics/btp368","title":"PhyloBayes 3: a Bayesian software package for phylogenetic reconstruction and molecular dating","year":2009,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1351,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"Université de Montréal; Agence Nationale de la Recherche","keywords":"Phylogenetic tree; Bayesian probability; Variety (cybernetics); Molecular clock; Probabilistic logic; Computer science; Software; Software package; Computational biology; Phylogenetics; Parametric statistics; Evolutionary biology; Biology; Data mining; Artificial intelligence; Genetics; Mathematics; Statistics; Programming language; Gene","score_opus":0.00846319551486009,"score_gpt":0.2270831567231505,"score_spread":0.21861996120829041,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2104398544","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0016102123,0.0006378241,0.8839901,0.00046072924,0.000242236,0.00023322173,0.029055825,0.08027506,0.0034948513],"genre_scores_gemma":[0.015816957,0.00060854206,0.9105462,0.00038341305,0.00015944988,0.0019326014,0.028323432,0.039043605,0.0031857633],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9965062,0.0015315968,0.0004235253,0.0005728948,0.00077894994,0.00018686555],"domain_scores_gemma":[0.98886263,0.007554643,0.0008469202,0.001132242,0.0011104998,0.0004931038],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.011285691,0.0027705466,0.004648385,0.0069398372,0.0023415745,0.0036552642,0.008340739,0.0027852114,0.096529536],"category_scores_gemma":[0.031157948,0.004313133,0.0036916968,0.005651732,0.0012989319,0.0042825146,0.0052672615,0.007039936,0.04365171],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00138021,0.0002338983,0.0063862335,0.00648504,0.0022317085,0.0013341958,0.0010625456,0.045377705,0.011118927,0.06758165,0.52966976,0.32713813],"study_design_scores_gemma":[0.0010087121,0.00015437239,0.005640521,0.0012230008,0.00078130834,0.002097492,0.00020338384,0.26490688,0.0093327025,0.17254974,0.5413842,0.00071764813],"about_ca_topic_score_codex":0.0064042546,"about_ca_topic_score_gemma":0.006465327,"teacher_disagreement_score":0.096529536,"about_ca_system_score_codex":0.0013395094,"about_ca_system_score_gemma":0.0043811034,"threshold_uncertainty_score":0.32292348},"labels":[],"label_agreement":null},{"id":"W2104474107","doi":"10.1093/bioinformatics/17.2.198","title":"COMBOSA3D: combining sequence alignments with three-dimensional structures","year":2001,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Documentation; Sequence (biology); Computer science; Source code; Code (set theory); Sequence alignment; Multiple sequence alignment; Scheme (mathematics); Information retrieval; Data mining; Programming language; Peptide sequence; Biology; Mathematics; Genetics; Set (abstract data type)","score_opus":0.020790341318182227,"score_gpt":0.24309067341316407,"score_spread":0.22230033209498185,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2104474107","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.018385353,0.0017316659,0.61786664,0.0007459702,0.0011091036,0.0006253516,0.06751904,0.28044605,0.011570829],"genre_scores_gemma":[0.035567436,0.0013737088,0.79659504,0.00020130599,0.00019714033,0.0012240079,0.11470801,0.045125794,0.005007489],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99866974,0.00025989107,0.00016999317,0.000365525,0.00043195055,0.00010289272],"domain_scores_gemma":[0.9981693,0.00069485046,0.00021575818,0.00033643132,0.0003323561,0.00025128454],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0031259581,0.0032848762,0.003337356,0.006490323,0.002986233,0.0060915165,0.005254338,0.0015099073,0.057204362],"category_scores_gemma":[0.0077107,0.003155071,0.0033139423,0.005173522,0.00079830457,0.0034128893,0.0038235702,0.0047306544,0.038824722],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0028413162,0.0003254618,0.0026419829,0.004521532,0.0010010103,0.0012380906,0.0006971762,0.0059072506,0.08341157,0.011923612,0.6113918,0.2740992],"study_design_scores_gemma":[0.0015389075,0.00028295178,0.005255164,0.00056497793,0.00064443983,0.0017641484,0.00045199916,0.1491147,0.13286886,0.040971898,0.66586274,0.0006792326],"about_ca_topic_score_codex":0.0029574893,"about_ca_topic_score_gemma":0.004201717,"teacher_disagreement_score":0.057204362,"about_ca_system_score_codex":0.0010337677,"about_ca_system_score_gemma":0.001863377,"threshold_uncertainty_score":0.19136769},"labels":[],"label_agreement":null},{"id":"W2104502840","doi":"10.1093/bioinformatics/btu808","title":"Kablammo: an interactive, web-based BLAST results visualizer","year":2014,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":92,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"Natural Sciences and Engineering Research Council of Canada; Alberta Innovates; Alberta Innovates - Health Solutions","keywords":"Computer science; MIT License; Visualization; License; Source code; Web application; Code (set theory); Interactive visualization; Sequence (biology); Software; World Wide Web; Programming language; Data mining; Biology; Operating system; Set (abstract data type); Genetics","score_opus":0.010736696061096446,"score_gpt":0.2563567379274806,"score_spread":0.24562004186638414,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2104502840","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0064589363,0.0009316656,0.13475147,0.0011097242,0.00033236956,0.00028626784,0.05850513,0.77895314,0.018671274],"genre_scores_gemma":[0.06977032,0.0016431217,0.504865,0.0018464752,0.00031685355,0.001844489,0.19752559,0.19433552,0.0278526],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9989706,0.00019117328,0.00010472974,0.0002125168,0.00041718388,0.000103830236],"domain_scores_gemma":[0.99798125,0.0009227163,0.00020542905,0.00026350428,0.00038137368,0.00024564232],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0024292136,0.0023579474,0.0012554821,0.003471075,0.0012591375,0.0021350638,0.0031861267,0.0016624611,0.08344719],"category_scores_gemma":[0.006193495,0.0014310866,0.0011587323,0.0020692537,0.00048630615,0.003778803,0.0029522649,0.0023925412,0.048799597],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0016175353,0.00013526069,0.0021795055,0.0018682297,0.00014276,0.0005391805,0.00044771968,0.0012539544,0.032782137,0.0039854394,0.85378844,0.10125976],"study_design_scores_gemma":[0.000982827,0.00021176782,0.0085664,0.0009079129,0.00016820378,0.0022548493,0.00035027138,0.05600054,0.07217555,0.02419354,0.83365935,0.00052875327],"about_ca_topic_score_codex":0.0027033067,"about_ca_topic_score_gemma":0.003031881,"teacher_disagreement_score":0.08344719,"about_ca_system_score_codex":0.0006487778,"about_ca_system_score_gemma":0.00095180445,"threshold_uncertainty_score":0.27915865},"labels":[],"label_agreement":null},{"id":"W2104873783","doi":"10.1093/bioinformatics/btm475","title":"PFRES: protein fold classification by using evolutionary information and predicted secondary structure","year":2007,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Machine Learning in Bioinformatics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":122,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Protein tertiary structure; Protein secondary structure; Computer science; Pseudo amino acid composition; Similarity (geometry); Classifier (UML); Pattern recognition (psychology); Protein Data Bank (RCSB PDB); Fold (higher-order function); Protein superfamily; Protein structure prediction; Representation (politics); Feature vector; Structural Classification of Proteins database; Protein structure; Sequence (biology); Structural alignment; Data mining; Algorithm; Artificial intelligence; Sequence alignment; Biology; Peptide sequence; Amino acid; Genetics","score_opus":0.006359821080226958,"score_gpt":0.23383970742686902,"score_spread":0.22747988634664207,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2104873783","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.08022684,0.002194039,0.66668457,0.00055562734,0.00025719687,0.00057644123,0.03260345,0.21111487,0.005786968],"genre_scores_gemma":[0.2992299,0.0010200592,0.6056753,0.00016017686,0.00023497379,0.00064593216,0.083676934,0.0034465124,0.0059102615],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99893826,0.00018261377,0.00008272678,0.00022903741,0.0004907784,0.00007662265],"domain_scores_gemma":[0.99881065,0.0004344647,0.00024619393,0.00017309588,0.00026546646,0.00007002835],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0017622682,0.0022719782,0.00096416124,0.0044589844,0.00051130005,0.00074001285,0.0018370977,0.0008691765,0.009428702],"category_scores_gemma":[0.0045295516,0.00034605528,0.000867346,0.0018037913,0.0003854776,0.0017124407,0.00082336174,0.0006871868,0.008114785],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013163773,0.00026587787,0.017256005,0.0014525737,0.00024743055,0.0005319269,0.00013421093,0.023488525,0.03655168,0.0035620013,0.11661959,0.79857373],"study_design_scores_gemma":[0.00017415875,0.00058609,0.017422952,0.00025864117,0.000100875164,0.0025207708,0.00011631296,0.83692056,0.084848195,0.009892185,0.0469894,0.00016983112],"about_ca_topic_score_codex":0.0010192882,"about_ca_topic_score_gemma":0.0010646924,"teacher_disagreement_score":0.009428702,"about_ca_system_score_codex":0.00042823332,"about_ca_system_score_gemma":0.00062791357,"threshold_uncertainty_score":0.031542182},"labels":[],"label_agreement":null},{"id":"W2105091000","doi":"10.1093/bioinformatics/btn295","title":"Poisson adjacency distributions in genome comparison: multichromosomal, circular, signed and unsigned cases","year":2008,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genome Rearrangement Algorithms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":18,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Ottawa","funders":"Natural Sciences and Engineering Research Council of Canada; Canada Research Chairs","keywords":"Poisson distribution; Phylogenetic tree; Mathematics; Statistics; Limiting; Adjacency list; Similarity (geometry); Chromosome; Probability distribution; Genome; Adjacency matrix; Poisson regression; Biology; Combinatorics; Genetics; Computer science; Artificial intelligence; Population","score_opus":0.02415182833651396,"score_gpt":0.2510118703173962,"score_spread":0.22686004198088222,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2105091000","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.18115112,0.0006052313,0.8141348,0.0003740978,0.00003135564,0.00012982171,0.00025517316,0.00048507936,0.0028333608],"genre_scores_gemma":[0.8765454,0.0004623439,0.1197246,0.00012197511,0.000109556975,0.0003122548,0.0008242025,0.00017370682,0.0017259541],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99596184,0.0019198798,0.00021826994,0.0006114737,0.0010752723,0.00021326923],"domain_scores_gemma":[0.93411076,0.057736095,0.0031314609,0.0026950995,0.0016581366,0.0006683498],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.011438671,0.00035003174,0.00065021426,0.0038106348,0.00067138375,0.001960898,0.0024821255,0.0012605783,0.0028736577],"category_scores_gemma":[0.07450792,0.00040801475,0.0008975279,0.0025519258,0.0023708798,0.0029736385,0.0020034676,0.0012085364,0.000549564],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00037120326,0.00015183473,0.046258382,0.00024848455,0.000113367256,0.0018434415,0.0017491753,0.30054796,0.0036766513,0.5388729,0.0030816812,0.10308493],"study_design_scores_gemma":[0.000020518693,0.000037108093,0.005386295,0.000044332,0.000018882489,0.0015359441,0.00029049013,0.7141091,0.0017878299,0.27517098,0.0015416548,0.000056923633],"about_ca_topic_score_codex":0.0011109393,"about_ca_topic_score_gemma":0.00096183165,"teacher_disagreement_score":0.011438671,"about_ca_system_score_codex":0.0010993456,"about_ca_system_score_gemma":0.0005323627,"threshold_uncertainty_score":0.060494184},"labels":[],"label_agreement":null},{"id":"W2105187843","doi":"10.1093/bioinformatics/btr288","title":"The role of indirect connections in gene networks in predicting function","year":2011,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":81,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canada's Michael Smith Genome Sciences Centre; University of British Columbia","funders":"National Institute of General Medical Sciences; Canadian Institutes of Health Research; National Institutes of Health; Michael Smith Health Research BC","keywords":"Computer science; Function (biology); Data mining; Code (set theory); Gene regulatory network; Source code; MATLAB; Algorithm; Theoretical computer science; Machine learning; Artificial intelligence; Gene; Gene expression; Programming language; Biology; Genetics","score_opus":0.009714114048023536,"score_gpt":0.1941080902456587,"score_spread":0.18439397619763517,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2105187843","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.3720003,0.0038862163,0.6072909,0.0028563095,0.00007043293,0.00011519838,0.0016810347,0.0008234322,0.011276197],"genre_scores_gemma":[0.925688,0.0014007421,0.07030648,0.00012238485,0.00008442666,0.00010023209,0.00079672254,0.00013853269,0.0013625111],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99895346,0.00047356967,0.000058823072,0.0002830142,0.0001797653,0.000051295],"domain_scores_gemma":[0.97569907,0.02072919,0.0012766897,0.0010100516,0.0009445869,0.00034044197],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0034243031,0.00091962685,0.00042326617,0.0018154338,0.0008347036,0.0016021446,0.0009331133,0.00068474055,0.0032580704],"category_scores_gemma":[0.028479803,0.00040679853,0.0003358529,0.002010367,0.0020397748,0.0033956342,0.0015885019,0.0011377345,0.0006990267],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00071951037,0.0001618618,0.2347094,0.0014739174,0.00040022592,0.0006454377,0.000953504,0.3175242,0.015438316,0.12758482,0.005566301,0.29482254],"study_design_scores_gemma":[0.00004418988,0.00014893476,0.042052202,0.0002002942,0.00019613582,0.00071166595,0.00012952763,0.6313758,0.008128515,0.31301662,0.0039390353,0.0000571388],"about_ca_topic_score_codex":0.002282092,"about_ca_topic_score_gemma":0.0030086408,"teacher_disagreement_score":0.0034243031,"about_ca_system_score_codex":0.0007027967,"about_ca_system_score_gemma":0.00066710054,"threshold_uncertainty_score":0.01810968},"labels":[],"label_agreement":null},{"id":"W2106350001","doi":"10.1093/bioinformatics/bts350","title":"Identifying aberrant pathways through integrated analysis of knowledge in pharmacogenomics","year":2012,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Biomedical Text Mining and Ontologies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":47,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Carleton University","funders":"National Human Genome Research Institute; National Institutes of Health; European Commission","keywords":"Pharmacogenomics; Ontology; Disease; Drug repositioning; Identification (biology); Computational biology; Computer science; Drug discovery; Data science; Bioinformatics; Drug; Biology; Medicine; Pharmacology","score_opus":0.05877106970867121,"score_gpt":0.3227743530747683,"score_spread":0.26400328336609713,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2106350001","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.12798133,0.005866545,0.8214412,0.0042548226,0.000061234336,0.00045225638,0.024873061,0.005732811,0.009336787],"genre_scores_gemma":[0.40210226,0.0060033016,0.5680935,0.0007082383,0.00006955628,0.00027402476,0.021468807,0.00022994225,0.0010503822],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9987435,0.00029446356,0.00017347722,0.00028979382,0.0004360339,0.00006280944],"domain_scores_gemma":[0.99690896,0.0018759187,0.0004366948,0.0003941388,0.00029844965,0.000085820066],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0020523085,0.00072270655,0.00076368486,0.0071757813,0.0005293324,0.0019803,0.0009143566,0.0005802698,0.0021387374],"category_scores_gemma":[0.0045564556,0.00027283892,0.0012503017,0.008308554,0.0008243576,0.002783734,0.0022308943,0.0008363014,0.00052445685],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014347882,0.000915057,0.08609868,0.0055433707,0.002048925,0.0035278038,0.0011143537,0.08460207,0.07073094,0.078113765,0.012038885,0.65383124],"study_design_scores_gemma":[0.00018675909,0.0002343914,0.04918236,0.0010861051,0.0020010737,0.0027547965,0.0008099175,0.4046531,0.08066529,0.3846085,0.07363238,0.0001852398],"about_ca_topic_score_codex":0.005245926,"about_ca_topic_score_gemma":0.0061256657,"teacher_disagreement_score":0.0071757813,"about_ca_system_score_codex":0.0014261024,"about_ca_system_score_gemma":0.0019741415,"threshold_uncertainty_score":0.010853767},"labels":[],"label_agreement":null},{"id":"W2106415954","doi":"10.1093/bioinformatics/bth257","title":"PUNS: transcriptomic- and genomic-<i>in silico</i> PCR for enhanced primer design","year":2004,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Molecular Biology Techniques and Applications","field":"Biochemistry, Genetics and Molecular Biology","cited_by":35,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canada Research Chairs; University of Toronto","funders":"","keywords":"Primer (cosmetics); In silico; UniGene; Computational biology; In silico PCR; Polymerase chain reaction; DNA microarray; Biology; Genetics; Computer science; Genome; Multiplex polymerase chain reaction; Gene; Expressed sequence tag; Chemistry","score_opus":0.014534450628128975,"score_gpt":0.25213249264830595,"score_spread":0.23759804202017698,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2106415954","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.009400206,0.0006667641,0.9460846,0.00026235075,0.00043793124,0.0011359657,0.009286491,0.028906118,0.0038195262],"genre_scores_gemma":[0.026891949,0.0005500143,0.9283189,0.00084800395,0.00011474756,0.0038622096,0.024791788,0.008021909,0.0066005057],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9955368,0.0015501807,0.00043133032,0.0012690222,0.0007847367,0.0004280588],"domain_scores_gemma":[0.9968238,0.0013554725,0.0003428065,0.00076025253,0.00049724785,0.00022036169],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0045275977,0.0032678964,0.0027344353,0.002038042,0.0009005235,0.0018337177,0.0033148827,0.001286842,0.040780224],"category_scores_gemma":[0.0060450807,0.002854833,0.0027681026,0.0018076908,0.0012409218,0.0016164698,0.0019476125,0.004215676,0.02986717],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0027644273,0.00060751487,0.0016809511,0.0039546704,0.00023162119,0.00071535516,0.0003603047,0.0044135754,0.78698134,0.0063333698,0.05709818,0.13485867],"study_design_scores_gemma":[0.00070462696,0.0011005105,0.0039004409,0.00040944992,0.000242178,0.0019061378,0.00011836619,0.042076644,0.7516814,0.0069140163,0.19065991,0.0002863193],"about_ca_topic_score_codex":0.00037884517,"about_ca_topic_score_gemma":0.0013340819,"teacher_disagreement_score":0.040780224,"about_ca_system_score_codex":0.00079334306,"about_ca_system_score_gemma":0.00168915,"threshold_uncertainty_score":0.13642341},"labels":[],"label_agreement":null},{"id":"W2106433904","doi":"10.1093/bioinformatics/btt659","title":"Fingerprinting protein structures effectively and efficiently","year":2013,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Protein Structure and Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":12,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; False positive paradox; Rendering (computer graphics); Algorithm; Protein structure; Similarity (geometry); Protein structure prediction; Local structure; Data structure; Artificial intelligence; Pattern recognition (psychology); Image (mathematics); Biology","score_opus":0.003844789569114486,"score_gpt":0.20698742703711143,"score_spread":0.20314263746799693,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2106433904","genre_codex":"empirical","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6049467,0.0018036237,0.37612408,0.00020337154,0.000027571657,0.00017121944,0.0015808298,0.011766124,0.0033765351],"genre_scores_gemma":[0.7986198,0.00068692677,0.1948142,0.000104102444,0.000043795564,0.000077167235,0.0035754275,0.00030992687,0.0017686354],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9992255,0.000099790246,0.000052791886,0.00014324274,0.00041575424,0.00006299553],"domain_scores_gemma":[0.9980628,0.0006399942,0.00034262644,0.00046560413,0.00038768072,0.000101367055],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00088862673,0.0008215362,0.0011912521,0.0031274285,0.00042922507,0.00087291305,0.00079585763,0.00072592456,0.00230183],"category_scores_gemma":[0.0031976264,0.00030507045,0.0005647369,0.0027274562,0.00040595187,0.0016922373,0.0009577542,0.0005299771,0.0017216941],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010364758,0.00048204785,0.036076423,0.00056471373,0.00019868836,0.0004765395,0.00025820377,0.037654936,0.2414103,0.0017212197,0.007193669,0.6729267],"study_design_scores_gemma":[0.00017870948,0.0011094473,0.043076523,0.0000610471,0.0002326698,0.003320644,0.00031191664,0.5675151,0.3662397,0.0066033727,0.011237311,0.00011368591],"about_ca_topic_score_codex":0.0019126938,"about_ca_topic_score_gemma":0.0015989953,"teacher_disagreement_score":0.0031274285,"about_ca_system_score_codex":0.0004890393,"about_ca_system_score_gemma":0.0005269985,"threshold_uncertainty_score":0.0077004433},"labels":[],"label_agreement":null},{"id":"W2106532429","doi":"10.1093/bioinformatics/btm422","title":"Multiple spaced seeds for homology search","year":2007,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genome Rearrangement Algorithms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":67,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Homology (biology); Exponential function; Heuristic; Algorithm; Computer science; Time complexity; Mathematics; Sensitivity (control systems); Smith–Waterman algorithm; Artificial intelligence; Sequence alignment; Biology; Genetics; Amino acid; Gene; Peptide sequence","score_opus":0.018724202756338102,"score_gpt":0.27138495041828953,"score_spread":0.2526607476619514,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2106532429","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.018447483,0.0007575124,0.97257304,0.00036462318,0.000110805326,0.00013929258,0.00022234513,0.0028018833,0.0045829834],"genre_scores_gemma":[0.13393392,0.0003751345,0.8614751,0.00014398017,0.00009520719,0.0001705566,0.0007242115,0.00048622393,0.0025956412],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9986547,0.00036422646,0.000089320456,0.00032239174,0.00048261692,0.0000866786],"domain_scores_gemma":[0.9963554,0.0020199358,0.0002988882,0.0006935388,0.00045892567,0.00017331759],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011471163,0.0009892649,0.001075533,0.0018070928,0.0008477544,0.001213626,0.0017172955,0.0016302251,0.010830717],"category_scores_gemma":[0.008444592,0.00060577306,0.00077217794,0.002963453,0.0014798957,0.0034254605,0.0018934329,0.0015386803,0.004336741],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00090456445,0.0002659057,0.0024882595,0.000899598,0.000082670835,0.00066259265,0.0004559658,0.16332577,0.035921335,0.23150165,0.022404114,0.54108757],"study_design_scores_gemma":[0.00015511546,0.00031706897,0.00049080886,0.00010196828,0.000033768094,0.0007924922,0.00008508831,0.6114537,0.024166444,0.34137094,0.020979464,0.000053111293],"about_ca_topic_score_codex":0.0006864952,"about_ca_topic_score_gemma":0.000713627,"teacher_disagreement_score":0.010830717,"about_ca_system_score_codex":0.0010526006,"about_ca_system_score_gemma":0.0008082291,"threshold_uncertainty_score":0.03623241},"labels":[],"label_agreement":null},{"id":"W2106763642","doi":"10.1093/bioinformatics/17.4.385","title":"PhyloBLAST: facilitating phylogenetic analysis of BLAST results","year":2001,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":27,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Phylogenetic tree; Perl; Multiple sequence alignment; The Internet; Sequence (biology); Computer science; Sequence analysis; Sequence alignment; Biology; World Wide Web; Genetics; Peptide sequence","score_opus":0.014662420360273883,"score_gpt":0.24227252337551633,"score_spread":0.22761010301524245,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2106763642","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.010252814,0.00094342354,0.349153,0.0010254883,0.000695965,0.000717802,0.071325906,0.5513527,0.01453293],"genre_scores_gemma":[0.040782392,0.00093889126,0.7603335,0.00043033296,0.00026041007,0.0019220617,0.13094032,0.05872376,0.0056683496],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.997743,0.00085289346,0.00029752916,0.00034349243,0.0005686331,0.00019436516],"domain_scores_gemma":[0.9942853,0.0034508894,0.00040683988,0.0007015743,0.0007093739,0.00044601157],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0075665377,0.0026871625,0.0018912525,0.006959563,0.0018094478,0.0034132542,0.0035173448,0.0014623592,0.0696416],"category_scores_gemma":[0.018931981,0.0019652655,0.0009909463,0.0054988153,0.0007546433,0.0033598375,0.0038083745,0.0031803453,0.06064588],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0033222982,0.0002061668,0.003812986,0.003405651,0.00020500006,0.0016560701,0.0014679853,0.0020968113,0.036983967,0.011200856,0.7592227,0.17641953],"study_design_scores_gemma":[0.0017747784,0.0003115115,0.00939426,0.0017521288,0.00028108322,0.0028378398,0.0009466772,0.08032119,0.05750058,0.03483398,0.80960125,0.00044468624],"about_ca_topic_score_codex":0.0015421391,"about_ca_topic_score_gemma":0.0020272657,"teacher_disagreement_score":0.0696416,"about_ca_system_score_codex":0.0008669379,"about_ca_system_score_gemma":0.0015316774,"threshold_uncertainty_score":0.23297435},"labels":[],"label_agreement":null},{"id":"W2106808791","doi":"10.1093/bioinformatics/btn565","title":"Slider—maximum use of probability information for alignment of short sequence reads and SNP detection","year":2008,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":59,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University; BC Cancer Agency","funders":"Michael Smith Health Research BC","keywords":"Computer science; Sequence (biology); Base (topology); Algorithm; SNP; Slider; Data mining; Mathematics; Genetics; Engineering; Single-nucleotide polymorphism","score_opus":0.06261196664988462,"score_gpt":0.25981289960732773,"score_spread":0.19720093295744312,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2106808791","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.025289606,0.00083750737,0.9512757,0.0002761191,0.000096194555,0.0001785856,0.0011416989,0.018994434,0.0019101923],"genre_scores_gemma":[0.109739706,0.00051503343,0.8833742,0.00022955016,0.00023220066,0.0003171232,0.0028451083,0.0010657116,0.0016814594],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99749905,0.00055628037,0.00013714288,0.000782567,0.00093906425,0.000085932224],"domain_scores_gemma":[0.99605775,0.0022217208,0.00065203645,0.0005928933,0.0003055497,0.00017013832],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0025800266,0.001355656,0.0012693916,0.0016085162,0.0005849986,0.0012130685,0.0017989435,0.00094165595,0.0061978223],"category_scores_gemma":[0.0083816815,0.000812923,0.00085643015,0.0015644357,0.00078691565,0.0028094517,0.0015052897,0.0012759707,0.0045104777],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009114734,0.0001758247,0.008574313,0.0011510951,0.0002779913,0.0005401937,0.00021609427,0.014549538,0.20084792,0.009732501,0.017501932,0.7455211],"study_design_scores_gemma":[0.00030209922,0.0014220617,0.016067825,0.00020827744,0.00020768539,0.004105265,0.000168739,0.53891826,0.36634797,0.029137503,0.042814452,0.00029988977],"about_ca_topic_score_codex":0.0004788239,"about_ca_topic_score_gemma":0.0007846342,"teacher_disagreement_score":0.0061978223,"about_ca_system_score_codex":0.00034725506,"about_ca_system_score_gemma":0.00067952974,"threshold_uncertainty_score":0.020733714},"labels":[],"label_agreement":null},{"id":"W2106848411","doi":"10.1093/bioinformatics/btt538","title":"Single assay-wide variance experimental (SAVE) design for high-throughput screening","year":2013,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Cell Image Analysis Techniques","field":"Biochemistry, Genetics and Molecular Biology","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University and Génome Québec Innovation Centre","funders":"","keywords":"Replication (statistics); Computer science; Bayes' theorem; Variance (accounting); Throughput; Statistical hypothesis testing; False discovery rate; Design of experiments; Data mining; Statistics; Algorithm; Bayesian probability; Artificial intelligence; Mathematics; Biology","score_opus":0.019963986038961396,"score_gpt":0.26029553693097496,"score_spread":0.24033155089201355,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2106848411","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.011080267,0.00009212089,0.98385,0.000087884604,0.00009638033,0.00093909644,0.0003295385,0.0030291518,0.00049554184],"genre_scores_gemma":[0.09565866,0.00011591137,0.89422584,0.0003075169,0.00007270142,0.007286589,0.00063317793,0.00055358536,0.0011459921],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9820024,0.0117339,0.00079483073,0.0020351117,0.003064309,0.0003694484],"domain_scores_gemma":[0.9773595,0.011634827,0.0026359134,0.0054924265,0.0024020153,0.00047539474],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.016677953,0.0015520339,0.0023870266,0.0013136293,0.0006910622,0.0013213363,0.002594348,0.0015768516,0.0053994376],"category_scores_gemma":[0.027121035,0.0014763739,0.0017622127,0.0011956892,0.0018624504,0.0009974958,0.0017995627,0.0030271541,0.0017311012],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.007534871,0.0022795147,0.010645526,0.0025879995,0.0015411675,0.00083116215,0.0005903593,0.07842256,0.5490413,0.04693813,0.010485727,0.28910178],"study_design_scores_gemma":[0.0024811057,0.011528111,0.010570574,0.00013696555,0.00076855125,0.0016182683,0.00011018474,0.3671866,0.5272388,0.04393902,0.03390059,0.00052131066],"about_ca_topic_score_codex":0.0003404373,"about_ca_topic_score_gemma":0.0008105773,"teacher_disagreement_score":0.016677953,"about_ca_system_score_codex":0.000838626,"about_ca_system_score_gemma":0.0017450276,"threshold_uncertainty_score":0.088202536},"labels":[],"label_agreement":null},{"id":"W2106855819","doi":"10.1093/bioinformatics/17.6.487","title":"Uniform integration of genome mapping data using intersection graphs","year":2001,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":40,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Intersection (aeronautics); Computer science; Contig; Software; Clique; Linkage (software); Data mining; Graph; Set (abstract data type); Data set; Algorithm; Genome; Theoretical computer science; Artificial intelligence; Biology; Mathematics; Genetics; Combinatorics","score_opus":0.03211698764524208,"score_gpt":0.25416274389606996,"score_spread":0.2220457562508279,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2106855819","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.023351083,0.00019699345,0.9676614,0.00016689036,0.000020433474,0.00018456878,0.0015165029,0.004519187,0.0023830915],"genre_scores_gemma":[0.18214968,0.00033062775,0.8061848,0.0001073982,0.000047685666,0.00038548865,0.008919308,0.0009767099,0.0008982758],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99187523,0.0020294108,0.0006012765,0.0023693612,0.0027727056,0.00035199613],"domain_scores_gemma":[0.98056364,0.00969184,0.0017986437,0.004252928,0.003375559,0.00031732162],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0045346953,0.0010823405,0.0012305124,0.0081373,0.0012569807,0.00384325,0.002510101,0.00088301685,0.002798765],"category_scores_gemma":[0.025351591,0.00074892846,0.0012700048,0.011643689,0.0011371537,0.005308251,0.0033468618,0.0015620594,0.0010992347],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006634481,0.00040321524,0.043742087,0.00094519893,0.00048050788,0.00081211387,0.0014874777,0.1398914,0.029439107,0.08838104,0.0103806285,0.68337375],"study_design_scores_gemma":[0.000074953714,0.00025461536,0.018608313,0.00017834853,0.00022237933,0.0010797421,0.00097406993,0.6896716,0.040261574,0.21217339,0.03634077,0.00016024274],"about_ca_topic_score_codex":0.004733931,"about_ca_topic_score_gemma":0.0041932804,"teacher_disagreement_score":0.0081373,"about_ca_system_score_codex":0.001618243,"about_ca_system_score_gemma":0.0017824145,"threshold_uncertainty_score":0.023982048},"labels":[],"label_agreement":null},{"id":"W2107141268","doi":"10.1093/bioinformatics/btn381","title":"Multi-dimensional classification of biomedical text: Toward automated, practical provision of high-utility text to diverse users","year":2008,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Biomedical Text Mining and Ontologies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":134,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Queen's University","funders":"Natural Sciences and Engineering Research Council of Canada; National Science Foundation","keywords":"Computer science; Categorization; Annotation; Sentence; Information retrieval; Task (project management); Variety (cybernetics); Artificial intelligence; Information extraction; Natural language processing; Focus (optics)","score_opus":0.05631967128910629,"score_gpt":0.3222240203340662,"score_spread":0.2659043490449599,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2107141268","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.07253602,0.0010016214,0.90424407,0.0036080026,0.00013496754,0.001111048,0.0030093428,0.010885166,0.003469885],"genre_scores_gemma":[0.14472787,0.00037584556,0.8456869,0.0005182525,0.0002553826,0.00066216127,0.0057548834,0.0004666181,0.0015521537],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99162674,0.0034674087,0.0009959948,0.0018059567,0.0018306477,0.0002732264],"domain_scores_gemma":[0.9609073,0.021452108,0.004210506,0.0042091208,0.008067651,0.0011533013],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008546948,0.0014343419,0.0014012114,0.008093144,0.0017650661,0.004641315,0.0027960713,0.002007478,0.002847018],"category_scores_gemma":[0.032611687,0.0005975825,0.0010046676,0.004910848,0.0017556103,0.007196731,0.0043941927,0.0024753155,0.003803959],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00072279293,0.00060465804,0.012592132,0.0012587814,0.00007070352,0.00041029786,0.0043614176,0.0060200817,0.05628081,0.008541122,0.020573912,0.8885633],"study_design_scores_gemma":[0.00018279371,0.0007082168,0.020288693,0.00061044656,0.00020832759,0.0015258769,0.0074936696,0.74549764,0.08810277,0.070644416,0.06445085,0.0002863524],"about_ca_topic_score_codex":0.0020124393,"about_ca_topic_score_gemma":0.0022142753,"teacher_disagreement_score":0.008546948,"about_ca_system_score_codex":0.0014648185,"about_ca_system_score_gemma":0.0019612666,"threshold_uncertainty_score":0.045201123},"labels":[],"label_agreement":null},{"id":"W2107249189","doi":"10.1093/bioinformatics/btq430","title":"Cytoscape Web: an interactive web-based network browser","year":2010,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":746,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"National Institutes of Health; National Institute of General Medical Sciences; Ontario Genomics Institute; Ontario Genomics; Genome Canada; University of Michigan; University of Toronto; Memorial Sloan-Kettering Cancer Center","keywords":"Computer science; JavaScript; World Wide Web; Web application; Web page; ActionScript; Visualization; Information retrieval; Data mining","score_opus":0.00650414083413331,"score_gpt":0.22936811716642294,"score_spread":0.22286397633228963,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2107249189","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0033716047,0.0008767695,0.3330193,0.0012245802,0.00084322836,0.00069280865,0.12524529,0.51641893,0.018307507],"genre_scores_gemma":[0.03630448,0.0014870863,0.57288927,0.0014566554,0.00042088225,0.00408572,0.266106,0.08733428,0.029915558],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99886584,0.00024047388,0.000117120064,0.00020582399,0.00048260708,0.000088161345],"domain_scores_gemma":[0.99697125,0.0013701001,0.0001916388,0.0003756492,0.0005500006,0.0005413606],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0023993398,0.0014668495,0.0012440995,0.0040315012,0.0008207138,0.001918862,0.0037008794,0.0017259077,0.10321303],"category_scores_gemma":[0.007214428,0.0015381149,0.0009969021,0.0024196,0.0004119936,0.0025098347,0.0028648106,0.0029628966,0.047616318],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005156388,0.00015669192,0.0009970167,0.0015314006,0.00025040205,0.00058665406,0.00029696742,0.0025817174,0.026726656,0.0074629304,0.85828835,0.10060563],"study_design_scores_gemma":[0.0006612504,0.00007563692,0.002571031,0.00040100704,0.0001357887,0.0010399616,0.000088798995,0.047788598,0.0393214,0.033129,0.8744833,0.00030415846],"about_ca_topic_score_codex":0.0026344003,"about_ca_topic_score_gemma":0.0049549635,"teacher_disagreement_score":0.10321303,"about_ca_system_score_codex":0.00055990386,"about_ca_system_score_gemma":0.0021249873,"threshold_uncertainty_score":0.34528196},"labels":[],"label_agreement":null},{"id":"W2107279733","doi":"10.1093/bioinformatics/btl415","title":"Nemo: an evolutionary and population genetics programming framework","year":2006,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Evolution and Genetic Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":196,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung","keywords":"Computer science; Executable; Population; Coalescent theory; Variety (cybernetics); Genetic programming; Programming language; Flexibility (engineering); Context (archaeology); Theoretical computer science; Biology; Machine learning; Artificial intelligence; Genetics; Phylogenetics","score_opus":0.005687145129834929,"score_gpt":0.24141533977967328,"score_spread":0.23572819464983835,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2107279733","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00131266,0.00012419053,0.9829417,0.00024317967,0.00004945784,0.00007124578,0.0010741728,0.010998842,0.0031846517],"genre_scores_gemma":[0.0284344,0.00034689403,0.9575382,0.0003685006,0.000066619534,0.0009224123,0.0030992664,0.004019387,0.0052044503],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99934655,0.00019104981,0.00006888454,0.00013587893,0.00019406193,0.00006362794],"domain_scores_gemma":[0.9991602,0.0005094035,0.00005941942,0.00007332829,0.00011965007,0.000078021905],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0022550495,0.0010530454,0.00081522466,0.0007170164,0.0008682379,0.0017168167,0.0030998548,0.001032501,0.013708142],"category_scores_gemma":[0.003884842,0.00061974383,0.0015983772,0.0008429002,0.0009254986,0.0021001976,0.0027079415,0.0017882566,0.0042858245],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006548472,0.00018231057,0.003502108,0.0014229647,0.00019032495,0.00087166904,0.00087932375,0.14546013,0.006869478,0.4283221,0.09582508,0.31581956],"study_design_scores_gemma":[0.00020002159,0.000040519633,0.00052426226,0.00023860596,0.000049676903,0.00045261995,0.00006096708,0.51104474,0.0044664782,0.21457553,0.2682816,0.000065032196],"about_ca_topic_score_codex":0.0013989453,"about_ca_topic_score_gemma":0.0015543093,"teacher_disagreement_score":0.013708142,"about_ca_system_score_codex":0.00063515705,"about_ca_system_score_gemma":0.0012538221,"threshold_uncertainty_score":0.045858324},"labels":[],"label_agreement":null},{"id":"W2107320771","doi":"10.1093/bioinformatics/bts457","title":"ANGES: reconstructing ANcestral GEnomeS maps","year":2012,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genome Rearrangement Algorithms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":70,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"Agence Nationale de la Recherche","keywords":"Genome; Extant taxon; Python (programming language); Merge (version control); Biology; Suite; Evolutionary biology; Computer science; Computational biology; Genetics; Geography; Archaeology; Programming language; Information retrieval","score_opus":0.02167448892732243,"score_gpt":0.24648847665906595,"score_spread":0.22481398773174352,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2107320771","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.036817454,0.00045410122,0.7344237,0.000424416,0.00026068508,0.00016982445,0.05305527,0.15853326,0.015861303],"genre_scores_gemma":[0.1397088,0.0005425074,0.6990315,0.0002213048,0.00009931942,0.00035865282,0.1238966,0.027990857,0.008150478],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9995209,0.000095972646,0.000027919594,0.00014049682,0.00015458692,0.000060274466],"domain_scores_gemma":[0.9990747,0.00038161717,0.00006119895,0.00022751096,0.0001755125,0.000079416735],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013341988,0.0009833064,0.00080846174,0.0031818196,0.001136116,0.0019680846,0.0021198168,0.0007946169,0.03648748],"category_scores_gemma":[0.005272473,0.0010877257,0.0019596051,0.0020314443,0.00065375236,0.0021597163,0.0029819359,0.0017794297,0.009992307],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013655855,0.00022022561,0.016800497,0.0024293936,0.00062097434,0.0013420455,0.0013347483,0.09654312,0.033356246,0.09764922,0.3432136,0.40512437],"study_design_scores_gemma":[0.00041817903,0.00014756067,0.008453042,0.00036285486,0.00018737392,0.001485589,0.0003904087,0.5122353,0.037387684,0.108795285,0.3299507,0.00018601898],"about_ca_topic_score_codex":0.0030654082,"about_ca_topic_score_gemma":0.004011122,"teacher_disagreement_score":0.03648748,"about_ca_system_score_codex":0.00059255096,"about_ca_system_score_gemma":0.001195851,"threshold_uncertainty_score":0.1220628},"labels":[],"label_agreement":null},{"id":"W2107414458","doi":"10.1093/bioinformatics/btv136","title":"EpIC: a rational pipeline for epitope immunogenicity characterization","year":2015,"lang":"en","type":"article","venue":"Bioinformatics","topic":"vaccines and immunoinformatics approaches","field":"Biochemistry, Genetics and Molecular Biology","cited_by":14,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Saskatchewan","funders":"","keywords":"Immunogenicity; Epitope; Computational biology; Peptide vaccine; Computer science; Peptide; Epitope mapping; Pipeline (software); Biology; Virology; Antibody; Immunology; Biochemistry; Programming language","score_opus":0.029183676362852008,"score_gpt":0.2518450514789635,"score_spread":0.2226613751161115,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2107414458","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.02312356,0.0007324517,0.9384847,0.0004020668,0.00007850355,0.0002825364,0.004467262,0.02782323,0.00460574],"genre_scores_gemma":[0.11333202,0.0014074366,0.8598768,0.0003265103,0.00004418893,0.0006201275,0.017388722,0.003731436,0.003272786],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9994773,0.000100838006,0.000031860138,0.0001134979,0.00021842124,0.000058133865],"domain_scores_gemma":[0.9993248,0.00025325423,0.00006910484,0.00009974318,0.00020846969,0.00004466563],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016673307,0.0018756284,0.0011566685,0.000911891,0.0005434777,0.0013191603,0.0014060523,0.0005193754,0.008163055],"category_scores_gemma":[0.0023031489,0.0006664834,0.0012732787,0.00088428485,0.00034735873,0.00095507107,0.00096710294,0.001812891,0.0056728604],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0015637864,0.00056689465,0.0073597743,0.0017393599,0.00045642888,0.0008840303,0.00033544487,0.1146024,0.23363462,0.036052976,0.07039353,0.53241074],"study_design_scores_gemma":[0.00027055352,0.0004187243,0.0025424976,0.00015861563,0.0002305068,0.0005652964,0.000088980225,0.6688922,0.18750493,0.030968688,0.10823356,0.00012552532],"about_ca_topic_score_codex":0.0011925027,"about_ca_topic_score_gemma":0.0015664852,"teacher_disagreement_score":0.008163055,"about_ca_system_score_codex":0.000588124,"about_ca_system_score_gemma":0.0016985226,"threshold_uncertainty_score":0.027308166},"labels":[],"label_agreement":null},{"id":"W2107861779","doi":"10.1093/bioinformatics/btp253","title":"Fluctuation AnaLysis CalculatOR: a web tool for the determination of mutation rate using Luria–Delbrück fluctuation analysis","year":2009,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Evolution and Genetic Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":405,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Brock University","funders":"U.S. Public Health Service; National Institutes of Health","keywords":"Calculator; Java applet; Computer science; Java; Estimator; Mutation testing; Mutation; Programming language; Operating system; Statistics; Biology; Genetics; Mathematics; Gene","score_opus":0.011083137509752405,"score_gpt":0.2770608778069041,"score_spread":0.26597774029715165,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2107861779","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.007161888,0.00036698912,0.614564,0.000205547,0.00016738924,0.00035526667,0.015824845,0.35801342,0.0033407395],"genre_scores_gemma":[0.05560883,0.00060396903,0.82916725,0.00042393655,0.00013774328,0.0037915166,0.033908002,0.064366244,0.011992467],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9987601,0.0002194994,0.00014333437,0.00027530474,0.0005195957,0.00008217471],"domain_scores_gemma":[0.99722904,0.0013402854,0.0002984709,0.00033895834,0.00065140513,0.00014179536],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0021164909,0.0019099977,0.0020918746,0.003910659,0.00077434763,0.00138152,0.002831221,0.00085775607,0.041089546],"category_scores_gemma":[0.0063095493,0.0009631705,0.0010146087,0.0022332985,0.00035580172,0.0013044768,0.0014219079,0.0018961924,0.023131996],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011313694,0.00045977076,0.008780741,0.0020328355,0.00041683656,0.00073552487,0.00044652467,0.0061548925,0.083430074,0.008711472,0.4561118,0.43158823],"study_design_scores_gemma":[0.0004716736,0.00048396268,0.02095181,0.0004018901,0.00028679913,0.001974032,0.00013236937,0.34457132,0.29382154,0.015614815,0.32055727,0.0007325007],"about_ca_topic_score_codex":0.0011469147,"about_ca_topic_score_gemma":0.0010745407,"teacher_disagreement_score":0.041089546,"about_ca_system_score_codex":0.0005609775,"about_ca_system_score_gemma":0.00091372046,"threshold_uncertainty_score":0.1374582},"labels":[],"label_agreement":null},{"id":"W2107867384","doi":"10.1093/bioinformatics/btg106","title":"Inference of transcriptional regulation relationshipsfrom gene expression data","year":2003,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":77,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"False positive paradox; Inference; Event (particle physics); Gene; Computer science; Computational biology; Data mining; Microarray analysis techniques; Microarray; Biology; Gene expression; Genetics; Artificial intelligence","score_opus":0.039213382559552604,"score_gpt":0.261241824872769,"score_spread":0.2220284423132164,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2107867384","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.23807475,0.0007134202,0.75440896,0.0003792794,0.000023423465,0.00014169414,0.0030671444,0.0020479958,0.0011433364],"genre_scores_gemma":[0.69972754,0.00063767406,0.28964406,0.00012755317,0.00008119829,0.00017749787,0.008725223,0.0001013347,0.00077795],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9989361,0.00026630238,0.00006668126,0.00038409492,0.00028613838,0.000060657443],"domain_scores_gemma":[0.9935824,0.005238205,0.000496675,0.00029186803,0.00030042324,0.00009050431],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0018884805,0.00095176144,0.0007304066,0.0018265862,0.00045530614,0.00078155636,0.001347364,0.0009783254,0.0019526408],"category_scores_gemma":[0.00826716,0.00043901373,0.0005216741,0.00170588,0.0008386041,0.0013667666,0.00064968277,0.0010309701,0.0005094092],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011659502,0.0008885813,0.090465024,0.0015762629,0.00023525431,0.00094734866,0.00024677048,0.40998796,0.09449289,0.016363889,0.0044442057,0.37918583],"study_design_scores_gemma":[0.000075685086,0.000113498616,0.015587633,0.0000537721,0.000058976555,0.0005637773,0.00007031568,0.90006864,0.056764893,0.023570092,0.0030335726,0.000039143993],"about_ca_topic_score_codex":0.0010292468,"about_ca_topic_score_gemma":0.0010202656,"teacher_disagreement_score":0.0019526408,"about_ca_system_score_codex":0.00044750414,"about_ca_system_score_gemma":0.0005301643,"threshold_uncertainty_score":0.009987354},"labels":[],"label_agreement":null},{"id":"W2108016183","doi":"10.1093/bioinformatics/bti189","title":"Finding differentially expressed genes for pattern generation","year":2004,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"","keywords":"Gene; Computational biology; Biology; Genetics; Computer science","score_opus":0.03261465458513153,"score_gpt":0.27135140593004875,"score_spread":0.23873675134491723,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2108016183","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.019757092,0.00051994034,0.9728966,0.0012031352,0.00008046495,0.00042166954,0.0017860828,0.0021041355,0.0012309258],"genre_scores_gemma":[0.15171975,0.00040675965,0.8389936,0.00052244816,0.00017896034,0.0008143635,0.0057036458,0.00035038183,0.0013101287],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9969031,0.00088627794,0.00024511848,0.0011605575,0.00067064574,0.00013418365],"domain_scores_gemma":[0.98796403,0.008161906,0.0008062538,0.0014157767,0.0013575056,0.00029453353],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004890803,0.0015558018,0.0014423385,0.002727826,0.00075249834,0.0018358382,0.0021863116,0.0016222247,0.005090621],"category_scores_gemma":[0.016027192,0.00054378936,0.0012868262,0.003714726,0.0014612233,0.0018129374,0.001481094,0.0016506762,0.002950741],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00070627476,0.0003818211,0.01990208,0.0021724608,0.00015325872,0.0009378615,0.0003106914,0.03891117,0.07024123,0.023757016,0.020160025,0.8223661],"study_design_scores_gemma":[0.0004931878,0.0005850871,0.019755496,0.00032296902,0.00019077865,0.0035456526,0.00033942363,0.5304348,0.10416685,0.30769283,0.032315627,0.00015725345],"about_ca_topic_score_codex":0.00047551712,"about_ca_topic_score_gemma":0.00045570408,"teacher_disagreement_score":0.005090621,"about_ca_system_score_codex":0.00084075675,"about_ca_system_score_gemma":0.001555202,"threshold_uncertainty_score":0.025865316},"labels":[],"label_agreement":null},{"id":"W2108102223","doi":"10.1093/bioinformatics/bts695","title":"ChromoZoom: a flexible, fluid, web-based genome browser","year":2012,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Lunenfeld-Tanenbaum Research Institute; University of Toronto","funders":"National Human Genome Research Institute; National Heart, Lung, and Blood Institute; University of California, Santa Cruz","keywords":"Computer science; Scrolling; Zoom; Rendering (computer graphics); Source code; Web server; Genome browser; User interface; Server; Web application; Computer graphics (images); Operating system; World Wide Web; The Internet; Genome; Genomics","score_opus":0.015099627374970379,"score_gpt":0.232037945268106,"score_spread":0.2169383178931356,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2108102223","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.011744277,0.0013330504,0.303139,0.0015688265,0.0010740499,0.0005491769,0.07842708,0.568608,0.03355661],"genre_scores_gemma":[0.09797514,0.0023319048,0.45119196,0.0015853724,0.00055449567,0.0012136331,0.2656648,0.12980342,0.049679283],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9993631,0.00007182657,0.000036289504,0.00014873972,0.00028781718,0.000092181945],"domain_scores_gemma":[0.998028,0.00059227675,0.00014305033,0.00036100336,0.00035302129,0.0005226943],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016058527,0.0011519251,0.00075751706,0.0014296563,0.0008179741,0.0027010683,0.0036432785,0.0012361701,0.07066032],"category_scores_gemma":[0.005085583,0.0010272147,0.00071367086,0.0019394886,0.0006375062,0.002339736,0.0024328295,0.002335016,0.03490107],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0019163011,0.00011819066,0.0035606609,0.0010882438,0.00009070934,0.0003948159,0.00039358583,0.0011105496,0.038078994,0.010888454,0.7234919,0.21886772],"study_design_scores_gemma":[0.00042650907,0.00012169803,0.008998543,0.00033231926,0.000085161155,0.0008535511,0.00016532505,0.020424223,0.030417142,0.01043672,0.92752415,0.00021464465],"about_ca_topic_score_codex":0.0069552436,"about_ca_topic_score_gemma":0.014844642,"teacher_disagreement_score":0.07066032,"about_ca_system_score_codex":0.0009060804,"about_ca_system_score_gemma":0.0017699323,"threshold_uncertainty_score":0.2363823},"labels":[],"label_agreement":null},{"id":"W2108125432","doi":"10.1093/bioinformatics/btq657","title":"A regression framework incorporating quantitative and negative interaction data improves quantitative prediction of PDZ domain–peptide interaction from primary sequence","year":2010,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Hippo pathway signaling and YAP/TAZ","field":"Biochemistry, Genetics and Molecular Biology","cited_by":32,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto; Western University","funders":"Canadian Institutes of Health Research; China Scholarship Council; National Natural Science Foundation of China","keywords":"PDZ domain; Context (archaeology); Sequence (biology); Computer science; Domain (mathematical analysis); Regression; Similarity (geometry); Construct (python library); Protein–protein interaction; Peptide; Computational biology; Data mining; Artificial intelligence; Biology; Mathematics; Statistics; Genetics","score_opus":0.0473592794905898,"score_gpt":0.3209091213810167,"score_spread":0.27354984189042686,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2108125432","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.14117366,0.00042874078,0.8527323,0.00035886286,0.000034681023,0.000070611364,0.00044085414,0.0039016602,0.0008587046],"genre_scores_gemma":[0.7382838,0.00017516228,0.2570421,0.00026323536,0.0000801193,0.00014588631,0.0018236279,0.00019724396,0.00198884],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.999203,0.00028244336,0.000042527357,0.00025047737,0.0001429993,0.00007845795],"domain_scores_gemma":[0.9976788,0.0013878979,0.0001936398,0.00015571166,0.00047206643,0.00011196062],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00432967,0.0012780129,0.0010951593,0.00071885897,0.0003613638,0.0006133365,0.0012106559,0.00076661963,0.0012203528],"category_scores_gemma":[0.0047016265,0.00036790612,0.0006872126,0.0004931096,0.0005345051,0.0008625881,0.00076545344,0.0016900403,0.0006352972],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006480221,0.0005036,0.007898769,0.00011946336,0.00020000758,0.000118567936,0.000049667593,0.7513358,0.030412225,0.0020311142,0.0026799052,0.20400287],"study_design_scores_gemma":[0.000011624606,0.000045982917,0.00035151228,0.0000020749073,0.000008948742,0.000012909264,0.0000020896082,0.9975076,0.0014934556,0.0004398972,0.000117716256,0.0000061860696],"about_ca_topic_score_codex":0.0052848174,"about_ca_topic_score_gemma":0.004920444,"teacher_disagreement_score":0.0052848174,"about_ca_system_score_codex":0.00053049007,"about_ca_system_score_gemma":0.0011152405,"threshold_uncertainty_score":0.02289772},"labels":[],"label_agreement":null},{"id":"W2108174382","doi":"10.1093/bioinformatics/btp456","title":"Modified variational Bayes EM estimation of hidden Markov tree model of cell lineages","year":2009,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Single-cell and spatial transcriptomics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":16,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Lunenfeld-Tanenbaum Research Institute; Mount Sinai Hospital","funders":"Engineering and Physical Sciences Research Council","keywords":"Lineage (genetic); Tree (set theory); Computer science; Bayes' theorem; Markov chain; Coalescent theory; Population; Markov model; Hidden Markov model; Bayesian probability; Algorithm; Expectation–maximization algorithm; Artificial intelligence; Biology; Mathematics; Machine learning; Statistics; Combinatorics; Phylogenetic tree; Maximum likelihood; Genetics","score_opus":0.014285339395681794,"score_gpt":0.22304628592734063,"score_spread":0.20876094653165883,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2108174382","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0102469195,0.00036260413,0.98689497,0.0003854253,0.000039718092,0.00007192662,0.00037912428,0.0003787808,0.0012405325],"genre_scores_gemma":[0.32275552,0.00081136194,0.6594814,0.0005175254,0.00019553784,0.00086534664,0.0043399036,0.0005893373,0.010444078],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99897134,0.00049923506,0.000053056174,0.00024684062,0.00012735381,0.00010215186],"domain_scores_gemma":[0.99317485,0.0057794983,0.00022528612,0.00016101717,0.0005190152,0.0001402777],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0033227946,0.0011074756,0.0019520415,0.0012040082,0.0006725092,0.0019672145,0.0031832824,0.0026919423,0.0063743345],"category_scores_gemma":[0.0140516255,0.0016426961,0.0017314192,0.0013410406,0.0011288996,0.0017065891,0.0014709654,0.002770623,0.0018322408],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00012569517,0.000030632633,0.0015219178,0.0001454355,0.00008463904,0.000108549306,0.000112908194,0.922609,0.0007933068,0.028278014,0.0028454908,0.043344334],"study_design_scores_gemma":[0.000011641369,0.0000053758836,0.000118372125,0.000013796677,0.000005784954,0.00001562237,0.000006190236,0.98827165,0.00012114029,0.011064696,0.00035879167,0.0000069387547],"about_ca_topic_score_codex":0.019036394,"about_ca_topic_score_gemma":0.019241812,"teacher_disagreement_score":0.019036394,"about_ca_system_score_codex":0.001935099,"about_ca_system_score_gemma":0.002596275,"threshold_uncertainty_score":0.037851155},"labels":[],"label_agreement":null},{"id":"W2108563376","doi":"10.1093/bioinformatics/btq639","title":"TE Displayer for post-genomic analysis of transposable elements","year":2010,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Chromosomal and Genetic Variations","field":"Agricultural and Biological Sciences","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Transposable element; In silico; Computer science; Computational biology; Biology; Data mining; Artificial intelligence; Genome; Genetics; Gene","score_opus":0.012886655885956541,"score_gpt":0.22270072485348852,"score_spread":0.20981406896753196,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2108563376","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.015017414,0.00031074375,0.5610974,0.00023442527,0.00021681991,0.00021804271,0.03228183,0.38687176,0.0037515038],"genre_scores_gemma":[0.090826906,0.00046860726,0.75693107,0.00036080912,0.00012767321,0.0013569549,0.07951963,0.05219614,0.018212246],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99944586,0.000090385,0.000059192327,0.00016821576,0.00017027104,0.0000660873],"domain_scores_gemma":[0.99819106,0.0009734516,0.00014862082,0.00031522798,0.0002599769,0.000111644615],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001811484,0.0015164121,0.0010774795,0.0015864862,0.00054174353,0.0011768523,0.0015243683,0.0010223974,0.06111825],"category_scores_gemma":[0.00418685,0.00079853815,0.001138488,0.0011797734,0.00024894468,0.0014156617,0.0011783526,0.0014773161,0.02011141],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0054417267,0.00043605483,0.007150303,0.002221961,0.0005931197,0.0014879336,0.0006497353,0.006703326,0.29197288,0.008347097,0.28960505,0.3853908],"study_design_scores_gemma":[0.00093911076,0.0005467954,0.013583061,0.00023632972,0.00026171494,0.0020970276,0.00015541169,0.15057269,0.56160337,0.01161605,0.25804123,0.0003472136],"about_ca_topic_score_codex":0.00070914376,"about_ca_topic_score_gemma":0.0009766198,"teacher_disagreement_score":0.06111825,"about_ca_system_score_codex":0.00038524778,"about_ca_system_score_gemma":0.0006302146,"threshold_uncertainty_score":0.20446092},"labels":[],"label_agreement":null},{"id":"W2108614159","doi":"10.1093/bioinformatics/btt001","title":"MaSC: mappability-sensitive cross-correlation for estimating mean fragment length of single-end short-read sequencing data","year":2013,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":33,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ottawa Hospital; University of Ottawa","funders":"Natural Sciences and Engineering Research Council of Canada; Ontario Ministry of Economic Development and Innovation; Ministero dello Sviluppo Economico; Ottawa Hospital Research Institute","keywords":"Fragment (logic); Computer science; Perl; Subroutine; Cross-correlation; Algorithm; Correlation; Reliability (semiconductor); Software; Data mining; Statistics; Mathematics","score_opus":0.04575954669842852,"score_gpt":0.28354031649707506,"score_spread":0.23778076979864654,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2108614159","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.010103609,0.00023297293,0.9653759,0.000086176486,0.000055310225,0.00010368107,0.0010064536,0.022575635,0.00046013022],"genre_scores_gemma":[0.052934457,0.00012657428,0.9419651,0.00012876398,0.00006253703,0.00029415707,0.0018595537,0.0019773338,0.0006514557],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9959345,0.0011822022,0.0002741703,0.0009051824,0.001532908,0.00017107438],"domain_scores_gemma":[0.9904295,0.0047201226,0.0015478847,0.0015596828,0.0014756285,0.00026721248],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007395383,0.0029121935,0.0012043627,0.003517188,0.0011214329,0.0014273342,0.002563876,0.0015835712,0.0045336476],"category_scores_gemma":[0.019787343,0.0013048253,0.0014710608,0.0031436682,0.0014053877,0.0016851823,0.002057724,0.0019946718,0.0038137238],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0021651203,0.00039123502,0.032221735,0.0021751039,0.0013851636,0.0013821861,0.0007429989,0.100676045,0.2316792,0.020868912,0.049769733,0.55654263],"study_design_scores_gemma":[0.00010687957,0.00024564995,0.009094755,0.00008504111,0.00014411354,0.0014791111,0.00007229227,0.7663455,0.19302562,0.012830749,0.016359424,0.00021089528],"about_ca_topic_score_codex":0.0020410887,"about_ca_topic_score_gemma":0.0030238088,"teacher_disagreement_score":0.007395383,"about_ca_system_score_codex":0.00086442864,"about_ca_system_score_gemma":0.002021728,"threshold_uncertainty_score":0.03911102},"labels":[],"label_agreement":null},{"id":"W2108998942","doi":"10.1093/bioinformatics/btg1025","title":"Detection and validation of single gene inversions","year":2003,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genome Rearrangement Algorithms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":56,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Ottawa; Université de Montréal; Université du Québec à Montréal","funders":"","keywords":"Computer science; Computational biology; Gene; Genetics; Biology","score_opus":0.013892619292329735,"score_gpt":0.21265513003258144,"score_spread":0.1987625107402517,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2108998942","genre_codex":"empirical","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.67296934,0.00032431888,0.3225387,0.00017851843,0.00003945925,0.00008693268,0.00043210463,0.0010593255,0.0023712975],"genre_scores_gemma":[0.89850473,0.000051052924,0.09992329,0.000047432284,0.000008775363,0.000052646352,0.001031847,0.00007560886,0.000304634],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99584806,0.001612866,0.00028282864,0.0011041738,0.00095099764,0.00020104769],"domain_scores_gemma":[0.97284734,0.018915882,0.0018924283,0.0035148591,0.0024654134,0.00036393627],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005511639,0.00044529288,0.0005065693,0.0017390472,0.00067222916,0.00092451024,0.0013292187,0.0009855194,0.0022389658],"category_scores_gemma":[0.041107025,0.00024382447,0.00046469513,0.0012658114,0.0014168331,0.0010188648,0.0010733717,0.00065709575,0.0005037347],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0025551438,0.0005262155,0.16553317,0.0012060043,0.0003603313,0.00057666615,0.0006830157,0.2968343,0.18753612,0.061546963,0.0015698242,0.2810723],"study_design_scores_gemma":[0.00012124624,0.00045229154,0.021249224,0.00004961256,0.00006145105,0.0007761273,0.00019228861,0.7325818,0.21270795,0.02962757,0.0021296847,0.000050735976],"about_ca_topic_score_codex":0.00083315646,"about_ca_topic_score_gemma":0.00068253605,"teacher_disagreement_score":0.005511639,"about_ca_system_score_codex":0.0007115626,"about_ca_system_score_gemma":0.000733698,"threshold_uncertainty_score":0.029148698},"labels":[],"label_agreement":null},{"id":"W2109028614","doi":"10.1093/bioinformatics/btp600","title":"Ancestors 1.0: a web server for ancestral sequence reconstruction","year":2009,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genome Rearrangement Algorithms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":43,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; Université du Québec à Montréal","funders":"","keywords":"Synteny; Inference; Computer science; Indel; Genome; Context (archaeology); Sequence (biology); Web server; Biology; Computational biology; Artificial intelligence; The Internet; Genetics; World Wide Web; Gene","score_opus":0.0301068154565614,"score_gpt":0.2728592275455003,"score_spread":0.24275241208893888,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2109028614","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.01057189,0.0009318163,0.32333234,0.00037560807,0.00027892092,0.0003687048,0.09398909,0.559301,0.01085063],"genre_scores_gemma":[0.07042555,0.0009594668,0.49374127,0.00034974585,0.00019748922,0.0008901766,0.35873333,0.06229067,0.012412309],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99919826,0.00015289239,0.00010129392,0.00018700834,0.00025438008,0.00010622003],"domain_scores_gemma":[0.9982728,0.0005399542,0.00018433837,0.0003788889,0.00040805663,0.00021597612],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002455977,0.0014309543,0.0012872424,0.0030450483,0.0009034425,0.0017343179,0.0034785762,0.0012413994,0.057391223],"category_scores_gemma":[0.0071768686,0.001709342,0.0011318756,0.002642569,0.00039338687,0.0020173667,0.002258249,0.0024597615,0.045765538],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.002028056,0.00023261746,0.0077202837,0.0018556027,0.00034413568,0.00093785126,0.00061235507,0.0075074118,0.020000558,0.01635679,0.6515255,0.29087883],"study_design_scores_gemma":[0.0012849849,0.00036230354,0.010455305,0.00057261286,0.00030025738,0.0029639455,0.00023720147,0.15694752,0.04492963,0.07191519,0.7095296,0.0005014393],"about_ca_topic_score_codex":0.0024575887,"about_ca_topic_score_gemma":0.0018298319,"teacher_disagreement_score":0.057391223,"about_ca_system_score_codex":0.00077881746,"about_ca_system_score_gemma":0.0017733664,"threshold_uncertainty_score":0.19199276},"labels":[],"label_agreement":null},{"id":"W2109363750","doi":"10.1093/bioinformatics/btq315","title":"Count: evolutionary analysis of phylogenetic profiles with parsimony and likelihood","year":2010,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":491,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Phylogenetic tree; Java; Phylogenetics; Computer science; Source code; Lineage (genetic); Maximum parsimony; Software; Biology; Gene; Programming language; Genetics; Clade","score_opus":0.005104281382279222,"score_gpt":0.203892047635604,"score_spread":0.19878776625332478,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2109363750","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.015547964,0.00071626087,0.89501935,0.00045614227,0.00021874992,0.00036863412,0.032654144,0.049934473,0.005084446],"genre_scores_gemma":[0.064205,0.00051642215,0.86897415,0.00015107827,0.00013439103,0.0014713856,0.03668071,0.02464291,0.0032239412],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9977634,0.00085229706,0.00025188184,0.00047052113,0.0005283203,0.00013353112],"domain_scores_gemma":[0.9933009,0.0045648147,0.00059039216,0.0006455933,0.0006509102,0.00024736143],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0042913794,0.002253696,0.003014641,0.0072597526,0.0021105653,0.0037591013,0.0035621128,0.0011463615,0.049022656],"category_scores_gemma":[0.020413024,0.0016473902,0.0029527757,0.006574394,0.0010051808,0.004896378,0.003100848,0.0036615334,0.014225093],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0018078818,0.00033945256,0.017195445,0.006283438,0.0022797387,0.0014099226,0.0023369414,0.060499344,0.014049836,0.084908076,0.32590124,0.48298866],"study_design_scores_gemma":[0.0005973096,0.00031396127,0.011558343,0.0006310372,0.0008346972,0.00200472,0.0004924831,0.60195434,0.011417674,0.16738743,0.20239446,0.00041352873],"about_ca_topic_score_codex":0.0018066457,"about_ca_topic_score_gemma":0.0022941292,"teacher_disagreement_score":0.049022656,"about_ca_system_score_codex":0.0009560562,"about_ca_system_score_gemma":0.0019062645,"threshold_uncertainty_score":0.16399711},"labels":[],"label_agreement":null},{"id":"W2109550992","doi":"10.1093/bioinformatics/bth946","title":"Into the heart of darkness: large-scale clustering of human non-coding DNA","year":2004,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genome Rearrangement Algorithms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":69,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"","keywords":"Human genome; Biology; Genome; Computational biology; Conserved sequence; Cluster analysis; Genetics; Gene; Coding region; Computer science; Peptide sequence; Artificial intelligence","score_opus":0.009885294791323542,"score_gpt":0.24885973694188881,"score_spread":0.23897444215056526,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2109550992","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.54751146,0.0021391325,0.44048175,0.0020989857,0.00013463928,0.0001349339,0.0018433934,0.002491508,0.0031641233],"genre_scores_gemma":[0.8110135,0.00051931565,0.1793771,0.00035584066,0.00009045187,0.00007028314,0.0056513585,0.00036490435,0.002557229],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99960333,0.000103657185,0.00001505429,0.00014067601,0.00010194624,0.000035287496],"domain_scores_gemma":[0.9984713,0.00083130377,0.00016602641,0.00021810464,0.00016627376,0.00014692945],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010176455,0.00036691836,0.00037858574,0.0016628087,0.0007406157,0.00091957214,0.0007892904,0.00047723137,0.0014889423],"category_scores_gemma":[0.005081486,0.00021861131,0.00027321794,0.0016637848,0.0007844294,0.0007982517,0.0010697005,0.00059545814,0.00047326187],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014261312,0.00030103407,0.09483663,0.0013269782,0.0004474193,0.00085405476,0.0028414272,0.1658921,0.14258271,0.074777775,0.040971503,0.47374225],"study_design_scores_gemma":[0.00014578867,0.00016377647,0.0795738,0.00014061871,0.00013555202,0.0008149602,0.0013263953,0.61880654,0.041758243,0.23207678,0.024958583,0.00009899122],"about_ca_topic_score_codex":0.0021811177,"about_ca_topic_score_gemma":0.0034285497,"teacher_disagreement_score":0.0021811177,"about_ca_system_score_codex":0.00047333472,"about_ca_system_score_gemma":0.00040997664,"threshold_uncertainty_score":0.005381882},"labels":[],"label_agreement":null},{"id":"W2109614625","doi":"10.1093/bioinformatics/btt041","title":"AbCD: arbitrary coverage design for sequencing-based genetic studies","year":2013,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":15,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"National Human Genome Research Institute; National Heart, Lung, and Blood Institute","keywords":"Computer science; Computational biology; Biology","score_opus":0.04451066268123887,"score_gpt":0.2829542883865914,"score_spread":0.23844362570535255,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2109614625","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0039165313,0.00026481072,0.97184414,0.00014791825,0.000108926404,0.0006330354,0.0021541775,0.019330263,0.0016001642],"genre_scores_gemma":[0.03630508,0.0001340626,0.9512413,0.00036202598,0.00003860863,0.0049321656,0.0019374641,0.0037803694,0.0012689258],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9906607,0.006062319,0.00064934476,0.0010219032,0.0013659315,0.00023982825],"domain_scores_gemma":[0.9567322,0.036097288,0.0017975314,0.0022578004,0.0025660004,0.00054906216],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.015566247,0.0018543395,0.0013315356,0.0015545742,0.00060770946,0.0015982266,0.002550393,0.0015548111,0.032357454],"category_scores_gemma":[0.047991,0.0016743491,0.0016474866,0.001137092,0.0009864767,0.0009314491,0.0023301688,0.0028222853,0.0045865388],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.008598047,0.0005805861,0.011614889,0.00561239,0.0015749842,0.0012146573,0.0012319442,0.18958984,0.051043335,0.08647348,0.090177074,0.5522887],"study_design_scores_gemma":[0.0034460095,0.0011706998,0.0031956583,0.0005211849,0.00040459927,0.0004446812,0.00008671164,0.72030467,0.05217018,0.08117261,0.13682589,0.00025704765],"about_ca_topic_score_codex":0.0013080082,"about_ca_topic_score_gemma":0.0017392045,"teacher_disagreement_score":0.032357454,"about_ca_system_score_codex":0.0010877884,"about_ca_system_score_gemma":0.0021221188,"threshold_uncertainty_score":0.108246446},"labels":[],"label_agreement":null},{"id":"W2109641307","doi":"10.1093/bioinformatics/18.6.894","title":"DMLE+: Bayesian linkage disequilibrium gene mapping","year":2002,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":137,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"National Institutes of Health; Canadian Institutes of Health Research; Killam Trusts; National Human Genome Research Institute; Fondation pour la Recherche Médicale","keywords":"Linkage disequilibrium; Bayesian probability; Linkage (software); Disequilibrium; Computational biology; Genetics; Computer science; Gene; Biology; Artificial intelligence; Haplotype; Allele; Medicine","score_opus":0.01966016657473886,"score_gpt":0.2320162652202325,"score_spread":0.21235609864549365,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2109641307","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.005622306,0.00034765646,0.84242237,0.00045250528,0.00013283378,0.00021492707,0.035523053,0.10747275,0.0078115473],"genre_scores_gemma":[0.061770696,0.0003585822,0.8727089,0.0005544783,0.00009483269,0.0019962818,0.037405793,0.01623612,0.008874319],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9986039,0.0006337389,0.00007728262,0.0002958009,0.00029673072,0.0000925634],"domain_scores_gemma":[0.99733126,0.0019849653,0.0001656399,0.00020680655,0.00020396413,0.00010731889],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0029730815,0.0012317472,0.0013755216,0.0017694917,0.00072729937,0.0016143749,0.0028260858,0.0007661376,0.056569047],"category_scores_gemma":[0.008841963,0.0012022651,0.0010836966,0.0015026812,0.00035846143,0.0011051586,0.0020672223,0.0016657424,0.020657234],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001334435,0.00024229594,0.006379881,0.0017438022,0.0008483466,0.00066160195,0.0004149635,0.06388234,0.007992024,0.04236867,0.48705032,0.38708127],"study_design_scores_gemma":[0.002072152,0.00012650255,0.004207274,0.00029307837,0.0002778549,0.0011038244,0.00007557847,0.5293057,0.019056201,0.12534824,0.31789228,0.00024136195],"about_ca_topic_score_codex":0.0016085521,"about_ca_topic_score_gemma":0.0022058485,"teacher_disagreement_score":0.056569047,"about_ca_system_score_codex":0.00042265814,"about_ca_system_score_gemma":0.000918559,"threshold_uncertainty_score":0.1892423},"labels":[],"label_agreement":null},{"id":"W2109643429","doi":"10.1093/bioinformatics/bti040","title":"Stepwise detection of recombination breakpoints in sequence alignments","year":2004,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":20,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"Eunice Kennedy Shriver National Institute of Child Health and Human Development; National Institute of Allergy and Infectious Diseases","keywords":"Breakpoint; Sequence (biology); Recombination; Genetics; Computer science; Computational biology; Biology; Algorithm; Gene; Chromosome","score_opus":0.01422088828083883,"score_gpt":0.23910917761121972,"score_spread":0.22488828933038088,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2109643429","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.12963626,0.0003467306,0.86582,0.00016520593,0.00003629733,0.00030695144,0.000467273,0.0025513738,0.0006699326],"genre_scores_gemma":[0.5101609,0.0001358094,0.48703182,0.00012841621,0.000043431977,0.000285948,0.0013686108,0.00028705786,0.0005579109],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99228173,0.003602226,0.000542434,0.00173382,0.0014815305,0.00035812342],"domain_scores_gemma":[0.94214624,0.042484395,0.0060911435,0.00466462,0.0035912462,0.001022467],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008507091,0.0008747098,0.0012375544,0.0037278065,0.00074002735,0.0015098088,0.0021110997,0.0013472558,0.0024678737],"category_scores_gemma":[0.051790718,0.00066327874,0.0016477369,0.0023922452,0.0010140202,0.0019301496,0.0022784662,0.0022325686,0.0014053201],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0027564198,0.00078380416,0.20734699,0.001597056,0.0012543252,0.0022421912,0.0019990797,0.089697234,0.12226732,0.014625166,0.004413966,0.5510165],"study_design_scores_gemma":[0.0003967264,0.0014745425,0.07165616,0.00024225049,0.00050664955,0.005259813,0.00058213435,0.7715357,0.092589654,0.04907416,0.0063798362,0.00030240646],"about_ca_topic_score_codex":0.0010220488,"about_ca_topic_score_gemma":0.0018962377,"teacher_disagreement_score":0.008507091,"about_ca_system_score_codex":0.0003790384,"about_ca_system_score_gemma":0.0012014032,"threshold_uncertainty_score":0.0449903},"labels":[],"label_agreement":null},{"id":"W2110019612","doi":"10.1093/bioinformatics/btp478","title":"Methods for labeling error detection in microarrays based on the effect of data perturbation on the regression model","year":2009,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":27,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"University of British Columbia","keywords":"Support vector machine; Computer science; Artificial intelligence; Pattern recognition (psychology); Algorithm; Regression; Data mining; DNA microarray; Source code; Mathematics; Statistics","score_opus":0.05680105849511773,"score_gpt":0.3634192508253875,"score_spread":0.30661819233026977,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2110019612","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0050753774,0.00009023504,0.99394304,0.00003904188,0.000012132445,0.000033283515,0.000017827875,0.00067616056,0.00011290227],"genre_scores_gemma":[0.14399478,0.00019675853,0.85436034,0.00007480649,0.000052992913,0.00028344145,0.00024032203,0.00017869781,0.00061780435],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9948035,0.0018198881,0.00031250992,0.0012092966,0.0016963806,0.00015833274],"domain_scores_gemma":[0.9856548,0.008614985,0.001584354,0.0017216839,0.0022503892,0.0001738733],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0056252005,0.0011080148,0.0011703416,0.0018915632,0.0005315773,0.0009920412,0.0016476121,0.0011291923,0.00081044756],"category_scores_gemma":[0.020547884,0.00059065636,0.00086659135,0.0013529435,0.0011282595,0.0011608785,0.0012281892,0.0015643962,0.0005670259],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00053776015,0.00015629434,0.0108165955,0.00040804822,0.00023095413,0.00019406158,0.00037826205,0.15803984,0.088064104,0.009849041,0.0025167416,0.7288083],"study_design_scores_gemma":[0.00001572973,0.00010250969,0.0028389855,0.000020618094,0.000038022972,0.0001746225,0.000030341223,0.9406049,0.050940264,0.003844112,0.0013524339,0.000037493002],"about_ca_topic_score_codex":0.0010389077,"about_ca_topic_score_gemma":0.00096374896,"teacher_disagreement_score":0.0056252005,"about_ca_system_score_codex":0.00091891916,"about_ca_system_score_gemma":0.0009534141,"threshold_uncertainty_score":0.029749274},"labels":[],"label_agreement":null},{"id":"W2110092182","doi":"10.1093/bioinformatics/bts349","title":"MolClass: a web portal to interrogate diverse small molecule screen datasets with different computational models","year":2012,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Computational Drug Discovery Methods","field":"Computer Science","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal; Institute for Research in Immunology and Cancer","funders":"Wellcome Trust","keywords":"Computer science; Software; Source code; Fingerprint (computing); Database; Web service; Python (programming language); Data mining; Cheminformatics; Function (biology); Open source; World Wide Web; Information retrieval; Bioinformatics; Programming language; Artificial intelligence; Biology","score_opus":0.046192672568302624,"score_gpt":0.2841637598046638,"score_spread":0.23797108723636118,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2110092182","genre_codex":"software","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00689427,0.00070954947,0.102453,0.0006697467,0.0002240687,0.0004889357,0.25531456,0.6251428,0.008103113],"genre_scores_gemma":[0.041674994,0.0009655404,0.15133587,0.0011968001,0.00012645945,0.0031069987,0.69749624,0.09547459,0.008622539],"study_design_codex":"not_applicable","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99842983,0.00026160266,0.00018212493,0.0003269039,0.0006469534,0.00015252427],"domain_scores_gemma":[0.9944402,0.0027485522,0.00039750052,0.0013660731,0.0005323146,0.0005153926],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005150741,0.0035120824,0.0032222527,0.004559073,0.0012685341,0.0050771697,0.007290416,0.002496653,0.11511081],"category_scores_gemma":[0.009332831,0.0019441425,0.0026220733,0.004204699,0.0006707918,0.0046821246,0.0049179792,0.0031263223,0.04583543],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013544048,0.00043279224,0.005758977,0.002490906,0.0005295192,0.00042644504,0.00023294524,0.006663589,0.0045046923,0.0069190846,0.8986067,0.0720798],"study_design_scores_gemma":[0.0024813854,0.00032778835,0.007072117,0.0005276543,0.00021926455,0.00078560377,0.00029074357,0.19281772,0.025767054,0.029940864,0.7393433,0.00042658392],"about_ca_topic_score_codex":0.003345817,"about_ca_topic_score_gemma":0.006301856,"teacher_disagreement_score":0.11511081,"about_ca_system_score_codex":0.0017524159,"about_ca_system_score_gemma":0.002830945,"threshold_uncertainty_score":0.38508403},"labels":[],"label_agreement":null},{"id":"W2110628507","doi":"10.1093/bioinformatics/18.suppl_2.s54","title":"Common intervals and sorting by reversals: a marriage of necessity","year":2002,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genome Rearrangement Algorithms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":41,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Montréal","funders":"","keywords":"Sorting; Computer science; Econometrics; Psychology; Algorithm; Mathematics","score_opus":0.014919944998585408,"score_gpt":0.23653411034470762,"score_spread":0.22161416534612222,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2110628507","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.021799585,0.0018265636,0.9593397,0.0018390426,0.00026952237,0.00004904118,0.00009394931,0.00030747207,0.014475086],"genre_scores_gemma":[0.49381006,0.0033750082,0.4922226,0.0012784171,0.001247253,0.00025734922,0.00030069822,0.00047035833,0.0070382548],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9943936,0.0016484556,0.00052094134,0.0013851918,0.0016208756,0.00043084775],"domain_scores_gemma":[0.97771126,0.016716603,0.0010336308,0.0025653448,0.0014723865,0.00050082896],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0058420873,0.00062506844,0.0014864774,0.0028765176,0.0023787166,0.005084956,0.0022869501,0.0019037358,0.0059887543],"category_scores_gemma":[0.030299373,0.0007765679,0.001903619,0.003491377,0.009510373,0.019218506,0.0055168164,0.004436852,0.0011588604],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00004978994,0.0000118210955,0.0003003992,0.00007520696,0.000008065777,0.00007109445,0.00028062655,0.0016014525,0.000587423,0.9656002,0.00069793244,0.030715875],"study_design_scores_gemma":[0.000013854711,0.000036052897,0.00018163692,0.000031551994,0.000010571678,0.00019221417,0.00011286753,0.0048240013,0.00069193303,0.9857091,0.008169621,0.000026520027],"about_ca_topic_score_codex":0.0009670229,"about_ca_topic_score_gemma":0.0005683513,"teacher_disagreement_score":0.0059887543,"about_ca_system_score_codex":0.0013133967,"about_ca_system_score_gemma":0.0012832223,"threshold_uncertainty_score":0.030896306},"labels":[],"label_agreement":null},{"id":"W2110783096","doi":"10.1093/bioinformatics/btq264","title":"GPCR-OKB: the G Protein Coupled Receptor Oligomer Knowledge Base","year":2010,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Receptor Mechanisms and Signaling","field":"Biochemistry, Genetics and Molecular Biology","cited_by":75,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal; Institute for Research in Immunology and Cancer","funders":"National Institute on Drug Abuse; National Institute of Mental Health; National Institutes of Health; Norwegian Biodiversity Information Centre","keywords":"Oligomer; G protein-coupled receptor; Base (topology); Receptor; Computer science; Computational biology; Chemistry; Biology; Biochemistry; Mathematics; Polymer chemistry","score_opus":0.009851811084422688,"score_gpt":0.23317525974567035,"score_spread":0.22332344866124768,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2110783096","genre_codex":"dataset","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.01255421,0.0018958984,0.34456143,0.002384507,0.0003859263,0.0010947684,0.40409172,0.21450488,0.018526684],"genre_scores_gemma":[0.036673807,0.0015449256,0.20874368,0.0009012752,0.0000800453,0.00092756376,0.7386092,0.00801631,0.004503096],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9991109,0.000119498654,0.00014223809,0.00018806789,0.00034714807,0.000092119226],"domain_scores_gemma":[0.99779946,0.0007670991,0.0002526941,0.00042252804,0.0004614974,0.0002967627],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0019465828,0.001213729,0.0011106589,0.0033368054,0.0008779542,0.0029611688,0.0033089956,0.0018739507,0.020163758],"category_scores_gemma":[0.006347743,0.0009559528,0.0011124102,0.0036314013,0.00049291947,0.003248442,0.0032524634,0.0025273846,0.018409094],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0019054003,0.00057247875,0.0045695077,0.0056967186,0.00033792213,0.0010702872,0.00059623877,0.01522306,0.038159747,0.045452196,0.6816305,0.2047859],"study_design_scores_gemma":[0.0007074309,0.00015524241,0.0067455615,0.0008966133,0.00026852379,0.0008934542,0.00022724534,0.06346443,0.032745216,0.038178936,0.85544765,0.0002697396],"about_ca_topic_score_codex":0.0080599785,"about_ca_topic_score_gemma":0.005549409,"teacher_disagreement_score":0.020163758,"about_ca_system_score_codex":0.00097214046,"about_ca_system_score_gemma":0.0028278776,"threshold_uncertainty_score":0.06745446},"labels":[],"label_agreement":null},{"id":"W2110845299","doi":"10.1093/bioinformatics/btn226","title":"Malin: maximum likelihood analysis of intron evolution in eukaryotes","year":2008,"lang":"en","type":"article","venue":"Bioinformatics","topic":"RNA and protein synthesis mechanisms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":51,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Intron; Biology; Computational biology; Exon; Phylogenetic tree; Gene; Genetics","score_opus":0.009916707369705392,"score_gpt":0.22100113472890803,"score_spread":0.21108442735920263,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2110845299","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.028297624,0.001163129,0.94137263,0.000583724,0.000085959335,0.000069943584,0.005853686,0.018660126,0.0039131944],"genre_scores_gemma":[0.32226276,0.00065848394,0.647163,0.00031774916,0.00017799703,0.00051349675,0.016012955,0.007852583,0.005040999],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99914396,0.00046128262,0.000040766536,0.00019167642,0.00011371797,0.000048616977],"domain_scores_gemma":[0.9978727,0.0016088237,0.00018277772,0.00015008525,0.00009555078,0.0000899978],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0021256993,0.0012097952,0.0010107977,0.0016402386,0.0009728428,0.0014105956,0.0020484487,0.0009949844,0.015926117],"category_scores_gemma":[0.007284397,0.00056937576,0.0011152597,0.0011815408,0.00063055166,0.0011892744,0.0013706686,0.0014401863,0.005924253],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0022584344,0.00027218353,0.03651487,0.0023787264,0.00070921675,0.0013686332,0.0008715814,0.2688434,0.04648379,0.07860927,0.09911707,0.46257287],"study_design_scores_gemma":[0.00016275814,0.00010175716,0.008439832,0.00015188084,0.00011711709,0.0008259685,0.00009881944,0.8480801,0.014687379,0.07566792,0.051577643,0.00008881057],"about_ca_topic_score_codex":0.00042740832,"about_ca_topic_score_gemma":0.0007115659,"teacher_disagreement_score":0.015926117,"about_ca_system_score_codex":0.00046277273,"about_ca_system_score_gemma":0.00052356155,"threshold_uncertainty_score":0.053278208},"labels":[],"label_agreement":null},{"id":"W2111044311","doi":"10.1093/bioinformatics/bts593","title":"SCALCE: boosting sequence compression algorithms using locally consistent encoding","year":2012,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Algorithms and Data Compression","field":"Computer Science","cited_by":158,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"National Human Genome Research Institute; Natural Sciences and Engineering Research Council of Canada; ExxonMobil Research and Engineering Company","keywords":"Boosting (machine learning); Encoding (memory); Computer science; Algorithm; Sequence (biology); Data compression; Compression (physics); Artificial intelligence","score_opus":0.08296651646590367,"score_gpt":0.3066863161757942,"score_spread":0.22371979970989056,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2111044311","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.028151931,0.00075192685,0.96184486,0.00025357504,0.00015312049,0.00015015787,0.00023823326,0.0062063974,0.0022497263],"genre_scores_gemma":[0.21090308,0.0005344778,0.77990866,0.00061365444,0.0001737837,0.00031019066,0.0016163932,0.00075844454,0.0051813056],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9993893,0.000105966225,0.00003702645,0.000091916954,0.00030899353,0.00006666286],"domain_scores_gemma":[0.9986547,0.00055283617,0.00011054203,0.00029210706,0.0003269042,0.000062757346],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011360662,0.0007434042,0.0007818696,0.0011437774,0.00053256424,0.00082305825,0.0017111396,0.0009251098,0.0029771521],"category_scores_gemma":[0.003878002,0.00027316794,0.00053964136,0.0013739335,0.0008825218,0.0013260893,0.0012438234,0.0016453085,0.0018355786],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00080255774,0.000223342,0.002414393,0.0002858059,0.00007268279,0.00033287256,0.00031011662,0.17430414,0.05288671,0.029117415,0.018803982,0.72044593],"study_design_scores_gemma":[0.00007370277,0.00023161626,0.0005238502,0.0000343046,0.000020617561,0.00021216633,0.000052855557,0.93806344,0.03338635,0.017943224,0.009420172,0.000037749025],"about_ca_topic_score_codex":0.0025228902,"about_ca_topic_score_gemma":0.002297467,"teacher_disagreement_score":0.0029771521,"about_ca_system_score_codex":0.0007019218,"about_ca_system_score_gemma":0.0010077399,"threshold_uncertainty_score":0.0099595785},"labels":[],"label_agreement":null},{"id":"W2111451842","doi":"10.1093/bioinformatics/btv284","title":"Identification of C2H2-ZF binding preferences from ChIP-seq data using RCADE","year":2015,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":35,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canadian Institute for Advanced Research; University of Toronto","funders":"Canadian Institutes of Health Research","keywords":"Identification (biology); Chip; Computer science; Software; Data mining; Programming language; Telecommunications; Biology","score_opus":0.061321560761717665,"score_gpt":0.2850528448438939,"score_spread":0.22373128408217627,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2111451842","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.21080786,0.0012333561,0.67977035,0.00061288814,0.00020764163,0.00037115093,0.028865471,0.07248677,0.0056444756],"genre_scores_gemma":[0.25911292,0.00055527757,0.70208436,0.00033470677,0.000038394657,0.0004986765,0.028582186,0.0056543,0.0031391995],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99928254,0.00009784603,0.000032777752,0.00032681227,0.00020717956,0.000052951113],"domain_scores_gemma":[0.998099,0.0010646754,0.00018079544,0.0003311018,0.00020428744,0.00012001128],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015195735,0.0012206447,0.0009411981,0.0012344221,0.0005881173,0.0009146098,0.0018664214,0.00085293193,0.008316258],"category_scores_gemma":[0.0030998606,0.0006526343,0.001046775,0.0010165621,0.00052620866,0.0005975099,0.0007685793,0.0014061936,0.0051403013],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.002530615,0.00036539804,0.035294496,0.002774922,0.0009037766,0.000742397,0.0005235833,0.056374777,0.6742396,0.009013712,0.045321617,0.17191519],"study_design_scores_gemma":[0.00014944309,0.0001915061,0.011797359,0.000067487046,0.00013507665,0.0007276745,0.000117313546,0.44477624,0.50609815,0.0033443354,0.03240339,0.00019199325],"about_ca_topic_score_codex":0.00263251,"about_ca_topic_score_gemma":0.0073308037,"teacher_disagreement_score":0.008316258,"about_ca_system_score_codex":0.0006066863,"about_ca_system_score_gemma":0.0011081917,"threshold_uncertainty_score":0.027820647},"labels":[],"label_agreement":null},{"id":"W2111510868","doi":"10.1093/bioinformatics/btq200","title":"Model-based detection of alternative splicing signals","year":2010,"lang":"en","type":"article","venue":"Bioinformatics","topic":"RNA Research and Splicing","field":"Biochemistry, Genetics and Molecular Biology","cited_by":23,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"National Cancer Institute; Natural Sciences and Engineering Research Council of Canada; Canadian Institutes of Health Research; Genome Canada; Ontario Genomics; Canadian Institute for Advanced Research; Ontario Genomics Institute; Ontario Innovation Trust","keywords":"Alternative splicing; RNA splicing; Exon; Computational biology; Probabilistic logic; Computer science; Proteome; Identification (biology); Exon skipping; Biology; Data mining; Gene; Bioinformatics; Artificial intelligence; Genetics; RNA","score_opus":0.016073067615979535,"score_gpt":0.2758594418579076,"score_spread":0.2597863742419281,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2111510868","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.07921438,0.0004264849,0.9164403,0.00046393502,0.00003639435,0.00006972746,0.00070351624,0.0017175955,0.0009276161],"genre_scores_gemma":[0.82039046,0.00037249023,0.17309609,0.000282779,0.00010892334,0.00025725257,0.0032466645,0.00019275688,0.0020526147],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9987424,0.00035588857,0.000075295895,0.00045279917,0.00026031307,0.00011345924],"domain_scores_gemma":[0.9948592,0.003965771,0.00038320664,0.00025428322,0.0004120227,0.0001256011],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0024513833,0.0010854806,0.0014724131,0.0012572429,0.00037324245,0.0014461413,0.0018571456,0.0017812313,0.0012802668],"category_scores_gemma":[0.006418166,0.00057863304,0.0012259416,0.0011648013,0.00089053804,0.0012749746,0.0010128263,0.0014859565,0.0006373428],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00032315863,0.00011118097,0.005790795,0.00016695684,0.00011680816,0.00021679382,0.00007678525,0.9290689,0.00873358,0.005643369,0.0015779828,0.048173584],"study_design_scores_gemma":[0.000007698653,0.000013058932,0.0002865244,0.0000025532634,0.000006526886,0.000034697085,0.000002463793,0.9955642,0.000947183,0.0029878716,0.00014203019,0.0000052129258],"about_ca_topic_score_codex":0.0039063687,"about_ca_topic_score_gemma":0.002810542,"teacher_disagreement_score":0.0039063687,"about_ca_system_score_codex":0.0009842194,"about_ca_system_score_gemma":0.0009977102,"threshold_uncertainty_score":0.012964368},"labels":[],"label_agreement":null},{"id":"W2111642038","doi":"10.1093/bioinformatics/btp415","title":"Statistical lower bounds on protein copy number from fluorescence expression images","year":2009,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomic variations and chromosomal abnormalities","field":"Biochemistry, Genetics and Molecular Biology","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Concordia University; Ottawa Hospital","funders":"Concordia University; Ottawa Hospital Research Institute; Natural Sciences and Engineering Research Council of Canada; McGill University","keywords":"Expression (computer science); Computer science; Noise (video); Estimator; Drosophila melanogaster; Computational biology; Segmentation; Algorithm; Pattern recognition (psychology); Biology; Gene; Artificial intelligence; Mathematics; Statistics; Genetics; Image (mathematics)","score_opus":0.006398281335425863,"score_gpt":0.23268936742304722,"score_spread":0.22629108608762136,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2111642038","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.039496537,0.00083536876,0.95572275,0.0005125075,0.000028877514,0.000034829245,0.00065191305,0.00075956446,0.0019576715],"genre_scores_gemma":[0.6159047,0.001787355,0.3722951,0.0004879411,0.0003393247,0.0003962702,0.0045499727,0.000860954,0.0033783081],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99627554,0.0009668566,0.00023345996,0.0010848282,0.0012126659,0.00022653733],"domain_scores_gemma":[0.895526,0.09009908,0.0040515657,0.0061916607,0.0033560416,0.00077558117],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008948093,0.0013199596,0.0015096155,0.002670885,0.0007426717,0.0027362588,0.0029636882,0.001978097,0.0036572637],"category_scores_gemma":[0.07283122,0.0010247993,0.0010364609,0.0020318918,0.0029566528,0.004441568,0.0027935186,0.0029297043,0.0009293593],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006574018,0.00012515111,0.023423582,0.00076530676,0.00026056656,0.00027935815,0.000431134,0.69399697,0.03517686,0.13560016,0.003614693,0.10566877],"study_design_scores_gemma":[0.000023245391,0.00005101713,0.0058428324,0.00006815105,0.000036950994,0.00020813094,0.00003101678,0.8976684,0.016558139,0.077626154,0.0018306783,0.00005530453],"about_ca_topic_score_codex":0.0018778077,"about_ca_topic_score_gemma":0.0022228316,"teacher_disagreement_score":0.008948093,"about_ca_system_score_codex":0.0025515761,"about_ca_system_score_gemma":0.0011402125,"threshold_uncertainty_score":0.04732257},"labels":[],"label_agreement":null},{"id":"W2111693807","doi":"10.1093/bioinformatics/btq266","title":"DETECT—a Density Estimation Tool for Enzyme ClassificaTion and its application to <i>Plasmodium falciparum</i>","year":2010,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Machine Learning in Bioinformatics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":60,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"SickKids Foundation; Canada Research Chairs; Hospital for Sick Children; University of Toronto","funders":"Ontario Ministry of Research and Innovation; Canadian Institutes of Health Research","keywords":"Annotation; Ranking (information retrieval); Computer science; Computational biology; Probabilistic logic; Plasmodium falciparum; Sequence alignment; Genomics; Data mining; Biology; Machine learning; Artificial intelligence; Genetics; Gene; Peptide sequence; Genome; Malaria","score_opus":0.009605648486673174,"score_gpt":0.2615764797752238,"score_spread":0.25197083128855063,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2111693807","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.013356314,0.00024779257,0.8897051,0.00026615997,0.000053282274,0.000108231805,0.0040221973,0.090537384,0.0017036325],"genre_scores_gemma":[0.14175212,0.00033760368,0.8388001,0.00018512974,0.00009954309,0.0004831448,0.00878013,0.005949579,0.0036125346],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9990722,0.00016961659,0.00006873184,0.00021158485,0.0004246054,0.000053226497],"domain_scores_gemma":[0.99734503,0.00169123,0.00032782613,0.00020846781,0.00033763947,0.00008987774],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0019748707,0.0013775247,0.0010217061,0.003232065,0.00082518597,0.0013848599,0.0019942645,0.0011156455,0.010065322],"category_scores_gemma":[0.009451518,0.0008713374,0.00094308757,0.001717621,0.00048883003,0.0011756221,0.0016801681,0.0010481093,0.0042508515],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00088976184,0.00028583038,0.021674521,0.0009242852,0.00028578023,0.0006091193,0.0003854935,0.053563084,0.024640173,0.012275259,0.124436006,0.76003075],"study_design_scores_gemma":[0.00013888681,0.00010051159,0.0032098822,0.00003659848,0.000041508272,0.00071601255,0.000059478494,0.92937124,0.020313159,0.013185812,0.03275328,0.00007364997],"about_ca_topic_score_codex":0.0036337364,"about_ca_topic_score_gemma":0.004399649,"teacher_disagreement_score":0.010065322,"about_ca_system_score_codex":0.00081888115,"about_ca_system_score_gemma":0.001081076,"threshold_uncertainty_score":0.033671856},"labels":[],"label_agreement":null},{"id":"W2111716459","doi":"10.1093/bioinformatics/btt611","title":"WaveCNV: allele-specific copy number alterations in primary tumors and xenograft models from next-generation sequencing","year":2013,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomic variations and chromosomal abnormalities","field":"Biochemistry, Genetics and Molecular Biology","cited_by":19,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto; Ontario Institute for Cancer Research","funders":"Ontario Ministry of Research and Innovation","keywords":"Copy-number variation; Copy number analysis; Genome; Computational biology; DNA sequencing; Biology; Gene duplication; DNA microarray; Segmental duplication; Executable; Genetics; Comparative genomic hybridization; Computer science; Gene","score_opus":0.03453435057544929,"score_gpt":0.21140131668597067,"score_spread":0.17686696611052138,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2111716459","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.023773748,0.00030574936,0.71538794,0.00043039312,0.00022097111,0.0003728111,0.019782027,0.23522767,0.004498706],"genre_scores_gemma":[0.10261614,0.00056690694,0.7801164,0.0005039375,0.00007082774,0.0029457007,0.042969663,0.060849074,0.009361319],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99912137,0.00011184249,0.000081305705,0.0002232661,0.00039382413,0.00006824353],"domain_scores_gemma":[0.9980915,0.0011350262,0.00023695343,0.00022267381,0.00023500723,0.000078811834],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0027483604,0.0012361112,0.0008789774,0.0023148882,0.00051885063,0.001562319,0.0025017038,0.0012612393,0.02223969],"category_scores_gemma":[0.005976012,0.0011212389,0.0012428635,0.0011688949,0.0004614642,0.0011755208,0.0014405752,0.0013837591,0.005914917],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0020266098,0.00028358033,0.014993118,0.0027573619,0.00069357,0.0015726595,0.00084849296,0.106500216,0.14692621,0.016257318,0.27711242,0.4300284],"study_design_scores_gemma":[0.00038279587,0.00028795202,0.010279213,0.0001946364,0.00015570867,0.0010966355,0.00008726596,0.72672844,0.14377439,0.021354245,0.09535859,0.00030015717],"about_ca_topic_score_codex":0.0033080932,"about_ca_topic_score_gemma":0.0052419486,"teacher_disagreement_score":0.02223969,"about_ca_system_score_codex":0.0010841272,"about_ca_system_score_gemma":0.0014563196,"threshold_uncertainty_score":0.07439917},"labels":[],"label_agreement":null},{"id":"W2112005039","doi":"10.1093/bioinformatics/btr629","title":"Feature-based classifiers for somatic mutation detection in tumour–normal paired sequencing data","year":2011,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Cancer Genomics and Diagnostics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":154,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canada's Michael Smith Genome Sciences Centre; University of British Columbia; BC Cancer Agency","funders":"Canadian Institutes of Health Research; Michael Smith Health Research BC","keywords":"Computer science; Discriminative model; Artificial intelligence; Feature selection; Random forest; Machine learning; Ground truth; Genome; Support vector machine; Germline mutation; Cluster analysis; Computational biology; Mutation; Data mining; Biology; Genetics; Gene","score_opus":0.047249894469341115,"score_gpt":0.2508523294330022,"score_spread":0.20360243496366107,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2112005039","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.4289642,0.0011234031,0.5470049,0.00051342574,0.000156165,0.00039245764,0.0073943892,0.012716702,0.0017343628],"genre_scores_gemma":[0.7761774,0.00013840101,0.21371318,0.00012768096,0.000049691967,0.0003085146,0.008533518,0.00015368823,0.0007978436],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9984401,0.0004364193,0.0001742438,0.00038573067,0.00043102689,0.00013248993],"domain_scores_gemma":[0.9940842,0.0036663737,0.00040990033,0.00054041296,0.0011752615,0.00012399237],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003874421,0.0008612051,0.0007372761,0.0017434342,0.0005458762,0.0006702611,0.0013941879,0.0010863194,0.0021537705],"category_scores_gemma":[0.01274817,0.00022500387,0.0007596847,0.0011551166,0.00035770427,0.0010261484,0.00066853216,0.0010461445,0.0011734908],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001377794,0.00081787363,0.05814926,0.0005156705,0.00033412932,0.00052428246,0.00015737805,0.2957037,0.02405124,0.002512352,0.014209706,0.6016466],"study_design_scores_gemma":[0.00003888101,0.00018817908,0.0066486686,0.00002851187,0.000037835798,0.00014687989,0.000035798374,0.97181183,0.016552014,0.0033486523,0.001138376,0.000024361067],"about_ca_topic_score_codex":0.0025098657,"about_ca_topic_score_gemma":0.0024616919,"teacher_disagreement_score":0.003874421,"about_ca_system_score_codex":0.00065399194,"about_ca_system_score_gemma":0.000732153,"threshold_uncertainty_score":0.02049017},"labels":[],"label_agreement":null},{"id":"W2112034631","doi":"10.1093/bioinformatics/19.2.169","title":"Computational and experimental analysis reveals a novel Src family kinase in the C. elegans genome","year":2003,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetics, Aging, and Longevity in Model Organisms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Biology; Proto-oncogene tyrosine-protein kinase Src; Gene; Genetics; Gene family; Open reading frame; Genome; Sequence analysis; Coding region; Phylogenetic tree; Myristoylation; Conserved sequence; Kinase; Computational biology; Peptide sequence; Phosphorylation","score_opus":0.013691928919307655,"score_gpt":0.24233260069530932,"score_spread":0.22864067177600167,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2112034631","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.983887,0.0004731136,0.009773745,0.00014196915,0.000018472903,0.000034775578,0.0027101438,0.000935648,0.0020251942],"genre_scores_gemma":[0.9337367,0.00043499644,0.053403016,0.00006494449,0.000009121034,0.00006319074,0.011608344,0.00007561498,0.0006041828],"study_design_codex":"simulation_or_modeling","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99993134,0.000009266527,0.0000037417774,0.000029130057,0.000017223505,0.000009391912],"domain_scores_gemma":[0.9998753,0.00007383959,0.000013445971,0.000009063125,0.00001941273,0.000008928276],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00026829503,0.00034573162,0.00028282465,0.0005247984,0.00043350089,0.0003719646,0.00042610083,0.0003322145,0.0011130701],"category_scores_gemma":[0.0006902855,0.00015591856,0.0005669535,0.0004875654,0.00018975521,0.0003117643,0.00014700193,0.00025925116,0.00015667311],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0027131215,0.0006811299,0.14008269,0.0022462192,0.0005441665,0.0024930502,0.00026937935,0.5086245,0.2097931,0.015238759,0.011011793,0.10630213],"study_design_scores_gemma":[0.00017341087,0.000219962,0.04747409,0.000034851204,0.00020152995,0.0003787685,0.000106513864,0.9229884,0.02007241,0.0038592848,0.0044656536,0.00002511809],"about_ca_topic_score_codex":0.006165377,"about_ca_topic_score_gemma":0.012428506,"teacher_disagreement_score":0.006165377,"about_ca_system_score_codex":0.0005056062,"about_ca_system_score_gemma":0.00064951467,"threshold_uncertainty_score":0.0122590065},"labels":[],"label_agreement":null},{"id":"W2112072385","doi":"10.1093/bioinformatics/btv006","title":"Integrating alignment-based and alignment-free sequence similarity measures for biological sequence classification","year":2015,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":43,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ontario Institute for Cancer Research","funders":"Government of Ontario; Ontario Institute for Cancer Research","keywords":"Weighting; Similarity (geometry); Sequence alignment; Sequence (biology); Classifier (UML); Multiple sequence alignment; Computer science; Artificial intelligence; Pattern recognition (psychology); Set (abstract data type); Metagenomics; Protein sequencing; Data mining; Computational biology; Biology; Genetics; Peptide sequence","score_opus":0.12212028373722353,"score_gpt":0.30315173594708955,"score_spread":0.18103145220986602,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2112072385","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.06315179,0.0019416136,0.9268516,0.00072280556,0.00018613599,0.0002862728,0.0014955713,0.002875872,0.0024883912],"genre_scores_gemma":[0.48490122,0.0009998638,0.50637287,0.00043460072,0.0003590842,0.0004974648,0.0043385983,0.00038257448,0.0017138091],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99570185,0.0011041565,0.0004097823,0.001092353,0.0015146908,0.0001771538],"domain_scores_gemma":[0.9914671,0.0042365286,0.0010702463,0.0010472792,0.0017706485,0.00040823952],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005754065,0.0012877432,0.0013746768,0.005126124,0.0006586735,0.0022530367,0.0024222166,0.001923267,0.0016360655],"category_scores_gemma":[0.01643543,0.00034694903,0.0013460045,0.004337546,0.0011664891,0.0045302464,0.0020311938,0.0022617446,0.0018908171],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008716943,0.00079836225,0.053598702,0.0012119283,0.00065298093,0.00025463125,0.0004680912,0.20024924,0.034405448,0.021192199,0.011511183,0.6747856],"study_design_scores_gemma":[0.000026746931,0.00020343772,0.004867818,0.00007852886,0.000072418865,0.00017407807,0.000068502704,0.94595355,0.008733617,0.034758154,0.0050105928,0.000052620115],"about_ca_topic_score_codex":0.002693151,"about_ca_topic_score_gemma":0.0028183642,"teacher_disagreement_score":0.005754065,"about_ca_system_score_codex":0.0015147063,"about_ca_system_score_gemma":0.0013056352,"threshold_uncertainty_score":0.030430794},"labels":[],"label_agreement":null},{"id":"W2112491476","doi":"10.1093/bioinformatics/btp367","title":"<i>De novo</i> transcriptome assembly with ABySS","year":2009,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":410,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"BC Cancer Agency","funders":"Genome British Columbia; Michael Smith Health Research BC; National Cancer Institute; Genome Canada","keywords":"Contig; Sequence assembly; Transcriptome; Genome; Computational biology; Java; Biology; De novo transcriptome assembly; Software; Computer science; Hybrid genome assembly; Source code; Shotgun sequencing; Reference genome; Perl; Genetics; Gene; Programming language; Gene expression","score_opus":0.009809137250629119,"score_gpt":0.24134644448210438,"score_spread":0.23153730723147525,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2112491476","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.07456221,0.000832986,0.7816455,0.0007854522,0.00041734052,0.0006638335,0.054497983,0.07887584,0.0077188127],"genre_scores_gemma":[0.05934261,0.0006425941,0.8372577,0.00025911623,0.00010186892,0.0010483995,0.08777719,0.008014159,0.0055563697],"study_design_codex":"bench_or_experimental","study_design_gemma":"not_applicable","domain_scores_codex":[0.9993291,0.000090236565,0.00007970218,0.00024759822,0.00018667002,0.00006675911],"domain_scores_gemma":[0.99904805,0.00023725293,0.00013801674,0.00021578837,0.00027674288,0.00008413003],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012552429,0.0013569703,0.0010772886,0.0013652081,0.0012409319,0.001417599,0.0013415323,0.00047318696,0.010857754],"category_scores_gemma":[0.0025085195,0.00071595935,0.0009406978,0.001998946,0.0003408731,0.0009607808,0.001017283,0.0016322738,0.01074004],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0018745812,0.00030615478,0.009438693,0.0024325699,0.00026164012,0.0007770505,0.001042124,0.012914445,0.66585934,0.007731067,0.10100094,0.19636144],"study_design_scores_gemma":[0.00024516968,0.0003221306,0.018098142,0.00021202053,0.0001896128,0.0011201572,0.00023852798,0.11804799,0.6131237,0.014261504,0.23396282,0.00017821087],"about_ca_topic_score_codex":0.0029052082,"about_ca_topic_score_gemma":0.0027951805,"teacher_disagreement_score":0.010857754,"about_ca_system_score_codex":0.00065088476,"about_ca_system_score_gemma":0.0011489388,"threshold_uncertainty_score":0.036322832},"labels":[],"label_agreement":null},{"id":"W2112492259","doi":"10.1093/bioinformatics/btu172","title":"HyperModules: identifying clinically and phenotypically significant network modules with disease mutations for biomarker discovery","year":2014,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":32,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"National Center for Research Resources; National Institute of General Medical Sciences; National Institutes of Health","keywords":"Personalized medicine; Biomarker discovery; Computational biology; Biomarker; Genomics; Disease; Phenotype; Computer science; Bioinformatics; Biology; Gene; Genetics; Medicine; Genome; Proteomics","score_opus":0.016452039156111462,"score_gpt":0.2520578626700897,"score_spread":0.23560582351397827,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2112492259","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.15329923,0.002753963,0.51046234,0.0027568329,0.0006668209,0.0013095771,0.19930205,0.11410551,0.0153436605],"genre_scores_gemma":[0.42588657,0.0018842083,0.3838562,0.00062425475,0.00036469454,0.0023719599,0.16604765,0.008578989,0.010385443],"study_design_codex":"not_applicable","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99946445,0.00012493173,0.000028184515,0.00019631776,0.00013962737,0.000046403136],"domain_scores_gemma":[0.9984029,0.0008809684,0.00021842147,0.00015102024,0.00016434336,0.00018238332],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015553945,0.0019323004,0.0008351018,0.0044769472,0.00066972506,0.0015786962,0.0014546411,0.0006525848,0.03253939],"category_scores_gemma":[0.0062815268,0.0006273985,0.0016819623,0.0016659271,0.00036724893,0.0015334206,0.002016766,0.0008488141,0.0054318947],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00258294,0.0005159688,0.08588734,0.0053565,0.0029901597,0.0013640239,0.0008382944,0.058291923,0.05967586,0.024149394,0.4094469,0.3489007],"study_design_scores_gemma":[0.0010868974,0.0007289106,0.0583352,0.000905785,0.0017238947,0.0024228995,0.0003730437,0.6035686,0.06363265,0.133253,0.1335985,0.00037061344],"about_ca_topic_score_codex":0.0015450771,"about_ca_topic_score_gemma":0.003249713,"teacher_disagreement_score":0.03253939,"about_ca_system_score_codex":0.0005651841,"about_ca_system_score_gemma":0.0013926094,"threshold_uncertainty_score":0.10885513},"labels":[],"label_agreement":null},{"id":"W2112861424","doi":"10.1093/bioinformatics/btt744","title":"ALEA: a toolbox for allele-specific epigenomics analysis","year":2013,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Epigenetics and DNA Methylation","field":"Biochemistry, Genetics and Molecular Biology","cited_by":23,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canada's Michael Smith Genome Sciences Centre; Simon Fraser University; University of British Columbia; BC Cancer Agency","funders":"Natural Sciences and Engineering Research Council of Canada; Canadian Institutes of Health Research; Genome British Columbia; Dermatology Foundation","keywords":"Epigenomics; Pipeline (software); Genome browser; Allele; Computational biology; Identification (biology); Biology; Computer science; Genetics; Genomics; Genome; Gene; DNA methylation; Gene expression; Programming language","score_opus":0.015137580687410666,"score_gpt":0.23971749601868483,"score_spread":0.22457991533127417,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2112861424","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0031440393,0.00060905993,0.6837793,0.00024050784,0.0001766249,0.00019223387,0.0450713,0.26382914,0.0029577964],"genre_scores_gemma":[0.03167486,0.0009598274,0.8406235,0.0007580434,0.00012053185,0.0023426185,0.06799389,0.049688887,0.0058378517],"study_design_codex":"not_applicable","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99890816,0.00026377497,0.0001025538,0.0003195152,0.0003062976,0.00009969313],"domain_scores_gemma":[0.99783,0.0012641021,0.00018136641,0.00035497325,0.00024862937,0.0001209678],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0038042285,0.0023998169,0.0019581618,0.0029203964,0.00088631053,0.0026767752,0.0037240998,0.0012265209,0.046507135],"category_scores_gemma":[0.0070592347,0.0017550422,0.0028130235,0.0017060018,0.00066722324,0.002023136,0.0032018851,0.0032007955,0.023873933],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0019781913,0.00033188597,0.012914833,0.0045646797,0.0021315024,0.0012953712,0.00079126976,0.042996764,0.04875462,0.029043533,0.5025628,0.35263452],"study_design_scores_gemma":[0.00093871314,0.00021314641,0.008374039,0.00040427875,0.00052056066,0.0013441847,0.0001998321,0.40388364,0.042905357,0.12496371,0.41585034,0.00040217297],"about_ca_topic_score_codex":0.0019565187,"about_ca_topic_score_gemma":0.00404318,"teacher_disagreement_score":0.046507135,"about_ca_system_score_codex":0.0006956168,"about_ca_system_score_gemma":0.0020318106,"threshold_uncertainty_score":0.15558189},"labels":[],"label_agreement":null},{"id":"W2112862832","doi":"10.1093/bioinformatics/btl414","title":"SANTA domain: a novel conserved protein module in <i>Eukaryota</i> with potential involvement in chromatin regulation","year":2006,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":35,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ontario Genomics; University of Ottawa","funders":"","keywords":"Chromatin; Biology; Chromatin remodeling; Protein domain; Gene; Genome; Genetics; Conserved sequence; Computational biology; Lineage (genetic); Gene duplication; B3 domain; Phylogenetic tree; ChIA-PET; Transcription factor; DNA-binding domain; Peptide sequence","score_opus":0.004771413853950824,"score_gpt":0.18169389461563537,"score_spread":0.17692248076168454,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2112862832","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98435193,0.002094443,0.0090222275,0.0001133985,0.00003289696,0.000034230074,0.0010618414,0.0003474267,0.002941558],"genre_scores_gemma":[0.9871229,0.00047727648,0.0076145856,0.00008153434,0.000022440276,0.000023696706,0.0025964791,0.000018201064,0.002042772],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99996746,0.0000027100987,0.0000022379957,0.000012535409,0.000008102682,0.0000068845243],"domain_scores_gemma":[0.99994195,0.000006825773,0.000019411189,0.0000057181674,0.0000073577353,0.000018696153],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000050486506,0.00026533147,0.00022163837,0.00025279506,0.00023447741,0.00024260057,0.00017837044,0.00028101198,0.0009040393],"category_scores_gemma":[0.00010872685,0.00010463075,0.00029177868,0.000250397,0.00024191478,0.0002217225,0.00019020899,0.00018909741,0.00056912354],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002253424,0.000018509954,0.0039364877,0.00015699049,0.000016288013,0.00055896846,0.000047550802,0.00026105595,0.9831619,0.0005719196,0.00034108062,0.0107038915],"study_design_scores_gemma":[0.00008422523,0.00068828656,0.24089356,0.00007867443,0.00017959942,0.014681183,0.00018469185,0.009409444,0.66978806,0.0018862423,0.062065452,0.000060623046],"about_ca_topic_score_codex":0.0003289215,"about_ca_topic_score_gemma":0.0003537912,"teacher_disagreement_score":0.0009040393,"about_ca_system_score_codex":0.00020210631,"about_ca_system_score_gemma":0.00016678822,"threshold_uncertainty_score":0.0030243397},"labels":[],"label_agreement":null},{"id":"W2112893969","doi":"10.1093/bioinformatics/btr303","title":"Sensitive and fast mapping of di-base encoded reads","year":2011,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":23,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"Natural Sciences and Engineering Research Council of Canada; National Institutes of Health; Michael Smith Health Research BC; National Human Genome Research Institute; Howard Hughes Medical Institute","keywords":"Computer science; Sequence (biology); Genome; Software; Source code; Code (set theory); Computational biology; Biology; Genetics; Gene; Set (abstract data type)","score_opus":0.021789904438875348,"score_gpt":0.21008830602175843,"score_spread":0.18829840158288308,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2112893969","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.06655125,0.001782217,0.8893343,0.0004036403,0.00040589538,0.0003223341,0.009205214,0.025227634,0.006767437],"genre_scores_gemma":[0.093529604,0.00045628363,0.8847775,0.00039453516,0.00007516854,0.00032529354,0.012008232,0.0019644813,0.0064689824],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9977527,0.00043112302,0.00012980361,0.0007768878,0.000766884,0.00014259716],"domain_scores_gemma":[0.9959948,0.0018435473,0.00036611586,0.0008637138,0.0007519944,0.00017964751],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0023315095,0.0013997069,0.0009408386,0.001395731,0.00066747476,0.0013513763,0.0019809315,0.0014429464,0.007896862],"category_scores_gemma":[0.009490674,0.001063632,0.0007815448,0.0019730614,0.00053931796,0.0019050846,0.0022714399,0.0026344615,0.008895647],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0012510758,0.00014008478,0.004279766,0.002001874,0.00021419882,0.00052880554,0.00046239953,0.008213939,0.6396215,0.00913387,0.0212901,0.31286234],"study_design_scores_gemma":[0.0001828643,0.00032301777,0.012459991,0.00019809866,0.00011993276,0.0030331619,0.00023171635,0.120163925,0.7509462,0.01936377,0.092751406,0.00022582793],"about_ca_topic_score_codex":0.00092151354,"about_ca_topic_score_gemma":0.0014128025,"teacher_disagreement_score":0.007896862,"about_ca_system_score_codex":0.0005449394,"about_ca_system_score_gemma":0.00068134867,"threshold_uncertainty_score":0.026417673},"labels":[],"label_agreement":null},{"id":"W2113127959","doi":"10.1093/bioinformatics/btr461","title":"Reconstructing the architecture of the ancestral amniote genome","year":2011,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genome Rearrangement Algorithms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":37,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"Agence Nationale de la Recherche","keywords":"Amniote; Genome; Biology; Vertebrate; Evolutionary biology; Extant taxon; Phylogenomics; Genomics; Most recent common ancestor; Ancestor; Python (programming language); Context (archaeology); Phylogenetics; Genetics; Computer science; Gene; Paleontology; Clade; Geography","score_opus":0.020882426988871196,"score_gpt":0.20813427935761294,"score_spread":0.18725185236874176,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2113127959","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9047986,0.0006022651,0.08029366,0.00031407268,0.000018314218,0.000035808258,0.0084886085,0.0024106328,0.003038044],"genre_scores_gemma":[0.80018324,0.00062480813,0.1683994,0.00013083406,0.000019638224,0.00007263425,0.029088592,0.00059963815,0.00088128453],"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9998652,0.000018970946,0.000007888425,0.00006302137,0.000027796455,0.00001720029],"domain_scores_gemma":[0.9996213,0.00014954786,0.00006264938,0.000055573386,0.00007140687,0.000039520604],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00034842017,0.0003655123,0.00030748206,0.0018158789,0.0004677198,0.00089117175,0.0005876597,0.000647059,0.0019198158],"category_scores_gemma":[0.0020742044,0.00045342097,0.0006569865,0.0014730596,0.0003355046,0.00062068045,0.00081125577,0.00061534956,0.0008852354],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00073671155,0.000166221,0.32632372,0.0011419285,0.000489672,0.0014233963,0.0019549557,0.21546146,0.31251386,0.013343864,0.0066647944,0.11977942],"study_design_scores_gemma":[0.000117333206,0.00028946262,0.35193133,0.00031397748,0.000522011,0.001982787,0.0012752973,0.52768993,0.053490955,0.033211395,0.029044563,0.00013086357],"about_ca_topic_score_codex":0.0054804618,"about_ca_topic_score_gemma":0.006064324,"teacher_disagreement_score":0.0054804618,"about_ca_system_score_codex":0.00044287086,"about_ca_system_score_gemma":0.0007901761,"threshold_uncertainty_score":0.01089716},"labels":[],"label_agreement":null},{"id":"W2113154626","doi":"10.1093/bioinformatics/btt407","title":"RACER: Rapid and accurate correction of errors in reads","year":2013,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":84,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"Natural Sciences and Engineering Research Council of Canada; Compute Canada","keywords":"Computer science; Software; Variety (cybernetics); Error detection and correction; DNA sequencing; Throughput; Data mining; Artificial intelligence; Algorithm; Programming language; Biology; Operating system","score_opus":0.010741544253993585,"score_gpt":0.22662187815072116,"score_spread":0.21588033389672756,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2113154626","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.024389641,0.0016859175,0.77403176,0.0010881614,0.0007997403,0.00063022884,0.023717243,0.16845156,0.005205665],"genre_scores_gemma":[0.041057132,0.0006326797,0.9015088,0.00083014957,0.0002643931,0.00087176,0.03317867,0.016091269,0.005565115],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.98891675,0.0021427898,0.0009727718,0.0033729628,0.004053249,0.0005414133],"domain_scores_gemma":[0.98583764,0.005959629,0.0023022327,0.0028541787,0.0025780783,0.00046826515],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.011097674,0.0026941628,0.0020025442,0.002334775,0.001815732,0.0031464535,0.005060586,0.0020501728,0.010165387],"category_scores_gemma":[0.027021261,0.001629848,0.0023372353,0.0025729868,0.0013391294,0.0036957383,0.0033216637,0.0044434136,0.017238883],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0027653768,0.00039893322,0.0113133155,0.0036023725,0.00071650505,0.0007934343,0.0008899455,0.016642679,0.38985574,0.013902668,0.21837345,0.34074554],"study_design_scores_gemma":[0.00041312104,0.00048053998,0.0063275727,0.00036573055,0.00035664006,0.0017002439,0.0001227038,0.10483267,0.66541797,0.011567146,0.20789516,0.0005205225],"about_ca_topic_score_codex":0.0014874699,"about_ca_topic_score_gemma":0.001963322,"teacher_disagreement_score":0.011097674,"about_ca_system_score_codex":0.0006816575,"about_ca_system_score_gemma":0.002251467,"threshold_uncertainty_score":0.058690786},"labels":[],"label_agreement":null},{"id":"W2113229776","doi":"10.1093/bioinformatics/btt527","title":"FPSAC: fast phylogenetic scaffolding of ancient contigs","year":2013,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Yersinia bacterium, plague, ectoparasites research","field":"Biochemistry, Genetics and Molecular Biology","cited_by":33,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Pacific Institute for the Mathematical Sciences; Simon Fraser University","funders":"","keywords":"Contig; Genome; Phylogenetic tree; Biology; Ancient DNA; Extant taxon; DNA sequencing; Clade; Whole genome sequencing; Evolutionary biology; Computational biology; Genetics; DNA; Gene","score_opus":0.011547503245307398,"score_gpt":0.25099856947682125,"score_spread":0.23945106623151385,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2113229776","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.06262506,0.0007124714,0.77289796,0.0003287647,0.0003344837,0.0003817604,0.028368806,0.12928657,0.005064106],"genre_scores_gemma":[0.123811275,0.0003049718,0.79125905,0.00014248732,0.00012938566,0.0006423256,0.06976482,0.010463456,0.0034822237],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9986896,0.00022666932,0.000059644535,0.00043066737,0.00047328975,0.00012014403],"domain_scores_gemma":[0.99491334,0.0022478316,0.000360109,0.0011897869,0.0009641916,0.00032466356],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002290381,0.0024191365,0.0015688549,0.003995167,0.0023896089,0.002359352,0.0026613013,0.0013136012,0.018462092],"category_scores_gemma":[0.0103131225,0.0014160507,0.0017257713,0.004784229,0.0009364987,0.0018913292,0.0026606002,0.0023577553,0.008787656],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0038422544,0.0004732573,0.013205909,0.0027015596,0.00075105496,0.0012488912,0.0022845836,0.044071443,0.09233962,0.019755477,0.17819583,0.6411302],"study_design_scores_gemma":[0.00146284,0.0006623451,0.019300565,0.00035787447,0.00029356117,0.001617082,0.00067677815,0.6726624,0.104662,0.051804386,0.1460915,0.00040866336],"about_ca_topic_score_codex":0.006996242,"about_ca_topic_score_gemma":0.007865409,"teacher_disagreement_score":0.018462092,"about_ca_system_score_codex":0.0008223692,"about_ca_system_score_gemma":0.002324968,"threshold_uncertainty_score":0.061761856},"labels":[],"label_agreement":null},{"id":"W2113432929","doi":"10.1093/bioinformatics/bth339","title":"A spline function approach for detecting differentially expressed genes in microarray data analysis","year":2004,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Lunenfeld-Tanenbaum Research Institute; Mount Sinai Hospital","funders":"Networks of Centres of Excellence of Canada; Canadian Institutes of Health Research","keywords":"Bayes' theorem; Parametric statistics; Computer science; Microarray analysis techniques; Statistical hypothesis testing; Covariate; Nonparametric statistics; Bayes factor; Parametric model; Sample size determination; False discovery rate; Inference; Data mining; Bayesian probability; Statistics; Mathematics; Gene; Machine learning; Artificial intelligence; Biology; Genetics; Gene expression","score_opus":0.0336993615275709,"score_gpt":0.2709232451001219,"score_spread":0.23722388357255098,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2113432929","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.003604249,0.00010023066,0.995752,0.00007131957,0.000015708869,0.000022463753,0.000043176682,0.0002818614,0.000109030676],"genre_scores_gemma":[0.1380907,0.0005178827,0.8581854,0.000120327255,0.00014455998,0.00048743695,0.00057406124,0.00018811286,0.0016914618],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9966223,0.0020605018,0.00011437188,0.00032456688,0.0007380415,0.00014020702],"domain_scores_gemma":[0.9920393,0.006134678,0.00045088,0.00044219042,0.0007371706,0.00019576418],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008192647,0.000930579,0.0012940026,0.0027905498,0.0006624458,0.0007782221,0.0018182159,0.0014755854,0.0015482258],"category_scores_gemma":[0.01473042,0.00040760497,0.0015337724,0.00371931,0.001271132,0.000733112,0.0011949525,0.0019569276,0.0007705164],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005507669,0.00022802007,0.011052218,0.00047756944,0.0002274903,0.00040799507,0.00030957718,0.46410486,0.024273086,0.050591543,0.0036551717,0.4441217],"study_design_scores_gemma":[0.000020895415,0.000091879694,0.0016173796,0.000015000006,0.000020462145,0.0001038241,0.000020539683,0.9753905,0.0019959318,0.018979603,0.0017142616,0.000029644298],"about_ca_topic_score_codex":0.0021319978,"about_ca_topic_score_gemma":0.0021088116,"teacher_disagreement_score":0.008192647,"about_ca_system_score_codex":0.0007972505,"about_ca_system_score_gemma":0.0017218277,"threshold_uncertainty_score":0.04332733},"labels":[],"label_agreement":null},{"id":"W2113574907","doi":"10.1093/bioinformatics/btl238","title":"Integrating copy number polymorphisms into array CGH analysis using a robust HMM","year":2006,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomic variations and chromosomal abnormalities","field":"Biochemistry, Genetics and Molecular Biology","cited_by":184,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"Michael Smith Health Research BC; Genome Canada","keywords":"Outlier; Hidden Markov model; Computer science; Comparative genomic hybridization; Segmentation; Pattern recognition (psychology); Copy-number variation; Benchmark (surveying); Artificial intelligence; Chromosome; Biology; Genome; Genetics; Gene","score_opus":0.00988208704385298,"score_gpt":0.22812102365176937,"score_spread":0.21823893660791638,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2113574907","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.007982888,0.00010237453,0.97962534,0.00016807916,0.000040129726,0.000043457712,0.00074708555,0.010936193,0.0003544864],"genre_scores_gemma":[0.12354365,0.00019140857,0.86945313,0.00017768177,0.00007937043,0.00021258972,0.0030271376,0.0013258288,0.0019892498],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9987978,0.000316032,0.00007263324,0.00044091212,0.00030096513,0.000071645874],"domain_scores_gemma":[0.9975642,0.001344996,0.00014126065,0.00058000657,0.00027199634,0.00009761864],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002078816,0.0008907796,0.0009686167,0.0010814416,0.00044082012,0.0010990833,0.0017967956,0.0010301822,0.004751234],"category_scores_gemma":[0.007065157,0.00069161004,0.0011369376,0.0013845447,0.00050215574,0.0009125164,0.0015163356,0.001701738,0.003818738],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00079465174,0.00018900445,0.010525814,0.0003327057,0.00053803885,0.0003576039,0.00027221238,0.45335567,0.061950568,0.006556815,0.011755703,0.45337117],"study_design_scores_gemma":[0.000028029646,0.000039293827,0.0019125862,0.00001320152,0.000051180825,0.000095815994,0.00001989437,0.9748644,0.01211099,0.007706305,0.003117414,0.000040829436],"about_ca_topic_score_codex":0.009444334,"about_ca_topic_score_gemma":0.010344826,"teacher_disagreement_score":0.009444334,"about_ca_system_score_codex":0.0007749936,"about_ca_system_score_gemma":0.0012716703,"threshold_uncertainty_score":0.018778682},"labels":[],"label_agreement":null},{"id":"W2113595151","doi":"10.1093/bioinformatics/btq422","title":"Multiple gene expression profile alignment for microarray time-series data clustering","year":2010,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Time Series Analysis and Forecasting","field":"Computer Science","cited_by":20,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Windsor","funders":"","keywords":"Cluster analysis; Pairwise comparison; Data mining; Computer science; Similarity (geometry); Series (stratigraphy); Expression (computer science); Time series; Data set; Dimension (graph theory); Data point; Similarity measure; Pattern recognition (psychology); Artificial intelligence; Mathematics; Machine learning; Biology","score_opus":0.024812553836522346,"score_gpt":0.2374502952753484,"score_spread":0.21263774143882605,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2113595151","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0064375736,0.00017050527,0.99237716,0.00008425664,0.000030136353,0.000039681163,0.00012389047,0.000452842,0.00028393883],"genre_scores_gemma":[0.093923256,0.00032224128,0.90337205,0.000056861983,0.00010368842,0.0003084771,0.0010198692,0.0001670236,0.0007265931],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99820626,0.0004241528,0.00015428835,0.00050564384,0.0006385975,0.00007096955],"domain_scores_gemma":[0.9979563,0.00076552515,0.0003651468,0.00032502643,0.00049613504,0.00009199706],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015482339,0.0008528174,0.0010931835,0.0015360657,0.0006953965,0.0009879947,0.0012874746,0.000937082,0.0014065384],"category_scores_gemma":[0.005530954,0.00033164266,0.0011196407,0.004170351,0.00062776636,0.00139523,0.001075489,0.0016467025,0.0012636689],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00037602778,0.00031051165,0.006154831,0.00084594515,0.00025654084,0.0003673451,0.00039115286,0.2184405,0.15442531,0.04158424,0.005068587,0.571779],"study_design_scores_gemma":[0.00002079549,0.00014325493,0.0042405566,0.00002968788,0.0000476519,0.00044232619,0.00009473613,0.8913942,0.059816424,0.034649666,0.009054242,0.000066433335],"about_ca_topic_score_codex":0.00061389036,"about_ca_topic_score_gemma":0.00068244763,"teacher_disagreement_score":0.0015482339,"about_ca_system_score_codex":0.00073132705,"about_ca_system_score_gemma":0.0009701396,"threshold_uncertainty_score":0.00818795},"labels":[],"label_agreement":null},{"id":"W2113768712","doi":"10.1093/bioinformatics/btt591","title":"ORMAN: Optimal resolution of ambiguous RNA-Seq multimappings in the presence of novel isoforms","year":2013,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":18,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University; University of British Columbia","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"RNA-Seq; Focus (optics); Gene isoform; Resolution (logic); Computational biology; Computer science; RNA; Biology; Artificial intelligence; Genetics; Transcriptome; Gene; Physics; Gene expression","score_opus":0.016720026813189442,"score_gpt":0.23036194235701427,"score_spread":0.21364191554382483,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2113768712","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.05066922,0.0019866384,0.93057597,0.0005028565,0.00022175374,0.00028051928,0.00084691355,0.0121698035,0.0027463587],"genre_scores_gemma":[0.09908297,0.0004483672,0.8946631,0.00053177756,0.00012830643,0.0002460512,0.0019582717,0.0013961749,0.0015450409],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99792516,0.0007049301,0.00011998625,0.0007045843,0.00028931652,0.0002559748],"domain_scores_gemma":[0.9957535,0.0026651924,0.00048259358,0.00057786435,0.00034160298,0.00017913357],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0044775466,0.0025016097,0.0019498727,0.0016546218,0.00091840705,0.0020192482,0.0026362436,0.0021124915,0.0043513486],"category_scores_gemma":[0.007320567,0.00111622,0.0021279997,0.0013933325,0.0011283024,0.00251505,0.0019048569,0.0020332828,0.0013557774],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0022996573,0.0005130867,0.00724959,0.0019059101,0.0005605787,0.0006185337,0.00037330732,0.41754085,0.056062333,0.018529618,0.023307465,0.47103903],"study_design_scores_gemma":[0.000105959094,0.00017090667,0.00072791404,0.000050764153,0.000065489265,0.00017283257,0.000081382874,0.9639714,0.018570377,0.011328667,0.0047106673,0.000043723347],"about_ca_topic_score_codex":0.002797703,"about_ca_topic_score_gemma":0.0048611965,"teacher_disagreement_score":0.0044775466,"about_ca_system_score_codex":0.0014400526,"about_ca_system_score_gemma":0.0028015052,"threshold_uncertainty_score":0.023679793},"labels":[],"label_agreement":null},{"id":"W2113892406","doi":"10.1093/bioinformatics/btm491","title":"<i>CellLine</i>, a stochastic cell lineage simulator","year":2007,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene Regulatory Network Analysis","field":"Biochemistry, Genetics and Molecular Biology","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"","keywords":"Lineage (genetic); Computer science; Cellular differentiation; Cell fate determination; Bistability; Gene regulatory network; Biology; Gene; Biological system; Computational biology; Genetics; Physics; Gene expression","score_opus":0.0066561639445456245,"score_gpt":0.2249542220159751,"score_spread":0.21829805807142946,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2113892406","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.08677465,0.00062145526,0.8286008,0.00071534584,0.00028847676,0.00035417086,0.011904315,0.035123326,0.035617508],"genre_scores_gemma":[0.4466911,0.0007898973,0.5145961,0.0005285939,0.000063949126,0.0013395137,0.013800122,0.005721879,0.016468743],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99983656,0.00004798201,0.0000093659155,0.00002799134,0.00005363715,0.000024337023],"domain_scores_gemma":[0.9991347,0.0004800925,0.000067025256,0.00008097359,0.00016493535,0.000072240146],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005298165,0.00062661024,0.00039636355,0.00034133135,0.0003812961,0.0006415723,0.0018159878,0.0010397453,0.015170244],"category_scores_gemma":[0.0016191701,0.00037416338,0.0005848135,0.00043833882,0.00041487615,0.0006210506,0.0006062591,0.00088107,0.0021867217],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00020699955,0.00011085385,0.0035748056,0.000366036,0.000079697056,0.00022389312,0.0001407203,0.92278284,0.007391278,0.025491059,0.018772485,0.020859247],"study_design_scores_gemma":[0.00005832209,0.0000303284,0.000129279,0.000013391291,0.000009658401,0.000028066259,0.0000054972843,0.9763222,0.0037818768,0.0017968685,0.017814916,0.0000094792895],"about_ca_topic_score_codex":0.0062112412,"about_ca_topic_score_gemma":0.0051468355,"teacher_disagreement_score":0.015170244,"about_ca_system_score_codex":0.0007492008,"about_ca_system_score_gemma":0.00094507274,"threshold_uncertainty_score":0.05074954},"labels":[],"label_agreement":null},{"id":"W2113944588","doi":"10.1093/bioinformatics/btn433","title":"Accurate sequence-based prediction of catalytic residues","year":2008,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Machine Learning in Bioinformatics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":85,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Nankai University","keywords":"Support vector machine; Curse of dimensionality; Sequence (biology); Computer science; Classifier (UML); Protein structure prediction; Feature selection; Artificial intelligence; Biological system; Data mining; Pattern recognition (psychology); Protein structure; Chemistry; Biology; Biochemistry","score_opus":0.028211413451712894,"score_gpt":0.26240301861543913,"score_spread":0.23419160516372622,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2113944588","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.48616365,0.008682016,0.48902553,0.00054752035,0.00018174836,0.000113732836,0.004077362,0.006864098,0.0043443562],"genre_scores_gemma":[0.8359466,0.0015785492,0.15180731,0.00016075745,0.00012112168,0.00006311066,0.008609102,0.00018727464,0.0015262858],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99914,0.00015812043,0.000081556514,0.00025181402,0.00031457882,0.000053853924],"domain_scores_gemma":[0.9973908,0.0010250177,0.000513437,0.0002741356,0.000697553,0.00009906089],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012643997,0.0009282973,0.00077398686,0.0012791546,0.00026635977,0.0008248955,0.00068768865,0.000984329,0.0012703103],"category_scores_gemma":[0.003775498,0.0002646203,0.0004586986,0.00083389634,0.0002762785,0.0011419088,0.0004813195,0.00093751645,0.0019098205],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008899393,0.00043211534,0.09294408,0.001790129,0.00025405013,0.00082336075,0.000091346956,0.18708448,0.22267623,0.0020635999,0.011974923,0.47897574],"study_design_scores_gemma":[0.00004775807,0.00022760233,0.022253044,0.00008687928,0.00008470315,0.0008967205,0.000043873235,0.8575967,0.10834159,0.003860252,0.0065178513,0.000042920965],"about_ca_topic_score_codex":0.0010015278,"about_ca_topic_score_gemma":0.0009817609,"teacher_disagreement_score":0.0012791546,"about_ca_system_score_codex":0.0003408814,"about_ca_system_score_gemma":0.0005245066,"threshold_uncertainty_score":0.0066868067},"labels":[],"label_agreement":null},{"id":"W2114013334","doi":"10.1093/bioinformatics/btr368","title":"SpEED: fast computation of sensitive spaced seeds","year":2011,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":62,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Toronto Metropolitan University; Western University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Computer science; Software; Computation; Speedup; Similarity (geometry); Source code; Sequence (biology); Code (set theory); Algorithm; Data mining; Artificial intelligence; Parallel computing; Programming language; Biology; Image (mathematics)","score_opus":0.019925179716295455,"score_gpt":0.22735844778708,"score_spread":0.20743326807078455,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2114013334","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.030664176,0.00036643763,0.87909037,0.0002509495,0.00022153587,0.00017700031,0.0018813332,0.08003893,0.007309262],"genre_scores_gemma":[0.22545186,0.00018860353,0.7494394,0.00020344982,0.00010936437,0.0004341235,0.0056706383,0.012530935,0.005971565],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9985862,0.00022086648,0.00011497513,0.00036689098,0.0005758397,0.00013533255],"domain_scores_gemma":[0.99572444,0.001849538,0.0002959801,0.0007172114,0.0011836946,0.00022908307],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013899589,0.0015997088,0.0011653756,0.0017472968,0.0008852177,0.0019006152,0.0026320645,0.0010866506,0.021279009],"category_scores_gemma":[0.011450247,0.00084748946,0.0010505004,0.0017577684,0.0007598965,0.002952665,0.0018581154,0.0014527093,0.008416888],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0026690161,0.00034045565,0.008138852,0.0012796222,0.00026844433,0.0004807634,0.00062344945,0.16373354,0.05315154,0.06680567,0.16055152,0.5419572],"study_design_scores_gemma":[0.00027030706,0.00016215772,0.000851959,0.00005129099,0.00004947843,0.00022356305,0.00006131244,0.8836356,0.047465805,0.04348666,0.02365203,0.00008991167],"about_ca_topic_score_codex":0.002396072,"about_ca_topic_score_gemma":0.0028099888,"teacher_disagreement_score":0.021279009,"about_ca_system_score_codex":0.0010125595,"about_ca_system_score_gemma":0.0019334843,"threshold_uncertainty_score":0.07118535},"labels":[],"label_agreement":null},{"id":"W2114143739","doi":"10.1093/bioinformatics/18.3.484","title":"Assembly of fingerprint contigs: parallelized FPC","year":2002,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Biometric Identification and Security","field":"Computer Science","cited_by":18,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"BC Cancer Agency","funders":"National Human Genome Research Institute; BC Cancer Agency","keywords":"Contig; Fingerprint (computing); Computer science; File Transfer Protocol; Artificial intelligence; Operating system; Biology; Genome; Genetics; The Internet","score_opus":0.039415668603740885,"score_gpt":0.24664642156911182,"score_spread":0.20723075296537094,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2114143739","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.03802522,0.0007373009,0.84115934,0.000360692,0.0002657135,0.0006558712,0.01236822,0.10021187,0.006215749],"genre_scores_gemma":[0.059157908,0.00034856878,0.87417346,0.00021346215,0.00007830361,0.0012299014,0.047578886,0.010908404,0.006311151],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99879706,0.00013656922,0.000097591284,0.0004964141,0.0003063953,0.00016595708],"domain_scores_gemma":[0.99794334,0.00059637526,0.00013274862,0.0005663057,0.0006007306,0.00016046583],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013004916,0.0022080054,0.0014248284,0.0018287958,0.001033788,0.0017864812,0.0024525765,0.00073354144,0.01656463],"category_scores_gemma":[0.0048259115,0.0011215018,0.0013266166,0.0030957842,0.0005311506,0.0010230663,0.0018011994,0.0024190352,0.015653528],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0039951834,0.00049703295,0.0046632783,0.0020906401,0.00041191964,0.00078580214,0.0010617586,0.025191262,0.25902435,0.0065800534,0.11331426,0.5823845],"study_design_scores_gemma":[0.00094089453,0.00053060614,0.009309564,0.00017164958,0.00029972097,0.0016578081,0.0002810084,0.22913758,0.48059568,0.022845153,0.25383335,0.00039694426],"about_ca_topic_score_codex":0.005658129,"about_ca_topic_score_gemma":0.003805987,"teacher_disagreement_score":0.01656463,"about_ca_system_score_codex":0.0012076072,"about_ca_system_score_gemma":0.001312702,"threshold_uncertainty_score":0.05541426},"labels":[],"label_agreement":null},{"id":"W2115480698","doi":"10.1093/bioinformatics/btn176","title":"A robust framework for detecting structural variations in a genome","year":2008,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":60,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of Toronto","funders":"National Institute of General Medical Sciences; Natural Sciences and Engineering Research Council of Canada; National Institutes of Health","keywords":"Genome; Structural variation; Computational biology; Biology; Identification (biology); Genetics; Source code; Reference genome; 1000 Genomes Project; Population; Human genome; Probabilistic logic; Computer science; Gene; Single-nucleotide polymorphism; Genotype; Artificial intelligence","score_opus":0.0291062581705459,"score_gpt":0.24351418172296851,"score_spread":0.2144079235524226,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2115480698","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0028118857,0.000218891,0.99532324,0.00012888621,0.000014275814,0.00003936652,0.00028104498,0.00094398,0.00023838652],"genre_scores_gemma":[0.13449086,0.00048069382,0.8584603,0.0004219823,0.00021690299,0.0004298421,0.003277506,0.00041155954,0.0018102681],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99479973,0.0011882546,0.00025805866,0.0020716374,0.0013404242,0.00034187973],"domain_scores_gemma":[0.9938186,0.0033392946,0.0007245779,0.0009146953,0.0009336972,0.000269166],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0065979147,0.0019137694,0.0024465607,0.0049267425,0.001169918,0.002783681,0.005300474,0.0031502543,0.0030387307],"category_scores_gemma":[0.0147782825,0.0011339649,0.0028031322,0.003084906,0.002406297,0.00262534,0.0030184344,0.002469635,0.0012635184],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00029891104,0.00014739833,0.0031309675,0.00028576335,0.0002783682,0.00039962647,0.00011890529,0.73645747,0.007890372,0.06242512,0.006366976,0.18220016],"study_design_scores_gemma":[0.000030114345,0.000055249526,0.00034719807,0.000016418606,0.00003308442,0.00012444939,0.000011197786,0.9490222,0.0010740529,0.047734253,0.0015285,0.000023337703],"about_ca_topic_score_codex":0.009249922,"about_ca_topic_score_gemma":0.0072041485,"teacher_disagreement_score":0.009249922,"about_ca_system_score_codex":0.001862383,"about_ca_system_score_gemma":0.002746709,"threshold_uncertainty_score":0.034893572},"labels":[],"label_agreement":null},{"id":"W2116062363","doi":"10.1093/bioinformatics/btp011","title":"SCARF: maximizing next-generation EST assemblies for evolutionary and population genomic analyses","year":2009,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":21,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Contig; License; Sequence assembly; Sequence (biology); Computer science; MIT License; Software; Genomics; Computational biology; Population; Biology; Programming language; Genome; Genetics; Operating system; Gene","score_opus":0.06561212450917692,"score_gpt":0.297059179803663,"score_spread":0.23144705529448606,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2116062363","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.019427106,0.00046818855,0.93048686,0.00024065208,0.00010954364,0.00025543172,0.0071580363,0.038768105,0.0030860186],"genre_scores_gemma":[0.06512257,0.00025589467,0.91193885,0.00011965323,0.00008438712,0.0007679752,0.013269232,0.006435401,0.0020060183],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9985769,0.0004961167,0.00009897253,0.00037231442,0.00034801636,0.00010768026],"domain_scores_gemma":[0.9977537,0.0013106233,0.00020078894,0.00031837742,0.0003056738,0.000110868845],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0052157273,0.0018932951,0.0015788843,0.0023858391,0.0010088,0.00143207,0.0023059028,0.0011402072,0.017148886],"category_scores_gemma":[0.012584687,0.0011433591,0.0016059176,0.001811923,0.00042725535,0.0018528317,0.0018536866,0.0014479706,0.009906993],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011837139,0.00027208016,0.010921609,0.0018325024,0.00055016135,0.00080847583,0.00064857624,0.06124709,0.09109012,0.014440729,0.13604705,0.680958],"study_design_scores_gemma":[0.00070121186,0.00037522,0.008826559,0.00027260228,0.0002993592,0.0014650939,0.00014736972,0.7418235,0.11308641,0.038124338,0.09464207,0.00023631191],"about_ca_topic_score_codex":0.0009347702,"about_ca_topic_score_gemma":0.0017786912,"teacher_disagreement_score":0.017148886,"about_ca_system_score_codex":0.000732407,"about_ca_system_score_gemma":0.0009849037,"threshold_uncertainty_score":0.057368755},"labels":[],"label_agreement":null},{"id":"W2116063068","doi":"10.1093/bioinformatics/btl379","title":"General framework for developing and evaluating database scoring algorithms using the TANDEM search engine","year":2006,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Advanced Proteomics Techniques and Applications","field":"Chemistry","cited_by":218,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"National Cancer Institute; U.S. Public Health Service","keywords":"Computer science; Database search engine; Database; Function (biology); Software; Suite; Information retrieval; Matching (statistics); Source code; Data mining; Open source; Search engine; Programming language; Mathematics","score_opus":0.07749868424509176,"score_gpt":0.3742599619674528,"score_spread":0.29676127772236105,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2116063068","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00040269483,0.000113732494,0.99157363,0.0000755265,0.000027941473,0.00045116243,0.00024291108,0.0064986385,0.00061368325],"genre_scores_gemma":[0.006672664,0.00026401083,0.98954517,0.00009671856,0.000042544052,0.0009775357,0.0010344171,0.00062057254,0.0007463073],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99094933,0.0023189364,0.0016566615,0.00096900377,0.003633905,0.00047211695],"domain_scores_gemma":[0.9920678,0.0025588602,0.00041198288,0.0013469897,0.0032988633,0.00031541736],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.016304573,0.0042277277,0.0023870696,0.005702578,0.0015881365,0.0069316956,0.01036084,0.0035504317,0.012694205],"category_scores_gemma":[0.028665503,0.0022693824,0.0036629983,0.0037400795,0.0015215744,0.0046630064,0.004279596,0.0040907455,0.011408289],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00054946955,0.0007464089,0.0042652474,0.0023495099,0.0006719183,0.0009243089,0.00054798025,0.13151199,0.023350736,0.22835751,0.060468353,0.54625666],"study_design_scores_gemma":[0.00035907497,0.00039691737,0.0010735765,0.00057077163,0.00020223601,0.0010245144,0.00014701315,0.7371512,0.023555417,0.13228446,0.102973,0.00026181273],"about_ca_topic_score_codex":0.008396735,"about_ca_topic_score_gemma":0.0052040378,"teacher_disagreement_score":0.016304573,"about_ca_system_score_codex":0.0017622833,"about_ca_system_score_gemma":0.0052544856,"threshold_uncertainty_score":0.086227834},"labels":[],"label_agreement":null},{"id":"W2116887561","doi":"10.1093/bioinformatics/btl553","title":"Dependence network modeling for biomarker identification","year":2006,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":28,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Biomarker discovery; Biomarker; Identification (biology); Computational biology; Consistency (knowledge bases); Cancer; Cancer biomarkers; Microarray analysis techniques; Computer science; Microarray; Relevance (law); Data mining; Bioinformatics; Gene; Biology; Proteomics; Artificial intelligence; Genetics; Gene expression","score_opus":0.022975766169512985,"score_gpt":0.26833696639057386,"score_spread":0.24536120022106087,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2116887561","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.015286762,0.00042110845,0.9810765,0.00054941233,0.000028645476,0.0000705743,0.0006602659,0.00031136462,0.0015953785],"genre_scores_gemma":[0.77508366,0.0017704688,0.21314937,0.0004440244,0.00021837302,0.0007742188,0.0031934683,0.00016499986,0.0052013947],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9984927,0.0007554147,0.00006263047,0.00034022398,0.00025009457,0.00009897004],"domain_scores_gemma":[0.9930507,0.005222115,0.0005913542,0.0003793927,0.0006054497,0.00015113704],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0028822154,0.0011864252,0.0010761793,0.0030684697,0.0004969005,0.0010015654,0.0015199294,0.0010915753,0.0027328753],"category_scores_gemma":[0.013497356,0.00051683804,0.0011396761,0.00227264,0.0009869448,0.0022325208,0.0009835651,0.001269564,0.0006134108],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00012809668,0.00007395773,0.0058201225,0.00013503332,0.000140673,0.00014135617,0.0000790885,0.86251235,0.0012594865,0.0820209,0.0022172737,0.045471556],"study_design_scores_gemma":[0.0000036534839,0.000009523182,0.00031865892,0.000008196134,0.000009925101,0.000023525188,0.0000043027217,0.9500656,0.00017852946,0.04885141,0.0005199479,0.0000066311563],"about_ca_topic_score_codex":0.0057434216,"about_ca_topic_score_gemma":0.003700626,"teacher_disagreement_score":0.0057434216,"about_ca_system_score_codex":0.0015918078,"about_ca_system_score_gemma":0.0010244674,"threshold_uncertainty_score":0.015242755},"labels":[],"label_agreement":null},{"id":"W2117368100","doi":"10.1093/bioinformatics/btm538","title":"GEIGER: investigating evolutionary radiations","year":2007,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Evolution and Paleontology Studies","field":"Earth and Planetary Sciences","cited_by":2550,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University; University of British Columbia","funders":"","keywords":"Geiger counter; Computer science; Software; R package; Software package; Theoretical computer science; Programming language; Physics; Nuclear physics","score_opus":0.026116039978048948,"score_gpt":0.2387691718893873,"score_spread":0.21265313191133833,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2117368100","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.04308348,0.0062783086,0.74010366,0.0030745456,0.0009251061,0.0002465697,0.10308901,0.07044161,0.03275763],"genre_scores_gemma":[0.27423432,0.0036265615,0.5862485,0.001199292,0.00044190232,0.001631461,0.079099014,0.03974891,0.013770003],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9976826,0.0009859222,0.00013755457,0.00071294175,0.00035749358,0.00012336957],"domain_scores_gemma":[0.9899704,0.0076340134,0.0005462202,0.001044768,0.00046583658,0.00033889394],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0046232445,0.0015276719,0.0019486507,0.003934299,0.0009803431,0.0030340466,0.0027764034,0.0009794761,0.039403245],"category_scores_gemma":[0.024071649,0.00082839647,0.0027319877,0.005227723,0.0008422281,0.0031709038,0.0035538631,0.0018364385,0.012674744],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006179842,0.000075399126,0.04845784,0.0044739633,0.0037669644,0.0011915718,0.0027054357,0.054771326,0.011713493,0.095056154,0.46484235,0.31232753],"study_design_scores_gemma":[0.00034851054,0.00031716388,0.055288013,0.00077479065,0.002226094,0.0025891052,0.0006432634,0.14617303,0.010339533,0.23136222,0.5494975,0.0004407086],"about_ca_topic_score_codex":0.0031089436,"about_ca_topic_score_gemma":0.002384169,"teacher_disagreement_score":0.039403245,"about_ca_system_score_codex":0.00050282775,"about_ca_system_score_gemma":0.0010307943,"threshold_uncertainty_score":0.13181692},"labels":[],"label_agreement":null},{"id":"W2117428399","doi":"10.1093/bioinformatics/btu665","title":"ExomeAI: detection of recurrent allelic imbalance in tumors using whole-exome sequencing data","year":2014,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Rare Diseases","field":"Biochemistry, Genetics and Molecular Biology","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"McGill University and Génome Québec Innovation Centre","funders":"Canadian Institutes of Health Research; Diamond Blackfan Anemia Foundation","keywords":"Exome sequencing; Allele; Exome; Genetics; Computational biology; Computer science; Biology; Gene; Mutation","score_opus":0.032638080909792394,"score_gpt":0.26518746949311117,"score_spread":0.23254938858331878,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2117428399","genre_codex":"dataset","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.11141179,0.001494397,0.32962966,0.0024585559,0.00044667284,0.0012889734,0.47483784,0.0644711,0.013961047],"genre_scores_gemma":[0.21505654,0.0010757373,0.3903717,0.0013711308,0.00028818657,0.0027896953,0.36643165,0.01154813,0.011067199],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9974825,0.00039930467,0.0003080207,0.000996936,0.00064734166,0.0001659987],"domain_scores_gemma":[0.9928883,0.0038152325,0.0010378367,0.0010198404,0.0007242166,0.0005146403],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0057075336,0.0011418614,0.0011478355,0.004909229,0.0008472202,0.0023118935,0.00153963,0.00095726567,0.027077906],"category_scores_gemma":[0.016831929,0.00067390146,0.0012370899,0.003924171,0.0004450729,0.0009293818,0.0028140296,0.0017260531,0.007476798],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.002305951,0.00022649349,0.12810773,0.0035863607,0.0024418351,0.0037998338,0.00095338863,0.009244938,0.09018844,0.007598957,0.4846422,0.2669038],"study_design_scores_gemma":[0.0012272744,0.0005342675,0.3514375,0.0007529394,0.0013879164,0.008256793,0.00050045346,0.0836217,0.119519025,0.031055134,0.40118092,0.00052608975],"about_ca_topic_score_codex":0.0029838402,"about_ca_topic_score_gemma":0.008201532,"teacher_disagreement_score":0.027077906,"about_ca_system_score_codex":0.0006993468,"about_ca_system_score_gemma":0.0017694287,"threshold_uncertainty_score":0.090584636},"labels":[],"label_agreement":null},{"id":"W2117628101","doi":"10.1093/bioinformatics/btu807","title":"flowCL: ontology-based cell population labelling in flow cytometry","year":2014,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Single-cell and spatial transcriptomics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":21,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"BC Cancer Agency","funders":"National Institute of Biomedical Imaging and Bioengineering; National Institute of Allergy and Infectious Diseases; National Institute of General Medical Sciences","keywords":"Labelling; Flow cytometry; Computer science; Ontology; Population; Biology; Molecular biology; Medicine; Biochemistry","score_opus":0.009206576639586,"score_gpt":0.2153353101133954,"score_spread":0.20612873347380942,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2117628101","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0023010436,0.0002747066,0.78271985,0.00039707654,0.0001774783,0.00073446066,0.02519054,0.18489538,0.0033095],"genre_scores_gemma":[0.023696361,0.0008279973,0.84501094,0.001030199,0.00011660292,0.0048494264,0.08707927,0.03247074,0.0049184966],"study_design_codex":"not_applicable","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99695444,0.0004754602,0.00046719043,0.00085890846,0.0010235873,0.00022038574],"domain_scores_gemma":[0.99395365,0.0030560393,0.0006582864,0.0011433021,0.0008897913,0.00029886147],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007830508,0.0029413956,0.0016736065,0.0063781617,0.0022227238,0.0046748016,0.004195989,0.0024278054,0.02089469],"category_scores_gemma":[0.011449427,0.002104984,0.0036141498,0.0038566291,0.0019296152,0.0046685026,0.0047323676,0.0036155444,0.0133441575],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0015329509,0.00051330385,0.013521076,0.0089910785,0.0010729478,0.001377262,0.003022,0.02396336,0.08961491,0.07628686,0.4105208,0.36958337],"study_design_scores_gemma":[0.0004679053,0.00023918842,0.008786505,0.0012843207,0.00038895622,0.0012383623,0.00043468073,0.15832819,0.067257434,0.12556003,0.63553256,0.00048187943],"about_ca_topic_score_codex":0.0072012264,"about_ca_topic_score_gemma":0.007400723,"teacher_disagreement_score":0.02089469,"about_ca_system_score_codex":0.0034244577,"about_ca_system_score_gemma":0.0059932666,"threshold_uncertainty_score":0.06989968},"labels":[],"label_agreement":null},{"id":"W2117658386","doi":"10.1093/bioinformatics/bth010","title":"The Hera database and its use in the characterization of endoplasmic reticulum proteins","year":2004,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Cellular transport and secretion","field":"Biochemistry, Genetics and Molecular Biology","cited_by":50,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"Canadian Institutes of Health Research; Genome Canada","keywords":"Endoplasmic reticulum; HERA; Transmembrane protein; Computational biology; Transmembrane domain; ER retention; Biology; Membrane protein; Chemistry; Cell biology; Biochemistry; Receptor; Gene; Membrane; Physics","score_opus":0.011676975472815937,"score_gpt":0.2055065723436951,"score_spread":0.19382959687087917,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2117658386","genre_codex":"empirical","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.34008408,0.016826568,0.24757546,0.00171004,0.00026995662,0.0012974638,0.28592685,0.08130475,0.025004894],"genre_scores_gemma":[0.19456418,0.0047856295,0.34709978,0.00034317956,0.0001580692,0.0008816292,0.44598854,0.0024332767,0.003745781],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9987657,0.00030724623,0.00031002797,0.0002578142,0.0002839661,0.00007527001],"domain_scores_gemma":[0.9955716,0.0017651874,0.0006079819,0.000746513,0.0008194246,0.0004892338],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002740121,0.0010476311,0.0011684329,0.008408062,0.0008321877,0.0018057661,0.0013386761,0.0007948967,0.0050329296],"category_scores_gemma":[0.0070468257,0.00038416075,0.0005533656,0.0075584073,0.00033272037,0.002470365,0.0014402398,0.0007299007,0.0044893245],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.008774282,0.0013406124,0.0403608,0.007755634,0.0006557851,0.004570743,0.0010044881,0.01682111,0.10787364,0.023913424,0.1156956,0.67123395],"study_design_scores_gemma":[0.0011277698,0.0011212875,0.06411815,0.001141254,0.0006343179,0.0104621,0.0008389454,0.18055314,0.10196957,0.026185406,0.6114467,0.00040137922],"about_ca_topic_score_codex":0.0010501357,"about_ca_topic_score_gemma":0.0007369315,"teacher_disagreement_score":0.008408062,"about_ca_system_score_codex":0.00041004154,"about_ca_system_score_gemma":0.0011856051,"threshold_uncertainty_score":0.016836822},"labels":[],"label_agreement":null},{"id":"W2117706019","doi":"10.1093/bioinformatics/btu793","title":"Adaptive settings for the nearest-neighbor particle tracking algorithm","year":2014,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Target Tracking and Data Fusion in Sensor Networks","field":"Computer Science","cited_by":18,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal; Hôpital Maisonneuve-Rosemont","funders":"","keywords":"k-nearest neighbors algorithm; Computer science; Tracking (education); Algorithm; Set (abstract data type); Series (stratigraphy); Data set; License; Particle filter; Data mining; Artificial intelligence","score_opus":0.022137062114337126,"score_gpt":0.24040703049046752,"score_spread":0.21826996837613039,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2117706019","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0071503883,0.00030231552,0.9896701,0.0002449482,0.00010502659,0.00015940894,0.00006945953,0.00034939722,0.001948944],"genre_scores_gemma":[0.3819131,0.00038346785,0.61189383,0.00036767792,0.00021962088,0.0010499312,0.00049450365,0.00022682156,0.003451066],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99591476,0.0018880556,0.0002513672,0.0007886124,0.00093106425,0.00022616156],"domain_scores_gemma":[0.9897864,0.006521136,0.0005898259,0.0009953603,0.0017449651,0.00036229606],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0059692366,0.0012542371,0.0017031676,0.0012264175,0.0011317029,0.0015330155,0.005127359,0.0032898898,0.004912454],"category_scores_gemma":[0.032728065,0.00079109083,0.0009195177,0.0010958465,0.0014486621,0.0033211617,0.0036866986,0.0032814357,0.0016415453],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0004112282,0.00018741644,0.0018090351,0.00025982232,0.00010502387,0.00020945772,0.00019860774,0.816482,0.0017615046,0.06067539,0.005332436,0.11256806],"study_design_scores_gemma":[0.000048012156,0.000033353703,0.000110575966,0.00001561272,0.000007196808,0.000057015466,0.0000107044525,0.9843991,0.00035616543,0.014037077,0.00091021915,0.000014932554],"about_ca_topic_score_codex":0.004727817,"about_ca_topic_score_gemma":0.0033538118,"teacher_disagreement_score":0.0059692366,"about_ca_system_score_codex":0.001372817,"about_ca_system_score_gemma":0.0019116475,"threshold_uncertainty_score":0.031568766},"labels":[],"label_agreement":null},{"id":"W2117716708","doi":"10.1093/bioinformatics/btq152","title":"Detection and characterization of novel sequence insertions using paired-end next-generation sequencing","year":2010,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomic variations and chromosomal abnormalities","field":"Biochemistry, Genetics and Molecular Biology","cited_by":130,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University; BC Cancer Agency","funders":"Natural Sciences and Engineering Research Council of Canada; Genome British Columbia; National Institutes of Health; National Human Genome Research Institute; Michael Smith Health Research BC; Howard Hughes Medical Institute","keywords":"Structural variation; Sequence (biology); Genome; Reference genome; Computational biology; Pipeline (software); Biology; DNA sequencing; Whole genome sequencing; Genetics; Sequence assembly; Genomics; Alignment-free sequence analysis; Sequence analysis; Sequence alignment; Computer science; Gene; Peptide sequence","score_opus":0.05710006682181176,"score_gpt":0.23859978664029263,"score_spread":0.18149971981848087,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2117716708","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.33672032,0.00090070826,0.6510309,0.0004449204,0.000070524104,0.0002662603,0.004798098,0.0041211033,0.0016471762],"genre_scores_gemma":[0.33799288,0.0003531793,0.6548412,0.00026217385,0.00005248194,0.00013170588,0.00546627,0.00016898922,0.0007311047],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9990927,0.00020377422,0.0000693178,0.0003264272,0.00026610776,0.000041703497],"domain_scores_gemma":[0.9975477,0.0013861408,0.00042206666,0.00020673311,0.00030285958,0.00013445658],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015589768,0.00066401716,0.0007874198,0.0018294654,0.00042267606,0.0007612355,0.0011532492,0.0010851315,0.0015778856],"category_scores_gemma":[0.0045607723,0.00034950208,0.00067314913,0.0012573556,0.0004990349,0.0008420869,0.0006328914,0.0008990844,0.0008705765],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0016188715,0.0007728976,0.098005496,0.0012788448,0.00045366577,0.0024495288,0.00039674251,0.10814069,0.48485976,0.007612231,0.004141895,0.2902695],"study_design_scores_gemma":[0.00014599756,0.0005003978,0.03467758,0.00007619694,0.0002595948,0.0031522694,0.00014355655,0.64869565,0.2896584,0.01546722,0.0070718382,0.00015124523],"about_ca_topic_score_codex":0.0011746332,"about_ca_topic_score_gemma":0.0022908093,"teacher_disagreement_score":0.0018294654,"about_ca_system_score_codex":0.00044952426,"about_ca_system_score_gemma":0.00064320484,"threshold_uncertainty_score":0.008244753},"labels":[],"label_agreement":null},{"id":"W2118070789","doi":"10.1093/bioinformatics/btl612","title":"Flexible empirical Bayes models for differential gene expression","year":2006,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":65,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Computer science; Bayes' theorem; Inference; Bayesian probability; Bayesian inference; Bayes factor; Data mining; Sensitivity (control systems); Hierarchical database model; Artificial intelligence; Machine learning","score_opus":0.029594441859634186,"score_gpt":0.2868024409490328,"score_spread":0.2572079990893986,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2118070789","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0038184994,0.0006921613,0.9932935,0.00038794993,0.000032279877,0.00008001471,0.00040520605,0.00021379339,0.0010766217],"genre_scores_gemma":[0.35224038,0.0032076377,0.6271363,0.0010362585,0.0005329513,0.0020822117,0.0031097615,0.00036925456,0.010285209],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9939084,0.0033188716,0.00024071867,0.0013290801,0.0009520204,0.00025083666],"domain_scores_gemma":[0.96301,0.032510344,0.0014681004,0.0013647474,0.0013834338,0.0002633402],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.018808536,0.0017864931,0.002783117,0.0023687424,0.00071817485,0.002601575,0.004995866,0.0026651642,0.007317272],"category_scores_gemma":[0.04913344,0.0012363855,0.0019037277,0.0034952578,0.002898179,0.0029266581,0.0016953141,0.0033104834,0.0018495867],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00019020592,0.00008682438,0.0044685393,0.00042379464,0.00020159455,0.000331837,0.00044758007,0.45657918,0.0009342757,0.44895563,0.0051830527,0.0821974],"study_design_scores_gemma":[0.00004284711,0.00001900823,0.0006267933,0.00005463811,0.000035677775,0.00011017673,0.00002253618,0.6651111,0.0002212004,0.3315631,0.0021660416,0.000026976186],"about_ca_topic_score_codex":0.004835487,"about_ca_topic_score_gemma":0.003904647,"teacher_disagreement_score":0.018808536,"about_ca_system_score_codex":0.0024369403,"about_ca_system_score_gemma":0.0013920288,"threshold_uncertainty_score":0.0994702},"labels":[],"label_agreement":null},{"id":"W2118085282","doi":"10.1093/bioinformatics/btl126","title":"HTS-Corrector: software for the statistical analysis and correction of experimental high-throughput screening data","year":2006,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Computational Drug Discovery Methods","field":"Computer Science","cited_by":46,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University and Génome Québec Innovation Centre; Université du Québec à Montréal","funders":"","keywords":"Computer science; Software; Throughput; Error detection and correction; Data mining; Raw data; Reliability engineering; Computer engineering; Algorithm; Operating system; Programming language","score_opus":0.03918408295329441,"score_gpt":0.3240491938539534,"score_spread":0.284865110900659,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2118085282","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0042297095,0.00033818142,0.77675956,0.00024455728,0.00020870748,0.0003775562,0.014337363,0.20201051,0.0014939068],"genre_scores_gemma":[0.042456094,0.00066544925,0.88807255,0.00043873454,0.0002455238,0.002277583,0.031405445,0.028468745,0.005969986],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9946143,0.0015472946,0.0005712991,0.0006009949,0.0024134277,0.00025273755],"domain_scores_gemma":[0.98567224,0.007884783,0.0014347675,0.0017477571,0.003008086,0.0002523496],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0069152904,0.0027390996,0.0023150644,0.0031764777,0.0009615458,0.0020075906,0.00402415,0.0011184957,0.03185917],"category_scores_gemma":[0.019671075,0.0013117103,0.0016952197,0.0032163959,0.0009359804,0.0020393285,0.0016491493,0.0033868013,0.01593354],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0027326553,0.00034589457,0.0084707355,0.0036556264,0.0009670159,0.00083024317,0.00049821264,0.037174594,0.049315993,0.014744066,0.44886297,0.432402],"study_design_scores_gemma":[0.00095381774,0.0006621872,0.010056778,0.00038322114,0.00045547404,0.0014976851,0.00012527699,0.48288184,0.25430787,0.02465598,0.22347453,0.0005454787],"about_ca_topic_score_codex":0.0024909629,"about_ca_topic_score_gemma":0.0023962315,"teacher_disagreement_score":0.03185917,"about_ca_system_score_codex":0.0008420195,"about_ca_system_score_gemma":0.0034102467,"threshold_uncertainty_score":0.10657948},"labels":[],"label_agreement":null},{"id":"W2118087641","doi":"10.1093/bioinformatics/btl530","title":"sGAL: a computational method for finding surface exposed sites in proteins suitable for Cys-mediated cross-linking","year":2006,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Protein Structure and Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"Canadian Institutes of Health Research","keywords":"Atom (system on chip); Surface (topology); Steric effects; Computer science; Protein Data Bank (RCSB PDB); Crystallography; Chemistry; Algorithm; Physics; Mathematics; Stereochemistry; Geometry","score_opus":0.0159941549849161,"score_gpt":0.2967423365172292,"score_spread":0.2807481815323131,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2118087641","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.03340505,0.00038787202,0.9253918,0.0003325858,0.00011789187,0.00023735638,0.0022294028,0.033493035,0.004405075],"genre_scores_gemma":[0.14859855,0.00041208608,0.83644974,0.00024886042,0.00007286589,0.0013803781,0.0050247917,0.0046420633,0.0031706532],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9997366,0.0000766262,0.000013424273,0.00005749238,0.00008466312,0.00003119889],"domain_scores_gemma":[0.99899215,0.000647945,0.0000603736,0.000087140266,0.00012658851,0.00008584166],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012801979,0.0017693916,0.001512134,0.0014038595,0.001239962,0.0014758293,0.004153612,0.0017844569,0.018308843],"category_scores_gemma":[0.00244953,0.0009187615,0.0016897531,0.0011953976,0.0008790943,0.001650457,0.0017321035,0.0014399703,0.0035134845],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0012600108,0.00039023333,0.005593362,0.0018023208,0.00044429657,0.00066519517,0.0004614075,0.63567793,0.012350385,0.034072645,0.047651246,0.25963098],"study_design_scores_gemma":[0.00021219961,0.000044257904,0.00015837036,0.000024281137,0.00002858699,0.000048355352,0.000030359673,0.98501104,0.0020611123,0.007174326,0.0051927096,0.000014430122],"about_ca_topic_score_codex":0.0037512542,"about_ca_topic_score_gemma":0.00608649,"teacher_disagreement_score":0.018308843,"about_ca_system_score_codex":0.00097259437,"about_ca_system_score_gemma":0.0021290977,"threshold_uncertainty_score":0.061249197},"labels":[],"label_agreement":null},{"id":"W2118290743","doi":"10.1093/bioinformatics/btp366","title":"Automated protein (re)sequencing with MS/MS and a homologous database yields almost full coverage and accuracy","year":2009,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Advanced Proteomics Techniques and Applications","field":"Chemistry","cited_by":40,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Bioinformatics Solutions (Canada); University of Waterloo; Western University","funders":"","keywords":"Protein sequencing; Sequence database; Sequence (biology); Proteomics; Protein methods; Computational biology; DNA sequencing; Tandem mass spectrometry; Database; Peptide sequence; Computer science; Sequence analysis; Mass spectrometry; Biology; Genetics; DNA; Chemistry; Gene; Chromatography","score_opus":0.014643112685008496,"score_gpt":0.26237445132770787,"score_spread":0.24773133864269936,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2118290743","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.33028963,0.0021848758,0.62673825,0.0009162183,0.00019070052,0.00033311677,0.0081123775,0.020974219,0.010260665],"genre_scores_gemma":[0.25086492,0.00053182436,0.7296687,0.00045969285,0.00006964033,0.00023978864,0.012099761,0.0016603553,0.0044053416],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9970259,0.0006809147,0.00027476513,0.000860271,0.0010407711,0.00011736602],"domain_scores_gemma":[0.9948619,0.0015337359,0.00072144286,0.0015181291,0.0012275609,0.00013716488],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0032923596,0.0009401812,0.0014139769,0.0019246497,0.00044200377,0.0010488824,0.0011842224,0.0010839315,0.0044212043],"category_scores_gemma":[0.006689619,0.00046664316,0.00065053185,0.0014416738,0.0005396221,0.0012493677,0.0009430409,0.00065765396,0.0066093514],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00063829497,0.00020423255,0.0054560527,0.0004892976,0.00013438244,0.0003496844,0.000109219116,0.003506913,0.80209005,0.0010732622,0.00835232,0.17759633],"study_design_scores_gemma":[0.00016931904,0.0003315302,0.013922009,0.0000951932,0.00018982193,0.0026976063,0.00009577398,0.073528506,0.8689689,0.003394691,0.036511444,0.00009520011],"about_ca_topic_score_codex":0.00088314177,"about_ca_topic_score_gemma":0.0011827743,"teacher_disagreement_score":0.0044212043,"about_ca_system_score_codex":0.00040625018,"about_ca_system_score_gemma":0.0006825191,"threshold_uncertainty_score":0.017411888},"labels":[],"label_agreement":null},{"id":"W2118379383","doi":"10.1093/bioinformatics/btq204","title":"Estimating genome-wide IBD sharing from SNP data via an efficient hidden Markov model of LD with application to gene mapping","year":2010,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":27,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Azrieli Foundation","keywords":"Linkage disequilibrium; Identity by descent; Locus (genetics); Genome-wide association study; Computer science; Genetic association; Multifactor dimensionality reduction; SNP; Genetics; Data mining; Computational biology; Biology; Single-nucleotide polymorphism; Allele; Haplotype; Gene; Genotype","score_opus":0.024204406816163513,"score_gpt":0.26412662976389323,"score_spread":0.23992222294772972,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2118379383","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.022173941,0.00009466049,0.9761713,0.00011192775,0.0000099573335,0.00003616041,0.00016280437,0.0009718421,0.00026745876],"genre_scores_gemma":[0.3093876,0.00015765005,0.6873738,0.00010672825,0.000036741654,0.00024951107,0.0010768527,0.0001617916,0.0014493732],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99947697,0.00021168147,0.000029374698,0.00013263854,0.000090948306,0.000058367164],"domain_scores_gemma":[0.99559,0.0036451386,0.00021160397,0.00025365417,0.00020818971,0.000091471695],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0018609212,0.00046113218,0.00084276317,0.000942453,0.00063809217,0.0008799194,0.0015047801,0.00092569133,0.002553136],"category_scores_gemma":[0.008929916,0.00056354346,0.00085888506,0.0011194817,0.0005563059,0.00094167393,0.0010822089,0.0011387803,0.000603739],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00021623613,0.00008850546,0.009346511,0.00009353337,0.000102572514,0.0001647864,0.00015346368,0.84822524,0.0024837554,0.01095971,0.0012540827,0.12691166],"study_design_scores_gemma":[0.000012213967,0.000009649712,0.0005817106,0.000004827699,0.000006865673,0.000026074136,0.0000060488956,0.9915575,0.00036518235,0.0072612893,0.00016174621,0.0000069194493],"about_ca_topic_score_codex":0.012727112,"about_ca_topic_score_gemma":0.014118424,"teacher_disagreement_score":0.012727112,"about_ca_system_score_codex":0.00093093765,"about_ca_system_score_gemma":0.0015606588,"threshold_uncertainty_score":0.025306046},"labels":[],"label_agreement":null},{"id":"W2118446779","doi":"10.1093/bioinformatics/btg415","title":"Functional topology in a network of protein interactions","year":2004,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":430,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Computational biology; Computer science; Network analysis; Function (biology); Interaction network; Construct (python library); Mutation; Protein–protein interaction; Network model; Set (abstract data type); Biological network; Biology; Genetics; Artificial intelligence; Gene; Computer network","score_opus":0.010753722658915683,"score_gpt":0.22999407287449528,"score_spread":0.21924035021557958,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2118446779","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.60540223,0.0013631873,0.38328746,0.0011206901,0.000024008265,0.00007942757,0.002356556,0.00074035104,0.005626101],"genre_scores_gemma":[0.9561072,0.00059725693,0.04083288,0.000055723453,0.000019867539,0.00007041793,0.0017310872,0.0000379019,0.0005477352],"study_design_codex":"simulation_or_modeling","study_design_gemma":"observational","domain_scores_codex":[0.99949944,0.0001998096,0.00002982245,0.00012977021,0.00011490482,0.000026332129],"domain_scores_gemma":[0.9965444,0.002625245,0.00040526217,0.00013234816,0.00018371743,0.00010913108],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006128907,0.00044410722,0.00034989396,0.002250802,0.0004987887,0.00092029606,0.0005447247,0.00057751045,0.0016982388],"category_scores_gemma":[0.004961294,0.00022769785,0.0003990685,0.001622444,0.0009784726,0.0015648893,0.0004903215,0.00038645134,0.00025841853],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001474229,0.00006410307,0.02313856,0.00047154864,0.00011434889,0.0005538298,0.00019644479,0.87934107,0.007877604,0.05905986,0.0010590744,0.027976157],"study_design_scores_gemma":[0.0000151630065,0.000048702936,0.006570697,0.000032240016,0.000050199884,0.0004542802,0.00005634917,0.8732211,0.0021739744,0.11501867,0.0023410239,0.00001766725],"about_ca_topic_score_codex":0.0016221551,"about_ca_topic_score_gemma":0.0013151265,"teacher_disagreement_score":0.002250802,"about_ca_system_score_codex":0.00075280137,"about_ca_system_score_gemma":0.00041733246,"threshold_uncertainty_score":0.005681157},"labels":[],"label_agreement":null},{"id":"W2118584916","doi":"10.1093/bioinformatics/btp552","title":"Reproducing the manual annotation of multiple sequence alignments using a SVM classifier","year":2009,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Dalhousie University","funders":"","keywords":"Computer science; Annotation; Classifier (UML); Support vector machine; Software; Artificial intelligence; Data mining; Sequence (biology); Machine learning; Multiple sequence alignment; Pattern recognition (psychology); Sequence alignment; Programming language; Peptide sequence","score_opus":0.040615098603383734,"score_gpt":0.2885442176537646,"score_spread":0.24792911905038084,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2118584916","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.20778021,0.00026491343,0.76577187,0.00023696465,0.00019981083,0.00021383772,0.0011311745,0.022167059,0.002234278],"genre_scores_gemma":[0.4110556,0.000098883575,0.5811105,0.000118272095,0.0000510913,0.00016107954,0.004000096,0.00080278737,0.0026017567],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99751794,0.000635622,0.0002701102,0.0006957261,0.00074416277,0.00013656626],"domain_scores_gemma":[0.9858354,0.0066779037,0.0014608615,0.003046178,0.0027502608,0.00022948497],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0041639064,0.0010425439,0.0009466237,0.0012475187,0.0007558773,0.0010180261,0.0015831423,0.0010865792,0.0030840193],"category_scores_gemma":[0.015856965,0.0003513875,0.0005426606,0.0011936198,0.0003623518,0.001249807,0.0006583171,0.00091198,0.0039705383],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00090834626,0.0004927495,0.0133140795,0.0004434464,0.00007827921,0.0005097008,0.00031001007,0.041388057,0.10483446,0.0010125288,0.0134462,0.8232622],"study_design_scores_gemma":[0.000043977267,0.00020752293,0.010079317,0.00005667261,0.000041033167,0.00061866635,0.00008923943,0.87070274,0.107043184,0.002867403,0.008195063,0.000055197317],"about_ca_topic_score_codex":0.0016079235,"about_ca_topic_score_gemma":0.0016386417,"teacher_disagreement_score":0.0041639064,"about_ca_system_score_codex":0.00045671235,"about_ca_system_score_gemma":0.00076769246,"threshold_uncertainty_score":0.022021115},"labels":[],"label_agreement":null},{"id":"W2118596665","doi":"10.1093/bioinformatics/btk027","title":"ORegAnno: an open access database and curation system for literature-derived promoters, transcription factor binding sites and regulatory variation","year":2006,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":123,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canada's Michael Smith Genome Sciences Centre","funders":"Genome British Columbia; Genome Canada","keywords":"Ensembl; Annotation; License; Database; Transcription factor; dbSNP; Computer science; DNA binding site; Promoter; World Wide Web; Computational biology; Biology; Gene; Bioinformatics; Genetics; Genomics; Genotype","score_opus":0.019126779821666915,"score_gpt":0.2672365525115915,"score_spread":0.2481097726899246,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2118596665","genre_codex":"dataset","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.008966615,0.002161416,0.1726666,0.000983355,0.00063029834,0.0014917923,0.60698247,0.18803838,0.018078985],"genre_scores_gemma":[0.016438087,0.0013833788,0.29824513,0.00041844277,0.00024025264,0.002878544,0.65879667,0.014340798,0.007258732],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9975574,0.0004121445,0.0005740113,0.0006978765,0.0006024865,0.00015610574],"domain_scores_gemma":[0.99237156,0.002615361,0.0013209013,0.001455893,0.001365045,0.0008712513],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.0052794977,0.0019530505,0.0021771663,0.016600618,0.0017879221,0.0035272962,0.0034794076,0.0011422917,0.031007929],"category_scores_gemma":[0.017185131,0.0008631179,0.0011967077,0.012407174,0.00067815813,0.00307618,0.0053021237,0.001441262,0.024924047],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014216206,0.00021240933,0.0059656496,0.009895573,0.00037672845,0.0014827515,0.0013074887,0.0017777117,0.032981984,0.015535273,0.6113657,0.31767717],"study_design_scores_gemma":[0.00027764996,0.00011569229,0.010219165,0.0011352464,0.00020753377,0.0010462763,0.00030196446,0.008571647,0.010928959,0.014656676,0.9522718,0.0002673882],"about_ca_topic_score_codex":0.0029720298,"about_ca_topic_score_gemma":0.0038900985,"teacher_disagreement_score":0.9965206,"about_ca_system_score_codex":0.0009936776,"about_ca_system_score_gemma":0.0063596494,"threshold_uncertainty_score":0.10373181},"labels":[],"label_agreement":null},{"id":"W2118651376","doi":"10.1093/bioinformatics/bth240","title":"A practical and robust sequence search strategy for structural genomics target selection","year":2004,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genome Rearrangement Algorithms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":14,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"Medical Research Council","keywords":"Prioritization; Genome; Selection (genetic algorithm); Computer science; Sequence (biology); Structural genomics; Computational biology; Genomics; Software; Sequence alignment; Similarity (geometry); Data mining; Gene; Biology; Genetics; Artificial intelligence; Protein structure; Peptide sequence; Image (mathematics)","score_opus":0.051150241365374265,"score_gpt":0.31169337020729015,"score_spread":0.2605431288419159,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2118651376","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.009728422,0.00011189978,0.9850348,0.00010772797,0.000021148178,0.00017850593,0.00021895295,0.003838997,0.0007594439],"genre_scores_gemma":[0.060017265,0.00011489997,0.9364727,0.00013855063,0.000020831498,0.00030756806,0.0013254617,0.00047383519,0.0011289025],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9980019,0.00060241064,0.000111210116,0.00040736003,0.00077749544,0.00009967661],"domain_scores_gemma":[0.9977717,0.0009365337,0.00019760079,0.00037359577,0.0006125805,0.00010798899],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0028958349,0.001224778,0.0011554406,0.0014357325,0.00063049304,0.0010517972,0.0016029625,0.0011796952,0.005971606],"category_scores_gemma":[0.006867078,0.0005836964,0.00087901304,0.0014982739,0.0006856887,0.0012318605,0.0012058354,0.0015953918,0.0043101055],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008947078,0.00042633957,0.002104715,0.0005679196,0.00017088829,0.0005226497,0.0001997817,0.052721832,0.4517286,0.0153781595,0.010516959,0.46476746],"study_design_scores_gemma":[0.00028287375,0.0011312484,0.0022223608,0.000057345147,0.00013482712,0.0019939626,0.00014461142,0.7484335,0.19811802,0.026704926,0.020638196,0.00013816931],"about_ca_topic_score_codex":0.00071846036,"about_ca_topic_score_gemma":0.0009895831,"teacher_disagreement_score":0.005971606,"about_ca_system_score_codex":0.00053189916,"about_ca_system_score_gemma":0.0009917425,"threshold_uncertainty_score":0.019977033},"labels":[],"label_agreement":null},{"id":"W2118986663","doi":"10.1093/bioinformatics/btu581","title":"Human structural proteome-wide characterization of Cyclosporine A targets","year":2014,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Signaling Pathways in Disease","field":"Biochemistry, Genetics and Molecular Biology","cited_by":43,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Canadian Institutes of Health Research","keywords":"Computational biology; Proteome; Drug discovery; Cyclophilin; Surface plasmon resonance; Human proteome project; Biology; Docking (animal); Cyclophilin A; Biochemistry; Proteomics; Molecular biology; Medicine; Gene; Nanotechnology","score_opus":0.007432989959109125,"score_gpt":0.23173405607790284,"score_spread":0.22430106611879372,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2118986663","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9778942,0.00056295283,0.0045792973,0.00010173979,0.000006020055,0.000022419412,0.015346736,0.0006354969,0.000851085],"genre_scores_gemma":[0.91808724,0.00044349345,0.013938762,0.000048476268,0.000009323567,0.000034095443,0.06696003,0.00007784707,0.00040075183],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9998907,0.00001801844,0.0000072507223,0.00004670028,0.000026441958,0.000010811182],"domain_scores_gemma":[0.9998149,0.000051683695,0.00004183721,0.000022391496,0.000040656814,0.000028491944],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00021189424,0.00042381583,0.00028035848,0.00065959105,0.00025793142,0.00035008346,0.00022585609,0.00025626944,0.0010683215],"category_scores_gemma":[0.0004138604,0.00010714616,0.00036764878,0.0009396461,0.00013664972,0.00016912588,0.0002217261,0.00024621875,0.00045174191],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0031537386,0.0005475977,0.14755192,0.0010302303,0.0004739371,0.0014420293,0.00030969386,0.043896966,0.728089,0.0011311246,0.009605537,0.062768206],"study_design_scores_gemma":[0.0002128837,0.00091200176,0.4272904,0.00005656344,0.00038321203,0.0033183456,0.00036564743,0.26047432,0.2861057,0.001934049,0.018895801,0.00005113659],"about_ca_topic_score_codex":0.0021876676,"about_ca_topic_score_gemma":0.0025105595,"teacher_disagreement_score":0.0021876676,"about_ca_system_score_codex":0.0003441876,"about_ca_system_score_gemma":0.00039894888,"threshold_uncertainty_score":0.0043498874},"labels":[],"label_agreement":null},{"id":"W2119191448","doi":"10.1093/bioinformatics/btr217","title":"Sorting the nuclear proteome","year":2011,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":22,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Dalhousie University","funders":"Australian Research Council; Canadian Institutes of Health Research","keywords":"Proteome; Computational biology; Nuclear protein; Biology; Nuclear transport; Sorting; Compartment (ship); Transcription factor; Computer science; Cell nucleus; Bioinformatics; Gene; Genetics","score_opus":0.012928176794443665,"score_gpt":0.20010404968187503,"score_spread":0.18717587288743137,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2119191448","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7613291,0.002971098,0.22469003,0.0012558772,0.00007254281,0.000080556514,0.004423914,0.0015389149,0.0036380515],"genre_scores_gemma":[0.904414,0.001381451,0.0878422,0.00027145707,0.000035907604,0.00006094526,0.004170827,0.00015174186,0.0016714011],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9995567,0.00007016552,0.00002261635,0.00019729689,0.000113925846,0.000039281185],"domain_scores_gemma":[0.9989496,0.0004095751,0.0002190358,0.00018071703,0.00017367341,0.00006753134],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000993904,0.0005199844,0.0005392719,0.0009198494,0.0006176632,0.0013876092,0.0005074636,0.00070854387,0.002223349],"category_scores_gemma":[0.0026097677,0.00023480471,0.00055207964,0.00093077315,0.0007396512,0.0011875243,0.00063494884,0.0005133728,0.0011993776],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00082350575,0.000108145236,0.06921327,0.0010309999,0.00017901727,0.00033732934,0.00034052308,0.056743547,0.7263063,0.019744335,0.003454545,0.1217184],"study_design_scores_gemma":[0.00008720731,0.00034132364,0.0711731,0.00018000632,0.00017477968,0.0015403182,0.00064457336,0.3760841,0.4649523,0.060068898,0.024631204,0.00012228581],"about_ca_topic_score_codex":0.002514466,"about_ca_topic_score_gemma":0.0023749562,"teacher_disagreement_score":0.002514466,"about_ca_system_score_codex":0.0010420971,"about_ca_system_score_gemma":0.0008436462,"threshold_uncertainty_score":0.0075609684},"labels":[],"label_agreement":null},{"id":"W2119610982","doi":"10.1093/bioinformatics/btn219","title":"quantiNemo: an individual-based program to simulate quantitative traits with explicit genetic architecture in a dynamic metapopulation","year":2008,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Evolution and Genetic Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":117,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"Université de Lausanne; National Science Foundation","keywords":"Executable; Genetic architecture; Selection (genetic algorithm); Quantitative trait locus; Computer science; Population; Metapopulation; Source code; Computer program; Programming language; Artificial intelligence; Biological dispersal","score_opus":0.020357586103040764,"score_gpt":0.29393550746773456,"score_spread":0.2735779213646938,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2119610982","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.16091438,0.00028188014,0.799953,0.00042336274,0.00017127518,0.00027418367,0.004589613,0.020044021,0.01334833],"genre_scores_gemma":[0.5111593,0.00033527322,0.4732515,0.00026496584,0.000055452372,0.001519911,0.0034679607,0.0033835734,0.00656215],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9998971,0.000032148702,0.000005422031,0.000026071733,0.000025857915,0.000013380245],"domain_scores_gemma":[0.9992531,0.0005192643,0.000056249428,0.000046244855,0.00006398805,0.000061145176],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00081770134,0.00054252124,0.0005685542,0.0004016395,0.00053875253,0.0005187416,0.0020337442,0.0008612662,0.009079282],"category_scores_gemma":[0.0018890926,0.00042953447,0.0006461433,0.0004292052,0.00050834013,0.0005555101,0.0008978786,0.0008176141,0.00068583176],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00017551506,0.00007637237,0.005617759,0.00022647623,0.00009867912,0.00013167677,0.00019909411,0.9487473,0.0039307605,0.014087438,0.0061891605,0.020519843],"study_design_scores_gemma":[0.000052909178,0.000014804715,0.00030753313,0.000009555826,0.000011221609,0.000021139636,0.0000069865464,0.99291056,0.000717801,0.00258511,0.0033546458,0.00000773723],"about_ca_topic_score_codex":0.0062034316,"about_ca_topic_score_gemma":0.0050806515,"teacher_disagreement_score":0.009079282,"about_ca_system_score_codex":0.000712466,"about_ca_system_score_gemma":0.001019993,"threshold_uncertainty_score":0.030373216},"labels":[],"label_agreement":null},{"id":"W2120173546","doi":"10.1093/bioinformatics/btl506","title":"Predicted function of the vaccinia virus G5R protein","year":2006,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Poxvirus research and outbreaks","field":"Immunology and Microbiology","cited_by":21,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Victoria","funders":"National Center for Research Resources; Natural Sciences and Engineering Research Council of Canada; National Institute of Allergy and Infectious Diseases; University of California, San Francisco; U.S. Public Health Service; National Institutes of Health","keywords":"Vaccinia; Virology; Virus; Function (biology); Biology; Orthopoxvirus; Computational biology; Computer science; Genetics; Gene","score_opus":0.008804086629035637,"score_gpt":0.21386878367333337,"score_spread":0.20506469704429772,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2120173546","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9583489,0.0024502522,0.028701268,0.00034782023,0.000060537302,0.000084322295,0.005931239,0.0013548338,0.002720875],"genre_scores_gemma":[0.95267445,0.00036264508,0.033538923,0.00007431034,0.000014574107,0.000023428433,0.012092765,0.000078020996,0.001140929],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99986696,0.000026515367,0.000010745254,0.000045140303,0.00002934642,0.000021338272],"domain_scores_gemma":[0.9997906,0.00006914338,0.000038757968,0.000017763383,0.000045920035,0.000037871876],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003303407,0.0007469659,0.00042887108,0.0006012649,0.0002942555,0.0005957925,0.00044445766,0.0007315487,0.0027180535],"category_scores_gemma":[0.00080420007,0.00013845356,0.0006314676,0.0003615289,0.00020834465,0.000284362,0.00013835721,0.00023705687,0.0020985452],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0053781075,0.0005460412,0.1341361,0.001888194,0.00027227422,0.00545373,0.0004681669,0.038475357,0.71388537,0.0056786262,0.008475924,0.085342176],"study_design_scores_gemma":[0.0005210011,0.0026254433,0.12604812,0.0002885152,0.00036017553,0.010653737,0.0006125805,0.39728057,0.418379,0.005548663,0.037564617,0.00011760849],"about_ca_topic_score_codex":0.0019593663,"about_ca_topic_score_gemma":0.0009697483,"teacher_disagreement_score":0.0027180535,"about_ca_system_score_codex":0.0008446584,"about_ca_system_score_gemma":0.0005564741,"threshold_uncertainty_score":0.009092808},"labels":[],"label_agreement":null},{"id":"W2120625059","doi":"10.1093/bioinformatics/btn246","title":"Estimating dynamic models for gene regulation networks","year":2008,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene Regulatory Network Analysis","field":"Biochemistry, Genetics and Molecular Biology","cited_by":53,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"National Institute of General Medical Sciences; Natural Sciences and Engineering Research Council of Canada; National Institutes of Health; National Science Foundation","keywords":"Ode; Gene regulatory network; Ordinary differential equation; Computer science; Smoothing; Systems biology; Biological network; Mathematical optimization; Process (computing); Algorithm; Theoretical computer science; Applied mathematics; Mathematics; Differential equation; Computational biology; Biology; Gene; Gene expression","score_opus":0.013247565527927451,"score_gpt":0.2320814970041585,"score_spread":0.21883393147623106,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2120625059","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.034796126,0.00020624803,0.96386683,0.00018555854,0.000008964108,0.000049390223,0.00020429435,0.0002601939,0.00042240642],"genre_scores_gemma":[0.7427627,0.0009032207,0.24920245,0.0001999125,0.00008443212,0.0007505221,0.0028285603,0.00019721051,0.0030709675],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9978745,0.00096217974,0.000076058044,0.00067258213,0.00030608862,0.00010867686],"domain_scores_gemma":[0.9803892,0.01729013,0.0010102735,0.000530878,0.00060983474,0.00016972817],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004297745,0.0014476929,0.0014588367,0.0019770213,0.00054259907,0.0012445783,0.0017198805,0.0019882198,0.0015972051],"category_scores_gemma":[0.030788768,0.0009587844,0.0013749648,0.0017662512,0.0014339075,0.002098834,0.001573814,0.0018181984,0.00041100543],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000020757943,0.000015079849,0.00097576814,0.00003475168,0.000025415658,0.000028304343,0.000031390016,0.98461086,0.000447628,0.0069263065,0.00015606111,0.006727656],"study_design_scores_gemma":[0.0000036605618,0.000006240213,0.00020204077,0.000004270449,0.000003898469,0.000009355134,0.000004556208,0.9886503,0.0001622553,0.010805545,0.00014243189,0.000005507761],"about_ca_topic_score_codex":0.008678144,"about_ca_topic_score_gemma":0.0052866247,"teacher_disagreement_score":0.008678144,"about_ca_system_score_codex":0.0018747331,"about_ca_system_score_gemma":0.0010792182,"threshold_uncertainty_score":0.02272892},"labels":[],"label_agreement":null},{"id":"W2120688086","doi":"10.1093/bioinformatics/btq262","title":"Fast integration of heterogeneous data sources for predicting gene function with limited annotation","year":2010,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":139,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"Ontario Genomics; Ontario Genomics Institute; Genome Canada","keywords":"Overfitting; Computer science; Function (biology); Data mining; Annotation; Construct (python library); Source code; Artificial intelligence; Gene regulatory network; Machine learning; Association (psychology); Artificial neural network; Gene; Biology; Genetics","score_opus":0.015437980484894882,"score_gpt":0.2300689522546752,"score_spread":0.2146309717697803,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2120688086","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.05476766,0.00090051425,0.9232993,0.0006857569,0.00007925962,0.00019802844,0.0057896045,0.012014389,0.0022655115],"genre_scores_gemma":[0.35537636,0.0009559357,0.61693907,0.00021414505,0.00010436955,0.0004819661,0.023452226,0.0010237854,0.0014521243],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9979532,0.0005021989,0.00011534492,0.0005605851,0.00079411565,0.00007454831],"domain_scores_gemma":[0.9881814,0.006422484,0.0008620023,0.0025891634,0.0016384417,0.0003065001],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003522856,0.0016254886,0.0013708384,0.005477747,0.00092269125,0.001754308,0.002152865,0.0010833812,0.0018688749],"category_scores_gemma":[0.02058071,0.00092873164,0.0010104255,0.006700854,0.0005161527,0.0037769093,0.0029631555,0.0016717836,0.0014851622],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0012431304,0.00045215068,0.03642985,0.0011734682,0.0007872112,0.0011282298,0.00046837976,0.22945514,0.04301911,0.014227136,0.020268453,0.6513477],"study_design_scores_gemma":[0.000054743836,0.00006678026,0.006639164,0.00008619464,0.00017046905,0.00027739847,0.00012440089,0.930553,0.01845847,0.035009068,0.008498976,0.000061357394],"about_ca_topic_score_codex":0.004545279,"about_ca_topic_score_gemma":0.0066952864,"teacher_disagreement_score":0.005477747,"about_ca_system_score_codex":0.000835095,"about_ca_system_score_gemma":0.001042988,"threshold_uncertainty_score":0.018630862},"labels":[],"label_agreement":null},{"id":"W2120817193","doi":"10.1093/bioinformatics/bts273","title":"MicrobeDB: a locally maintainable database of microbial genomic sequences","year":2012,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":22,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"BC Centre for Disease Control; Simon Fraser University; Dalhousie University","funders":"Canadian Institutes of Health Research","keywords":"Upload; Computer science; Parsing; Genome; Bacterial genome size; Key (lock); Database; Automation; Interface (matter); World Wide Web; Biology; Programming language; Genetics; Operating system","score_opus":0.01155662655029321,"score_gpt":0.22623209520254178,"score_spread":0.21467546865224857,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2120817193","genre_codex":"dataset","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.017140403,0.0055721872,0.10111392,0.0011879324,0.00058000314,0.00033474594,0.67537385,0.18196584,0.016731115],"genre_scores_gemma":[0.025491955,0.0014533708,0.07414729,0.00027763873,0.00011352792,0.00039966343,0.8851537,0.008349533,0.004613383],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9992644,0.00009195808,0.00007329803,0.0002486124,0.00022176877,0.00009991861],"domain_scores_gemma":[0.9978709,0.00043864708,0.0002849346,0.00050990557,0.000301232,0.00059441],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016355452,0.0018890431,0.001941145,0.0044184346,0.0012325529,0.0030707112,0.0047834963,0.0015196963,0.039340835],"category_scores_gemma":[0.005647055,0.0011232777,0.0008343446,0.0075479406,0.00055632717,0.0029962813,0.0033807117,0.0021948894,0.04732115],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0051099574,0.00031628058,0.007312518,0.006910322,0.00033898442,0.0007805633,0.00063017104,0.0022609364,0.08455941,0.009831818,0.65283513,0.22911389],"study_design_scores_gemma":[0.0005242307,0.00023170213,0.009287973,0.00060336717,0.0002455297,0.0007566361,0.00022043654,0.0072150026,0.037353545,0.0063940613,0.9369877,0.0001798142],"about_ca_topic_score_codex":0.0027641815,"about_ca_topic_score_gemma":0.0030045,"teacher_disagreement_score":0.039340835,"about_ca_system_score_codex":0.00063097547,"about_ca_system_score_gemma":0.0020121338,"threshold_uncertainty_score":0.13160819},"labels":[],"label_agreement":null},{"id":"W2120905579","doi":"10.1093/bioinformatics/btv458","title":"NRGsuite: a PyMOL plugin to perform docking simulations in real time using FlexAID","year":2015,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Protein Degradation and Inhibitors","field":"Biochemistry, Genetics and Molecular Biology","cited_by":96,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Sherbrooke","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Plug-in; Computer science; Docking (animal); Software; Computer graphics (images); Computational science; Programming language","score_opus":0.03695817191731657,"score_gpt":0.29749589377417157,"score_spread":0.260537721856855,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2120905579","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.021103064,0.0010955943,0.32290444,0.00071112445,0.0006534398,0.00034462288,0.026463045,0.61199176,0.014732921],"genre_scores_gemma":[0.18934508,0.001992561,0.4589681,0.0012136336,0.0001807649,0.0036493216,0.0670734,0.2482658,0.029311234],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99942833,0.00010403832,0.00004284547,0.00010958821,0.00021878327,0.0000964093],"domain_scores_gemma":[0.9990256,0.00042611442,0.00007096065,0.00016818126,0.00017557536,0.0001336367],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0017310055,0.0026301574,0.001734587,0.0010697412,0.0010258167,0.001440074,0.006473268,0.0014054626,0.079453066],"category_scores_gemma":[0.0033014868,0.0015048698,0.0014282723,0.0010425146,0.0006223717,0.0018101213,0.0025604481,0.003413543,0.02299701],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0022552374,0.00036737122,0.0046505737,0.003260912,0.00095983787,0.0011215217,0.0007382019,0.05418907,0.02520133,0.016183807,0.7128321,0.17824015],"study_design_scores_gemma":[0.0009010073,0.00023693925,0.0032851195,0.00028370763,0.00020352835,0.0007245257,0.00012747609,0.5799187,0.05915814,0.012592447,0.34203178,0.0005367366],"about_ca_topic_score_codex":0.005155577,"about_ca_topic_score_gemma":0.005645093,"teacher_disagreement_score":0.079453066,"about_ca_system_score_codex":0.0010682275,"about_ca_system_score_gemma":0.001820016,"threshold_uncertainty_score":0.26579696},"labels":[],"label_agreement":null},{"id":"W2120942948","doi":"10.1093/bioinformatics/btv241","title":"Using kernelized partial canonical correlation analysis to study directly coupled side chains and allostery in small G proteins","year":2015,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Protein Structure and Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Allosteric regulation; Canonical correlation; Partial correlation; Side chain; Inference; Computer science; Correlation; Graph; Computational biology; Biological system; Chemistry; Physics; Theoretical computer science; Biology; Artificial intelligence; Mathematics; Biochemistry; Nuclear magnetic resonance","score_opus":0.03087710946533728,"score_gpt":0.28210783536686584,"score_spread":0.25123072590152856,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2120942948","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.11946528,0.00042383248,0.86894166,0.00034818737,0.00004437078,0.0000838095,0.0013790225,0.007452637,0.0018612935],"genre_scores_gemma":[0.58567476,0.0003969434,0.4072707,0.00016823065,0.00004850484,0.00015901715,0.0029003478,0.0012036436,0.0021778194],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99970776,0.00009824174,0.0000105107065,0.00007477816,0.000073756426,0.0000348946],"domain_scores_gemma":[0.99867487,0.00065513584,0.00013541196,0.00016945618,0.00026688733,0.00009827736],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00097469136,0.0006915656,0.00049650896,0.0011718508,0.00045986602,0.00058127556,0.00065818697,0.00045107774,0.0019537003],"category_scores_gemma":[0.004180616,0.00027495518,0.00068873435,0.0011134307,0.00062681193,0.00080457644,0.0008224491,0.0007721731,0.00067833724],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00035083148,0.00016944784,0.022445237,0.00055203855,0.0003466967,0.0005300692,0.0003827566,0.6683485,0.039248448,0.041395757,0.022662474,0.20356782],"study_design_scores_gemma":[0.000008043859,0.000008746602,0.0011726884,0.0000046717864,0.000010192381,0.000033795444,0.000013730044,0.98891884,0.0023921714,0.0058529754,0.0015702059,0.000014013454],"about_ca_topic_score_codex":0.014819481,"about_ca_topic_score_gemma":0.017940829,"teacher_disagreement_score":0.014819481,"about_ca_system_score_codex":0.00051848334,"about_ca_system_score_gemma":0.0016180766,"threshold_uncertainty_score":0.02946645},"labels":[],"label_agreement":null},{"id":"W2121166244","doi":"10.1093/bioinformatics/btt056","title":"Non-parametric Bayesian approach to post-translational modification refinement of predictions from tandem mass spectrometry","year":2013,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Advanced Proteomics Techniques and Applications","field":"Chemistry","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of Toronto","funders":"Canadian Institutes of Health Research","keywords":"Bayesian probability; Posttranslational modification; Tandem mass spectrometry; Computer science; Parametric statistics; Mass spectrometry; Tandem; Chemistry; Computational biology; Chromatography; Mathematics; Artificial intelligence; Statistics; Biology; Materials science","score_opus":0.014431690440249959,"score_gpt":0.2489121045829948,"score_spread":0.23448041414274484,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2121166244","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.011937841,0.00022312267,0.9857991,0.00020495968,0.000021968097,0.00005400151,0.00013917168,0.0013042382,0.00031562353],"genre_scores_gemma":[0.2917577,0.00035694675,0.7037114,0.00046309427,0.00016077477,0.00036657572,0.0013858959,0.0005798262,0.0012177671],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99676776,0.0013259875,0.00018869164,0.00067304465,0.0008991194,0.00014536447],"domain_scores_gemma":[0.9841733,0.011111822,0.001364154,0.0010840187,0.0019887283,0.0002780594],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007830186,0.0014535264,0.0015536353,0.0016702103,0.00094336754,0.0012774889,0.0042053876,0.00173426,0.001506244],"category_scores_gemma":[0.027107015,0.0011415874,0.0016855147,0.0014122381,0.001307368,0.0019169116,0.0019580566,0.0026788216,0.00070710876],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00023585498,0.00008417989,0.002508843,0.00014135108,0.00014077047,0.000106850326,0.00010885744,0.9193244,0.0037672326,0.0059896694,0.0015774674,0.06601456],"study_design_scores_gemma":[0.000007827768,0.000013610143,0.00018939897,0.0000054106026,0.000007212269,0.000016530237,0.0000046198993,0.9943792,0.001030687,0.0040858476,0.00025038078,0.000009117762],"about_ca_topic_score_codex":0.010403629,"about_ca_topic_score_gemma":0.011164204,"teacher_disagreement_score":0.010403629,"about_ca_system_score_codex":0.0017705593,"about_ca_system_score_gemma":0.0027653293,"threshold_uncertainty_score":0.041410506},"labels":[],"label_agreement":null},{"id":"W2121305702","doi":"10.1093/bioinformatics/btp054","title":"Fast computation of neighbor seeds","year":2009,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Toronto Metropolitan University; Western University","funders":"","keywords":"Computation; Heuristic; Computer science; Exponential function; Current (fluid); Class (philosophy); Algorithm; Exponential growth; Sensitivity (control systems); k-nearest neighbors algorithm; Polynomial; Time complexity; Artificial intelligence; Mathematics","score_opus":0.009562008365173272,"score_gpt":0.233085311769209,"score_spread":0.22352330340403573,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2121305702","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.030812236,0.00048793628,0.962226,0.00015536373,0.00008980211,0.000117167714,0.00026739942,0.002459495,0.0033846502],"genre_scores_gemma":[0.16786869,0.0002393227,0.8283814,0.00007030662,0.000058448324,0.00011805636,0.0010278403,0.00045062584,0.0017853132],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99882954,0.00024714947,0.0000968431,0.00026657962,0.00045775794,0.00010218081],"domain_scores_gemma":[0.9954359,0.0024534771,0.0002979461,0.0007548378,0.0008723832,0.00018539741],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010285121,0.0008528384,0.0012141464,0.0015402597,0.00090277137,0.0012834858,0.0013621326,0.0010665715,0.007374756],"category_scores_gemma":[0.011048221,0.00052945875,0.0006488857,0.0017545468,0.00083478075,0.0026011062,0.001367262,0.00087033055,0.0030302512],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009860361,0.00019047617,0.004484749,0.00080842205,0.000099739904,0.00047481866,0.0004406777,0.21772088,0.038956355,0.072792284,0.01895582,0.64408976],"study_design_scores_gemma":[0.00013127974,0.00018981783,0.0009118731,0.000048655103,0.000038729428,0.00056472817,0.000091516755,0.86419404,0.026644945,0.09638144,0.010756165,0.00004686735],"about_ca_topic_score_codex":0.0017548316,"about_ca_topic_score_gemma":0.0024983254,"teacher_disagreement_score":0.007374756,"about_ca_system_score_codex":0.0007744299,"about_ca_system_score_gemma":0.0012520774,"threshold_uncertainty_score":0.024671018},"labels":[],"label_agreement":null},{"id":"W2121367512","doi":"10.1093/bioinformatics/btt529","title":"Ontology-aware classification of tissue and cell-type signals in gene expression profiles across platforms and technologies","year":2013,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":29,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Human Genome Research Institute; National Institute of General Medical Sciences; Canadian Institute for Advanced Research; Princeton University; National Institutes of Health; National Science Foundation","keywords":"Preprocessor; Source code; Computational biology; Computer science; Gene expression; Gene ontology; Annotation; Gene; Data mining; Biology; Artificial intelligence; Genetics","score_opus":0.017833288188142147,"score_gpt":0.27234277418341923,"score_spread":0.25450948599527706,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2121367512","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.2482033,0.0020221493,0.664402,0.0015991667,0.0003475902,0.0004928524,0.052511226,0.016707113,0.013714665],"genre_scores_gemma":[0.56322193,0.00096407183,0.3614695,0.0005610798,0.00014063898,0.0006160258,0.066755064,0.0019306011,0.0043410114],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99799407,0.0003374156,0.00015155457,0.00066413713,0.00066161924,0.00019126984],"domain_scores_gemma":[0.9952886,0.001902951,0.0006666497,0.00093252066,0.0009948609,0.00021436115],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0026117992,0.00073692907,0.00070603203,0.003592503,0.00095208373,0.0019831492,0.0009335598,0.00059650047,0.002465281],"category_scores_gemma":[0.00771343,0.00031259793,0.0013104591,0.0032880704,0.0005761747,0.0015793572,0.0016799571,0.001160371,0.0015156933],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0012639967,0.00041230416,0.23717263,0.0023707252,0.0011152212,0.00053944055,0.0011043801,0.03754432,0.18338744,0.019394627,0.051750507,0.4639445],"study_design_scores_gemma":[0.00011953568,0.00024955705,0.22182775,0.0003901676,0.00082478527,0.000867231,0.0010387823,0.4788965,0.10800635,0.084441304,0.10312804,0.00021000845],"about_ca_topic_score_codex":0.004291497,"about_ca_topic_score_gemma":0.010621257,"teacher_disagreement_score":0.004291497,"about_ca_system_score_codex":0.0010874677,"about_ca_system_score_gemma":0.0015962776,"threshold_uncertainty_score":0.013812661},"labels":[],"label_agreement":null},{"id":"W2121621283","doi":"10.1093/bioinformatics/btg004","title":"IslandPath: aiding detection of genomic islands in prokaryotes","year":2003,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Salmonella and Campylobacter epidemiology","field":"Agricultural and Biological Sciences","cited_by":395,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"BC Cancer Agency; Simon Fraser University","funders":"","keywords":"Computational biology; Computer science; Biology","score_opus":0.023017004468561517,"score_gpt":0.21639375044336323,"score_spread":0.1933767459748017,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2121621283","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.055242386,0.0014058725,0.5658828,0.0010578152,0.00033427367,0.00049444113,0.030737622,0.33890936,0.005935413],"genre_scores_gemma":[0.12505451,0.00077626924,0.82046115,0.00026940007,0.00012268513,0.00066226255,0.039879724,0.008291608,0.0044823214],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9995178,0.00012405943,0.00003178289,0.00014649815,0.0001401852,0.000039595318],"domain_scores_gemma":[0.9982534,0.0010739943,0.00016187372,0.00018219261,0.00017750937,0.00015101649],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012953996,0.0017047467,0.0010417869,0.0020080418,0.0006668088,0.0012425435,0.0017426569,0.00093017257,0.010354891],"category_scores_gemma":[0.0050515984,0.00064706977,0.000786912,0.0019165804,0.00030900256,0.0014875727,0.00138516,0.0010968398,0.0072312467],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.002707924,0.000364794,0.026651993,0.0036051082,0.00047257825,0.0008387154,0.0009848278,0.008002048,0.109813645,0.0045487955,0.1962157,0.6457939],"study_design_scores_gemma":[0.00096419186,0.0005951552,0.04131516,0.0004740287,0.00054919533,0.0025144557,0.0005115681,0.49600792,0.16980328,0.024893608,0.2619899,0.00038154388],"about_ca_topic_score_codex":0.0015694996,"about_ca_topic_score_gemma":0.0027206852,"teacher_disagreement_score":0.010354891,"about_ca_system_score_codex":0.00028410187,"about_ca_system_score_gemma":0.0006446707,"threshold_uncertainty_score":0.03464061},"labels":[],"label_agreement":null},{"id":"W2121628067","doi":"10.1093/bioinformatics/btt064","title":"Phylogenomic clustering for selecting non-redundant genomes for comparative genomics","year":2013,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":42,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Wilfrid Laurier University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Genomics; Comparative genomics; Cluster analysis; Genome; Computational biology; Computer science; Biology; Evolutionary biology; Genetics; Artificial intelligence; Gene","score_opus":0.026802228151749517,"score_gpt":0.2642351388127584,"score_spread":0.23743291066100888,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2121628067","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.08396255,0.00071322336,0.8778645,0.0005440511,0.00012452455,0.00081717357,0.009827832,0.023703858,0.002442265],"genre_scores_gemma":[0.083623335,0.00015641272,0.9009396,0.0000901431,0.000031436113,0.0005315855,0.011905321,0.0023075896,0.00041454556],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.997535,0.00082364155,0.00020622717,0.0006782382,0.00057669956,0.00018019973],"domain_scores_gemma":[0.99569917,0.001659261,0.00050896034,0.0008170179,0.0010228056,0.00029285043],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0043435716,0.0020680055,0.001742112,0.0075007807,0.002192369,0.002046049,0.002648287,0.0011822234,0.005299388],"category_scores_gemma":[0.012877911,0.0008904397,0.001960575,0.006710153,0.00072643807,0.002005932,0.0018224621,0.0017145217,0.0033321388],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0037250821,0.0015110645,0.056456443,0.0054804296,0.0019141819,0.0009472065,0.0030826803,0.04816973,0.2754182,0.019475587,0.055144027,0.5286754],"study_design_scores_gemma":[0.000517373,0.000546187,0.053634312,0.00046348321,0.00062274304,0.001567704,0.0018748099,0.7286621,0.12803192,0.043898735,0.039813805,0.00036684264],"about_ca_topic_score_codex":0.0021363583,"about_ca_topic_score_gemma":0.0033225825,"teacher_disagreement_score":0.0075007807,"about_ca_system_score_codex":0.0012798267,"about_ca_system_score_gemma":0.0016495498,"threshold_uncertainty_score":0.022971272},"labels":[],"label_agreement":null},{"id":"W2121653135","doi":"10.1093/bioinformatics/btr472","title":"Conserved and differential gene interactions in dynamical biological systems","year":2011,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene Regulatory Network Analysis","field":"Biochemistry, Genetics and Molecular Biology","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"Core Research for Evolutional Science and Technology; National Cancer Institute; U.S. Department of Agriculture; New Mexico State University; National Science Foundation","keywords":"Computational biology; Differential (mechanical device); Gene; Biology; Computer science; Genetics; Physics","score_opus":0.02893718377708309,"score_gpt":0.23776513239720018,"score_spread":0.20882794862011708,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2121653135","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.03933968,0.00032722516,0.9530129,0.0005164359,0.000042260494,0.00009745461,0.0016752394,0.001992716,0.002996142],"genre_scores_gemma":[0.5414862,0.0005348735,0.45099434,0.00017404665,0.00007468198,0.0005436824,0.003354543,0.0006542626,0.0021834546],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99923134,0.00026657703,0.000058673686,0.00020850607,0.00019415007,0.000040725456],"domain_scores_gemma":[0.99623376,0.0028586101,0.00030794743,0.0002657431,0.00023735092,0.000096592434],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014999326,0.0008559516,0.00054942555,0.0008609007,0.00055715605,0.00076399086,0.0010348093,0.0009788456,0.004698263],"category_scores_gemma":[0.0073881308,0.0003917853,0.0010103798,0.00075514684,0.0012422972,0.0011233055,0.0011474005,0.00106021,0.000795005],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00013802024,0.000060902403,0.007114871,0.0004245893,0.00006811334,0.00033467606,0.00023243375,0.83851355,0.008108025,0.118184805,0.0026302189,0.024189673],"study_design_scores_gemma":[0.000014381701,0.000023115492,0.0007732294,0.000019246787,0.000013866142,0.00007407843,0.000015676265,0.9603227,0.0019876743,0.034467228,0.0022747053,0.000014108727],"about_ca_topic_score_codex":0.0032806278,"about_ca_topic_score_gemma":0.0016766527,"teacher_disagreement_score":0.004698263,"about_ca_system_score_codex":0.001014584,"about_ca_system_score_gemma":0.00088849256,"threshold_uncertainty_score":0.015717328},"labels":[],"label_agreement":null},{"id":"W2122062351","doi":"10.1093/bioinformatics/btr389","title":"BRISK—research-oriented storage kit for biology-related data","year":2011,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Biomedical Text Mining and Ontologies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":22,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"Canadian Institutes of Health Research","keywords":"Computer science; Java; Documentation; Data sharing; Data management; Software; Data access; Web application; World Wide Web; Data science; Database; Operating system","score_opus":0.17065333150659484,"score_gpt":0.373670698899958,"score_spread":0.20301736739336318,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2122062351","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.010532199,0.0015856702,0.27345476,0.0025804546,0.0008406073,0.0018654127,0.15422761,0.5300828,0.024830492],"genre_scores_gemma":[0.07235617,0.0018726273,0.47585046,0.0026313437,0.0006158757,0.0044354303,0.3565519,0.05262035,0.033065826],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9962586,0.0006432381,0.00079000276,0.0006444556,0.00136551,0.00029812925],"domain_scores_gemma":[0.9794049,0.0047652507,0.0025381814,0.0070306323,0.0045251446,0.0017357699],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00783029,0.001786459,0.0016122236,0.0052083475,0.0010865725,0.0057658064,0.0051324526,0.0010475068,0.04806086],"category_scores_gemma":[0.01787144,0.0014717164,0.0011162652,0.006812205,0.0010034245,0.007183745,0.005653737,0.0020607589,0.07470085],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0027811965,0.00023536543,0.0052507916,0.00368831,0.00027815625,0.0006383753,0.0013690756,0.0019013591,0.026054226,0.016422411,0.66867745,0.27270323],"study_design_scores_gemma":[0.000796615,0.00040815555,0.00835936,0.0008435685,0.0002519446,0.0010360224,0.0005440875,0.015682325,0.059029933,0.025012923,0.8876004,0.00043466248],"about_ca_topic_score_codex":0.0020084116,"about_ca_topic_score_gemma":0.0014229165,"teacher_disagreement_score":0.04806086,"about_ca_system_score_codex":0.0013588772,"about_ca_system_score_gemma":0.0031262033,"threshold_uncertainty_score":0.1607796},"labels":[],"label_agreement":null},{"id":"W2122270525","doi":"10.1093/bioinformatics/16.5.425","title":"Prediction of genetic structure in eukaryotic DNA using reference point logistic regression and sequence alignment","year":2000,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":19,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Logistic regression; Sequence (biology); Computer science; Point (geometry); Computational biology; DNA; Statistics; Biology; Genetics; Mathematics; Geometry","score_opus":0.02673682026311662,"score_gpt":0.24565773206027303,"score_spread":0.2189209117971564,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2122270525","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.14152798,0.00028530913,0.8489266,0.00017379447,0.000013626506,0.00006313079,0.0006920623,0.00719622,0.001121204],"genre_scores_gemma":[0.40735483,0.00033235655,0.58607477,0.00007058052,0.000025702739,0.00021846766,0.0034586438,0.00061674754,0.0018479703],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9995265,0.00018794618,0.00002473666,0.00013621237,0.00010103203,0.000023556257],"domain_scores_gemma":[0.99782914,0.0015644048,0.00024443396,0.00007135393,0.00023845467,0.000052138672],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013044502,0.0008964638,0.0005333791,0.0011771045,0.0004056641,0.00053425704,0.000898285,0.0005800842,0.0023487078],"category_scores_gemma":[0.006237844,0.0004252955,0.00075687916,0.0013542514,0.00036832548,0.00070989935,0.00053481443,0.00078683207,0.0013296126],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005344192,0.00018509942,0.04393153,0.0002738146,0.0001478336,0.0003446709,0.00014576432,0.75888205,0.019229475,0.0033318931,0.0036138087,0.16937965],"study_design_scores_gemma":[0.000015845158,0.000040710715,0.0022950796,0.0000088476245,0.000010966318,0.00009244378,0.000009254031,0.9913025,0.0044827894,0.0012218609,0.0005098134,0.000009921823],"about_ca_topic_score_codex":0.0037572838,"about_ca_topic_score_gemma":0.002896708,"teacher_disagreement_score":0.0037572838,"about_ca_system_score_codex":0.0004484432,"about_ca_system_score_gemma":0.0010219188,"threshold_uncertainty_score":0.007857263},"labels":[],"label_agreement":null},{"id":"W2122812051","doi":"10.1093/bioinformatics/btq653","title":"HiTEC: accurate error correction in high-throughput sequencing data","year":2010,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":138,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Toronto Metropolitan University; Western University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Computer science; Throughput; DNA sequencing; Source code; Algorithm; Biology; DNA; Genetics","score_opus":0.03351851025080672,"score_gpt":0.2747700612401669,"score_spread":0.2412515509893602,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2122812051","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00943461,0.00072381,0.90066785,0.0003288764,0.00045715648,0.00028775225,0.0047455267,0.08216344,0.0011910029],"genre_scores_gemma":[0.040591303,0.0002583315,0.93820876,0.00038022242,0.00013534937,0.0004669173,0.010930358,0.006167039,0.0028618192],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99220216,0.0017121386,0.0007031832,0.0015041846,0.0034826978,0.00039555994],"domain_scores_gemma":[0.98070174,0.0073605464,0.0019580736,0.0038823306,0.005680249,0.00041706234],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0075054783,0.0025268064,0.001587426,0.0035749837,0.0020887286,0.002641282,0.0038679924,0.0026287006,0.0113555305],"category_scores_gemma":[0.032029394,0.0011613158,0.0014175916,0.004362407,0.0010913876,0.002837161,0.002308971,0.0029334214,0.011284014],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0016187095,0.00027334708,0.013690876,0.0029951439,0.0010819916,0.0013675699,0.0007457035,0.036734115,0.11245038,0.012438978,0.2498811,0.56672204],"study_design_scores_gemma":[0.00031170913,0.00041172767,0.010144375,0.0003778737,0.0002976531,0.0033443435,0.00022959408,0.4431077,0.412083,0.020618016,0.108572796,0.00050117646],"about_ca_topic_score_codex":0.002184624,"about_ca_topic_score_gemma":0.0031340595,"teacher_disagreement_score":0.0113555305,"about_ca_system_score_codex":0.00089169876,"about_ca_system_score_gemma":0.002657119,"threshold_uncertainty_score":0.039693236},"labels":[],"label_agreement":null},{"id":"W2122898583","doi":"10.1093/bioinformatics/btu681","title":"mimicMe: a web server for prediction and analysis of host-like proteins in microbial pathogens","year":2014,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bacteriophages and microbial interactions","field":"Environmental Science","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Host (biology); Proteome; Pathogen; Biology; Web server; Function (biology); Computational biology; Microbiology; Bioinformatics; Computer science; World Wide Web; Genetics; The Internet","score_opus":0.00748936859276984,"score_gpt":0.20999672857677043,"score_spread":0.2025073599840006,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2122898583","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.103645496,0.0040034824,0.24057421,0.0009107787,0.00069025403,0.0007117953,0.24263345,0.39117476,0.015655812],"genre_scores_gemma":[0.18044221,0.0019545075,0.29396144,0.0005821364,0.00025609936,0.001332155,0.48562524,0.018482221,0.017363934],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9995372,0.0000920922,0.000041919357,0.00012261214,0.00015163295,0.00005442002],"domain_scores_gemma":[0.9993318,0.00020097577,0.00008256056,0.00009223393,0.00015110178,0.00014139467],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000891706,0.0017702322,0.0012286181,0.0014036614,0.00085071806,0.0015219961,0.0016922856,0.0011292886,0.019950276],"category_scores_gemma":[0.0026135163,0.00093117583,0.0013768262,0.001287468,0.0002726686,0.0016120571,0.0014659648,0.0010252226,0.016210444],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.007016321,0.00044047518,0.015557634,0.0039498517,0.0008181742,0.0011751584,0.00037343716,0.011814234,0.19269554,0.006376491,0.6124691,0.14731365],"study_design_scores_gemma":[0.0014281186,0.00095957366,0.031785663,0.00043504676,0.00060175045,0.0031802668,0.0002793592,0.26302293,0.1587469,0.016837455,0.5222327,0.00049031875],"about_ca_topic_score_codex":0.0015430966,"about_ca_topic_score_gemma":0.0023986574,"teacher_disagreement_score":0.019950276,"about_ca_system_score_codex":0.00068800885,"about_ca_system_score_gemma":0.001524173,"threshold_uncertainty_score":0.066740274},"labels":[],"label_agreement":null},{"id":"W2123032056","doi":"10.1093/bioinformatics/btu166","title":"relax: the analysis of biomolecular kinetics and thermodynamics using NMR relaxation dispersion data","year":2014,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Protein Structure and Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":62,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université Laval","funders":"H2020 European Research Council","keywords":"Kinetics; Dispersion (optics); Relaxation (psychology); Thermodynamics; Chemistry; Physical chemistry; Statistical physics; Materials science; Physics","score_opus":0.011225486809506306,"score_gpt":0.24577192386449237,"score_spread":0.23454643705498607,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2123032056","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.027948437,0.00089677674,0.9292801,0.0005249408,0.00016517869,0.00015132314,0.007306311,0.026162947,0.0075639],"genre_scores_gemma":[0.24650319,0.0022940568,0.7052696,0.0003256572,0.00034981626,0.00093563244,0.02039883,0.011547378,0.012375905],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99931335,0.00015698615,0.00005531314,0.00019009768,0.00023198858,0.000052364736],"domain_scores_gemma":[0.9986091,0.00058631063,0.00023001248,0.0002500448,0.0002447515,0.000079854435],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015854299,0.0014019723,0.0009076163,0.0010140539,0.0005733502,0.0011990971,0.0022648005,0.00069990894,0.010962358],"category_scores_gemma":[0.0040028687,0.0005980976,0.0015416265,0.0011262456,0.0006736151,0.0013006787,0.001281679,0.0015707937,0.007747905],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009781056,0.00039318224,0.014831063,0.003840233,0.0006941909,0.00057900907,0.000532857,0.27272704,0.16476472,0.0755319,0.07195302,0.39317462],"study_design_scores_gemma":[0.000106861415,0.00019902256,0.0042207255,0.0001358358,0.00015751878,0.00057514576,0.00011754544,0.8104972,0.08853629,0.018476944,0.07677878,0.00019804254],"about_ca_topic_score_codex":0.0022480772,"about_ca_topic_score_gemma":0.0018065142,"teacher_disagreement_score":0.010962358,"about_ca_system_score_codex":0.00047028906,"about_ca_system_score_gemma":0.0011752644,"threshold_uncertainty_score":0.03667277},"labels":[],"label_agreement":null},{"id":"W2123128942","doi":"10.1093/bioinformatics/btr184","title":"Comrad: detection of expressed rearrangements by integrated analysis of RNA-Seq and low coverage genome sequence data","year":2011,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":45,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University; BC Cancer Agency","funders":"Canadian Institutes of Health Research","keywords":"Genome; Computational biology; LNCaP; Biology; RNA-Seq; Gene; Genetics; Gene expression; Prostate cancer; Cancer; Transcriptome","score_opus":0.03551003974175679,"score_gpt":0.24134292339191143,"score_spread":0.20583288365015465,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2123128942","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.009708745,0.00017121571,0.91202885,0.00016457759,0.0000699388,0.00015064195,0.0052796584,0.07117787,0.0012484526],"genre_scores_gemma":[0.07399499,0.00016840592,0.90750957,0.00036014276,0.00004547526,0.00045598677,0.010755467,0.0049930634,0.0017168962],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99719316,0.00046727265,0.00021043762,0.0009664877,0.0009725776,0.00019004817],"domain_scores_gemma":[0.9962704,0.0017899777,0.00051400886,0.00080086506,0.00046196984,0.00016268544],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0045676855,0.0022360256,0.0012924925,0.0022385314,0.0007714785,0.0031644863,0.003628044,0.0015345752,0.008647256],"category_scores_gemma":[0.007931181,0.0010089807,0.0016096415,0.0020425548,0.0013239344,0.002076556,0.0015633323,0.0020759562,0.005407715],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0032983085,0.0005339732,0.018727254,0.0035804354,0.0012094086,0.0011826168,0.0005833423,0.10520005,0.34297353,0.03165004,0.07110454,0.41995645],"study_design_scores_gemma":[0.00018328414,0.000283682,0.004586594,0.00013724169,0.00011158063,0.0006313683,0.0001356535,0.66967964,0.27845678,0.015337657,0.030242292,0.00021426004],"about_ca_topic_score_codex":0.0020640385,"about_ca_topic_score_gemma":0.0041572466,"teacher_disagreement_score":0.008647256,"about_ca_system_score_codex":0.0011907705,"about_ca_system_score_gemma":0.0015448425,"threshold_uncertainty_score":0.028927982},"labels":[],"label_agreement":null},{"id":"W2123271845","doi":"10.1093/bioinformatics/bts555","title":"Phylogenetics, likelihood, evolution and complexity","year":2012,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"National Institutes of Health; National Institute of General Medical Sciences; National Natural Science Foundation of China","keywords":"Phylogenetics; Evolutionary biology; Computational biology; Computer science; Biology; Genetics; Gene","score_opus":0.022273995039399662,"score_gpt":0.23694598210685225,"score_spread":0.21467198706745258,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2123271845","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0077523966,0.006484533,0.9620589,0.003277696,0.0003632124,0.00008520426,0.0035478373,0.0024957296,0.01393456],"genre_scores_gemma":[0.26613978,0.009595601,0.699273,0.0012821201,0.0015231784,0.00081304065,0.010222358,0.0034344494,0.0077165347],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9965553,0.0015439107,0.00025909184,0.0006864629,0.0008486866,0.00010652658],"domain_scores_gemma":[0.9833402,0.013088034,0.0011108967,0.0010647402,0.0010283261,0.00036774407],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0047222357,0.0010948771,0.0013154692,0.0047383406,0.0010934856,0.0042244466,0.0012977219,0.0012214832,0.016964916],"category_scores_gemma":[0.031479865,0.00061581365,0.0012838611,0.005110642,0.0030939525,0.0046498813,0.0026545273,0.00380986,0.003814666],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00013688765,0.00003687491,0.008421622,0.0018434642,0.00047073295,0.00034940755,0.0005391208,0.070994094,0.0019755098,0.60367477,0.06794288,0.24361472],"study_design_scores_gemma":[0.000019596373,0.000017937597,0.0018459017,0.00017975723,0.00005301041,0.00043525346,0.000063937405,0.0970062,0.00077570014,0.8494324,0.050112553,0.000057819118],"about_ca_topic_score_codex":0.0015339857,"about_ca_topic_score_gemma":0.0012462545,"teacher_disagreement_score":0.016964916,"about_ca_system_score_codex":0.0019348982,"about_ca_system_score_gemma":0.0018287688,"threshold_uncertainty_score":0.056753278},"labels":[],"label_agreement":null},{"id":"W2123317029","doi":"10.1093/bioinformatics/bti035","title":"tPatternHunter: gapped, fast and sensitive translated homology search","year":2004,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":28,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University; University of Waterloo; Bioinformatics Solutions (Canada)","funders":"Killam Trusts; Canada Research Chairs","keywords":"Software; Computer science; Homology (biology); Frame (networking); Translation (biology); Information retrieval; Theoretical computer science; Artificial intelligence; Programming language; Biology; Genetics; Amino acid; Gene","score_opus":0.01029322687539442,"score_gpt":0.22948157494675012,"score_spread":0.2191883480713557,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2123317029","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.018611936,0.003035054,0.76514363,0.00078214967,0.0013979347,0.0010459434,0.014580265,0.1903612,0.005041815],"genre_scores_gemma":[0.056834232,0.0014539824,0.86734504,0.00036972584,0.000242074,0.0015633148,0.046037924,0.017430743,0.008722942],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9977203,0.0006039754,0.00022950965,0.00074072706,0.00050827925,0.00019724981],"domain_scores_gemma":[0.99779403,0.0009905379,0.00030849472,0.0003719365,0.00029718335,0.00023780699],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0032221884,0.002937828,0.0024542292,0.002386226,0.0014186815,0.0022195915,0.004269639,0.001995853,0.03141484],"category_scores_gemma":[0.009740601,0.0016230544,0.0016146346,0.0030775664,0.0009799382,0.0023673994,0.0017785304,0.0038462726,0.038393736],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0061501297,0.0004985485,0.002696272,0.005644128,0.00046465578,0.0016064163,0.00068849744,0.006426067,0.19280697,0.008290763,0.24282983,0.5318978],"study_design_scores_gemma":[0.0016588639,0.0017051597,0.0055294354,0.0011574796,0.0005099359,0.006388824,0.00036639412,0.18442601,0.39158928,0.0136627145,0.392383,0.00062290946],"about_ca_topic_score_codex":0.00068050605,"about_ca_topic_score_gemma":0.0012788415,"teacher_disagreement_score":0.03141484,"about_ca_system_score_codex":0.0007260176,"about_ca_system_score_gemma":0.0016664943,"threshold_uncertainty_score":0.10509312},"labels":[],"label_agreement":null},{"id":"W2124012659","doi":"10.1093/bioinformatics/btr378","title":"Biologically inspired EM image alignment and neural reconstruction","year":2011,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Advanced Electron Microscopy Techniques and Applications","field":"Biochemistry, Genetics and Molecular Biology","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Dalhousie University","funders":"Biotechnology and Biological Sciences Research Council","keywords":"Computer science; Artificial intelligence; Pattern recognition (psychology); Computer vision; Dynamic programming; Segmentation; Shortest path problem; Algorithm; Theoretical computer science","score_opus":0.01341737314837253,"score_gpt":0.2679032938747742,"score_spread":0.25448592072640164,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2124012659","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.005367648,0.00008022132,0.99199003,0.00013163657,0.000015594296,0.000019152301,0.00008799831,0.0008080506,0.0014995815],"genre_scores_gemma":[0.08197471,0.00017679567,0.9148855,0.00008028288,0.00001917883,0.00010323854,0.0003435468,0.00038189616,0.0020348679],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99968636,0.00009237609,0.000017072758,0.00007425034,0.000113253685,0.000016723807],"domain_scores_gemma":[0.9993734,0.00030450383,0.00006938643,0.00012159604,0.00010898645,0.000022149712],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008546851,0.00052276364,0.00048140623,0.0003962406,0.00027078777,0.0007127723,0.0010963128,0.0009500545,0.0040189303],"category_scores_gemma":[0.003100149,0.00037069622,0.0005057665,0.0006982946,0.0006983638,0.00089197414,0.0010819173,0.0010249506,0.0018585392],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001596366,0.000058295354,0.000782221,0.00051140477,0.0000686912,0.00019702698,0.00021827478,0.589633,0.062388245,0.081598535,0.004716694,0.25966802],"study_design_scores_gemma":[0.000012918849,0.0000256018,0.00028477932,0.000020212981,0.000006844086,0.00020762264,0.00001842912,0.9489202,0.020577312,0.022644598,0.0072663375,0.000015170647],"about_ca_topic_score_codex":0.00090404047,"about_ca_topic_score_gemma":0.00084559544,"teacher_disagreement_score":0.0040189303,"about_ca_system_score_codex":0.0004901514,"about_ca_system_score_gemma":0.0004380059,"threshold_uncertainty_score":0.013444662},"labels":[],"label_agreement":null},{"id":"W2124018109","doi":"10.1093/bioinformatics/btr241","title":"Predicting site-specific human selective pressure using evolutionary signatures","year":2011,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"McGill University; Université du Québec à Montréal","funders":"","keywords":"Computer science; Computational biology; Artificial intelligence; Biological system; Biology","score_opus":0.015591928864699595,"score_gpt":0.22112838834859713,"score_spread":0.20553645948389754,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2124018109","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8585691,0.0025145619,0.12724327,0.0009831514,0.000040234634,0.000048655413,0.005270856,0.0020006723,0.0033295648],"genre_scores_gemma":[0.95956266,0.00063734956,0.03551113,0.00011979555,0.00004162419,0.000019937364,0.003252365,0.00006749808,0.0007877305],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9997985,0.00005655458,0.000012205233,0.00006188288,0.000051784384,0.000019235398],"domain_scores_gemma":[0.99854726,0.0009372885,0.00022248714,0.0000722802,0.00013258026,0.00008804922],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005374825,0.00026535903,0.00033685853,0.0010844537,0.00019091909,0.0004150775,0.0003677408,0.00042192911,0.0017928521],"category_scores_gemma":[0.002590048,0.00013985341,0.00017598135,0.0010352466,0.00023680252,0.0003980776,0.00029639562,0.0003610756,0.0007736185],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008383715,0.00018999133,0.3818236,0.00076396274,0.00017637103,0.00061366986,0.00018848028,0.13605587,0.06013237,0.0038289614,0.009494564,0.40589377],"study_design_scores_gemma":[0.00004524942,0.00019180588,0.1506651,0.0000840337,0.00008853882,0.0009650591,0.00008047295,0.7960236,0.035435997,0.011665595,0.0047038835,0.000050600403],"about_ca_topic_score_codex":0.0028301454,"about_ca_topic_score_gemma":0.003433865,"teacher_disagreement_score":0.0028301454,"about_ca_system_score_codex":0.00024516534,"about_ca_system_score_gemma":0.00029663113,"threshold_uncertainty_score":0.005997658},"labels":[],"label_agreement":null},{"id":"W2124137325","doi":"10.1093/bioinformatics/btv008","title":"Deep profiling of multitube flow cytometry data","year":2015,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Single-cell and spatial transcriptomics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":12,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Vancouver General Hospital; University of British Columbia; BC Cancer Agency","funders":"National Institute of Biomedical Imaging and Bioengineering; Canadian Institutes of Health Research; National Institutes of Health; Michael Smith Health Research BC","keywords":"Bioconductor; Cluster analysis; Computer science; Profiling (computer programming); Imputation (statistics); Computational biology; Data mining; Flow cytometry; Artificial intelligence; Pattern recognition (psychology); Biological system; Biology; Molecular biology; Machine learning; Genetics; Missing data; Gene","score_opus":0.0601042930517372,"score_gpt":0.2783396199601503,"score_spread":0.21823532690841307,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2124137325","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.48575035,0.00030707434,0.50278085,0.0006531745,0.00004820857,0.0001448657,0.0030493452,0.0054319096,0.0018342376],"genre_scores_gemma":[0.84310734,0.00013643906,0.15090813,0.00029380276,0.000025149666,0.00023331922,0.0033864079,0.000286026,0.0016233479],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99940956,0.00012097768,0.00003302813,0.00016016397,0.00020243203,0.000073847026],"domain_scores_gemma":[0.99848515,0.0005584429,0.00016251105,0.00033021334,0.0003333801,0.00013031147],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014829639,0.0005877729,0.00064588164,0.0010269695,0.00053511147,0.0009839517,0.001052725,0.0009014119,0.002333821],"category_scores_gemma":[0.003096545,0.00021606703,0.00045426417,0.0011406476,0.0006812948,0.00091641414,0.001170501,0.0011980138,0.00051509455],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011252287,0.0005082615,0.04570498,0.00040279157,0.00024886878,0.00049741915,0.00041105296,0.4261503,0.38213333,0.011622107,0.007066021,0.12412974],"study_design_scores_gemma":[0.000025151381,0.00009204363,0.011742549,0.000012891895,0.000016440245,0.00013811694,0.000062129984,0.84554106,0.13076746,0.008645933,0.002913623,0.000042519467],"about_ca_topic_score_codex":0.0026673146,"about_ca_topic_score_gemma":0.0028329056,"teacher_disagreement_score":0.0026673146,"about_ca_system_score_codex":0.0015091863,"about_ca_system_score_gemma":0.00085060345,"threshold_uncertainty_score":0.010949969},"labels":[],"label_agreement":null},{"id":"W2124312647","doi":"10.1093/bioinformatics/bth186","title":"Exploiting the kernel trick to correlate fragment ions for peptide identification via tandem mass spectrometry","year":2004,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Advanced Proteomics Techniques and Applications","field":"Chemistry","cited_by":111,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"","keywords":"Computer science; Kernel (algebra); Tandem mass spectrometry; Identification (biology); Software; Fragment (logic); Correlative; Mass spectrometry; Artificial intelligence; Data mining; Algorithm; Chemistry; Mathematics; Operating system","score_opus":0.016772802314836435,"score_gpt":0.2691784092850094,"score_spread":0.252405606970173,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2124312647","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.043434966,0.00010992604,0.95428944,0.000053859167,0.00001132898,0.00002831763,0.00006535588,0.0017594676,0.00024724234],"genre_scores_gemma":[0.28780067,0.00017649108,0.71063507,0.000041060124,0.000023950477,0.00006625065,0.00034230758,0.00014742417,0.0007668067],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9993723,0.00018060188,0.000044924807,0.00010700716,0.00026365728,0.000031541545],"domain_scores_gemma":[0.99821633,0.0007847203,0.00030232276,0.00036049157,0.00025597724,0.00008007425],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015054209,0.0006432781,0.00071751024,0.0010755355,0.0002781585,0.00054574694,0.0009694951,0.00047930225,0.0012912092],"category_scores_gemma":[0.0037109004,0.0002717009,0.0003708825,0.0012640287,0.00053160114,0.0013278585,0.0012157145,0.0007175811,0.0011009041],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00071886886,0.00038340455,0.008282826,0.0005010121,0.00019671803,0.0007583198,0.000118624215,0.10675967,0.20073777,0.012768061,0.0044157,0.664359],"study_design_scores_gemma":[0.000038770024,0.000098226854,0.001719704,0.000003777112,0.000023918068,0.00049709686,0.000011526525,0.9458808,0.04290476,0.007713146,0.001080921,0.000027469945],"about_ca_topic_score_codex":0.00045266157,"about_ca_topic_score_gemma":0.00054587435,"teacher_disagreement_score":0.0015054209,"about_ca_system_score_codex":0.00028267258,"about_ca_system_score_gemma":0.00070567755,"threshold_uncertainty_score":0.007961512},"labels":[],"label_agreement":null},{"id":"W2124337127","doi":"10.1093/bioinformatics/bts650","title":"HD-CNV: hotspot detector for copy number variants","year":2012,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomic variations and chromosomal abnormalities","field":"Biochemistry, Genetics and Molecular Biology","cited_by":21,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Copy-number variation; Computer science; Copy number analysis; Source code; Biology; Genetics; Genome; Gene; Programming language","score_opus":0.013509275045211431,"score_gpt":0.24548352605813237,"score_spread":0.23197425101292093,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2124337127","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.044219177,0.000571664,0.5692329,0.0005685245,0.00023364398,0.0005257626,0.07546123,0.29728207,0.011905],"genre_scores_gemma":[0.15648417,0.0002339173,0.7397241,0.00047687461,0.0001105045,0.0009626942,0.07365299,0.014298468,0.014056257],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9991035,0.00011498227,0.00006827993,0.00031557807,0.00032529395,0.000072465446],"domain_scores_gemma":[0.99835724,0.0008323884,0.00021349754,0.00021394614,0.00026641067,0.0001165727],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011004905,0.00089072477,0.0007542349,0.0038089934,0.00059778674,0.000875819,0.0015293173,0.00090282765,0.0271674],"category_scores_gemma":[0.0048445472,0.0005811274,0.000609316,0.0021075415,0.0003432573,0.0010871787,0.0013362208,0.0006795038,0.006838083],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0018486411,0.0002574986,0.03716116,0.0010609836,0.00033047618,0.0013295158,0.000505127,0.007034252,0.08124176,0.014110981,0.42202115,0.43309838],"study_design_scores_gemma":[0.0009856548,0.0003326301,0.07413284,0.00016880665,0.00029107125,0.0059009683,0.00025193192,0.2934722,0.23683123,0.05621589,0.33096653,0.0004503009],"about_ca_topic_score_codex":0.0026113642,"about_ca_topic_score_gemma":0.003538969,"teacher_disagreement_score":0.0271674,"about_ca_system_score_codex":0.0005866625,"about_ca_system_score_gemma":0.0007892197,"threshold_uncertainty_score":0.09088403},"labels":[],"label_agreement":null},{"id":"W2124714582","doi":"10.1093/bioinformatics/btr452","title":"OrganismTagger: detection, normalization and grounding of organism entities in biomedical documents","year":2011,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Biomedical Text Mining and Ontologies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":50,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Concordia University; University of New Brunswick","funders":"","keywords":"Computer science; Organism; Information retrieval; Taxonomy (biology); Natural language processing; Precision and recall; Named-entity recognition; Artificial intelligence; Task (project management); Biology","score_opus":0.014569750501862155,"score_gpt":0.2312963566121826,"score_spread":0.21672660611032044,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2124714582","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.11555527,0.007690686,0.50466543,0.0018104544,0.00084349664,0.0011761705,0.08185356,0.2745901,0.0118147135],"genre_scores_gemma":[0.101216905,0.0015020831,0.80899864,0.0004985353,0.00014659272,0.00045683794,0.07945493,0.0025699562,0.005155517],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99719113,0.00038859848,0.00050452666,0.0010622572,0.0007568059,0.00009676625],"domain_scores_gemma":[0.99368095,0.0028428682,0.0011430964,0.0008884022,0.0012541712,0.00019044994],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002935746,0.0013054307,0.00086773164,0.012545528,0.00083969673,0.0028812701,0.0016621384,0.0017453092,0.0043003885],"category_scores_gemma":[0.010837156,0.00073019817,0.00085301726,0.005605407,0.0007813099,0.0037207047,0.0019180411,0.0008375224,0.006540365],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00048616872,0.00029278835,0.029108413,0.0046235407,0.00036060184,0.0014566985,0.0014503995,0.0041622547,0.09488849,0.0047795773,0.08869346,0.76969755],"study_design_scores_gemma":[0.0002519022,0.00064381206,0.08641172,0.0021945303,0.0007810085,0.007306713,0.0017553314,0.111499116,0.33558607,0.015700804,0.4373908,0.0004781778],"about_ca_topic_score_codex":0.0040796353,"about_ca_topic_score_gemma":0.0065145013,"teacher_disagreement_score":0.012545528,"about_ca_system_score_codex":0.0011150924,"about_ca_system_score_gemma":0.0022583953,"threshold_uncertainty_score":0.015525937},"labels":[],"label_agreement":null},{"id":"W2124935821","doi":"10.1093/bioinformatics/bth413","title":"EC_oligos: automated and whole-genome primer design for exons within one or between two genomes","year":2004,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Agriculture and Agri-Food Canada; McGill University; Ste. Anne's Hospital","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Executable; Oligonucleotide; Genome; Primer (cosmetics); Computational biology; Exon; Biology; File Transfer Protocol; Genetics; Computer science; Sequence (biology); DNA; Programming language; Gene; World Wide Web","score_opus":0.041176228304203685,"score_gpt":0.27209719614512895,"score_spread":0.23092096784092525,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2124935821","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.022656983,0.001422595,0.73894584,0.00046403662,0.0004382649,0.0016334167,0.09454129,0.12605412,0.0138433585],"genre_scores_gemma":[0.026385339,0.00053159875,0.8227485,0.0005125748,0.00006491985,0.0019866715,0.12174051,0.014622062,0.011407891],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9988431,0.00019387779,0.00014580584,0.00038595556,0.00028618023,0.00014511058],"domain_scores_gemma":[0.9983992,0.00066780974,0.00019901033,0.00029049264,0.0003139908,0.00012948416],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0018593522,0.0029916512,0.001431201,0.0017875776,0.00091799634,0.0014307711,0.0025445991,0.001248274,0.048562847],"category_scores_gemma":[0.0038219031,0.0024132677,0.001446852,0.0013726504,0.000740113,0.0013404519,0.0014212007,0.002069111,0.031528663],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0029970126,0.00036357803,0.0039962665,0.0036625245,0.00034231084,0.0006172061,0.00047291978,0.0060744807,0.35886008,0.009203331,0.33026594,0.28314447],"study_design_scores_gemma":[0.00074989325,0.00052821747,0.0059970496,0.0003042261,0.00026910356,0.0016193857,0.0001188006,0.033250786,0.56260514,0.00647644,0.38777053,0.00031042742],"about_ca_topic_score_codex":0.0026690036,"about_ca_topic_score_gemma":0.0065062605,"teacher_disagreement_score":0.048562847,"about_ca_system_score_codex":0.0010694885,"about_ca_system_score_gemma":0.0022281373,"threshold_uncertainty_score":0.16245896},"labels":[],"label_agreement":null},{"id":"W2124965915","doi":"10.1093/bioinformatics/btu355","title":"GenomeVISTA—an integrated software package for whole-genome alignment and visualization","year":2014,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":36,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"SickKids Foundation; Hospital for Sick Children; University of Toronto","funders":"National Institute of General Medical Sciences","keywords":"Perl; Computer science; Visualization; Software; Java; Genome; Web application; Source code; Flexibility (engineering); World Wide Web; Data mining; Operating system; Biology; Genetics","score_opus":0.011843090348213053,"score_gpt":0.242854442158818,"score_spread":0.23101135181060495,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2124965915","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0022939108,0.000844172,0.3884386,0.00061505346,0.00061926525,0.00033402347,0.06897676,0.5298181,0.008060102],"genre_scores_gemma":[0.016000835,0.00121603,0.6305701,0.00078125636,0.0002655024,0.0026253448,0.18697444,0.15140074,0.010165783],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9988298,0.00024432325,0.00012391717,0.00028550802,0.00040988112,0.00010662705],"domain_scores_gemma":[0.99810946,0.00066218426,0.00027101242,0.0003176156,0.00044799907,0.00019174034],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0022969958,0.0026741363,0.0020387045,0.00337475,0.001125712,0.0024694437,0.0050795306,0.0015102718,0.071361035],"category_scores_gemma":[0.0060547777,0.0014966426,0.0016977142,0.003680486,0.00058293476,0.0028231416,0.0030075512,0.0042740908,0.05441622],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00080504006,0.000092118775,0.0011924324,0.0022642454,0.00040523,0.0003267561,0.00042803783,0.0031447913,0.024711547,0.0075967778,0.77683145,0.18220161],"study_design_scores_gemma":[0.00039921547,0.00015656232,0.003758128,0.0005079546,0.00019491168,0.0008747707,0.00012347961,0.036764987,0.029468043,0.026337558,0.90112126,0.00029317834],"about_ca_topic_score_codex":0.0027361782,"about_ca_topic_score_gemma":0.0022765074,"teacher_disagreement_score":0.071361035,"about_ca_system_score_codex":0.0006927899,"about_ca_system_score_gemma":0.0024977126,"threshold_uncertainty_score":0.23872644},"labels":[],"label_agreement":null},{"id":"W2125045287","doi":"10.1093/bioinformatics/btq491","title":"Structural genomics of histone tail recognition","year":2010,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":12,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Structural Genomics Consortium; University of Toronto","funders":"Knut och Alice Wallenbergs Stiftelse; Karolinska Institutet; Stiftelsen för Strategisk Forskning; Ontario Genomics Institute; Canadian Institutes of Health Research; National Institute for Health and Care Research; Ontario Genomics; Genome Canada; Ontario Innovation Trust; Wellcome Trust; GlaxoSmithKline","keywords":"Histone; Computational biology; Computer science; Epigenetics; Genomics; Graphics; Biology; Genetics; Genome; Gene; Computer graphics (images)","score_opus":0.008085528060864508,"score_gpt":0.23325997232090495,"score_spread":0.22517444426004044,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2125045287","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7125972,0.015373022,0.15547448,0.004239472,0.00054014596,0.00022842587,0.060848903,0.008095463,0.042602785],"genre_scores_gemma":[0.81248146,0.0049485713,0.05755472,0.0006991337,0.0002389102,0.00014001048,0.10912967,0.00093672535,0.013870718],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9998983,0.000011777461,0.000005568199,0.000033750763,0.00003489064,0.000015652213],"domain_scores_gemma":[0.9998636,0.000035841622,0.00002586388,0.0000151887925,0.000028016133,0.000031463354],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00022399395,0.0004405743,0.00039201727,0.00037404196,0.00038710324,0.0005167947,0.0005005882,0.00037019933,0.005857464],"category_scores_gemma":[0.00029842526,0.00017475415,0.00040696803,0.0006212593,0.00014320509,0.00036165386,0.00024647667,0.0006955959,0.0025055204],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009246696,0.00023461203,0.004794004,0.0011512047,0.00009768255,0.0004674573,0.00019522122,0.011983044,0.84757036,0.02365231,0.027394194,0.081535265],"study_design_scores_gemma":[0.00020859827,0.00063998014,0.049070526,0.00012614967,0.0002540801,0.00142827,0.00016018971,0.059522606,0.7001309,0.034705903,0.15363342,0.00011943223],"about_ca_topic_score_codex":0.002539992,"about_ca_topic_score_gemma":0.00252298,"teacher_disagreement_score":0.005857464,"about_ca_system_score_codex":0.000693389,"about_ca_system_score_gemma":0.0005289821,"threshold_uncertainty_score":0.019595206},"labels":[],"label_agreement":null},{"id":"W2125122419","doi":"10.1093/bioinformatics/bts317","title":"An integrated open framework for thermodynamics of reactions that combines accuracy and coverage","year":2012,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Microbial Metabolic Engineering and Bioproduction","field":"Biochemistry, Genetics and Molecular Biology","cited_by":131,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Azrieli Foundation; Israel Academy of Sciences and Humanities","keywords":"Python (programming language); Computer science; Gibbs free energy; Ionic strength; Ionic bonding; Open source; Thermodynamics; Data mining; Chemistry; Software; Programming language; Physical chemistry; Physics; Ion; Aqueous solution","score_opus":0.018575142589848163,"score_gpt":0.2830708436086184,"score_spread":0.26449570101877024,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2125122419","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00058144954,0.00030573062,0.98179686,0.00017320992,0.0000829785,0.000087413704,0.00097615627,0.01435099,0.001645189],"genre_scores_gemma":[0.04675224,0.0010695307,0.9298411,0.00033918716,0.00033783514,0.0012761018,0.005736913,0.011036534,0.0036105546],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99334186,0.0013199678,0.0006490984,0.001399336,0.0029023027,0.00038743488],"domain_scores_gemma":[0.99214435,0.0033301401,0.0005944196,0.002138645,0.0014294691,0.00036296385],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.009206251,0.0037999754,0.0035646711,0.0044536283,0.0013756924,0.0056564636,0.007589521,0.003043931,0.01537289],"category_scores_gemma":[0.024047526,0.002129651,0.007856213,0.0028227055,0.002813881,0.007085092,0.007896582,0.004939675,0.0071819834],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002629237,0.0002499388,0.0037873597,0.0017571948,0.00043494595,0.00037237932,0.00039734648,0.28106263,0.008938243,0.51238245,0.024666548,0.16568796],"study_design_scores_gemma":[0.00005471554,0.000043245018,0.00042450885,0.00015383889,0.00008877897,0.00017031742,0.000028889486,0.5836439,0.0049296613,0.3493261,0.06103107,0.0001050642],"about_ca_topic_score_codex":0.006565086,"about_ca_topic_score_gemma":0.0054159723,"teacher_disagreement_score":0.01537289,"about_ca_system_score_codex":0.0031351054,"about_ca_system_score_gemma":0.0054387823,"threshold_uncertainty_score":0.051427484},"labels":[],"label_agreement":null},{"id":"W2125339166","doi":"10.1093/bioinformatics/btn389","title":"DrugViz: a Cytoscape plugin for visualizing and analyzing small molecule drugs in biological networks","year":2008,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Computational Drug Discovery Methods","field":"Computer Science","cited_by":19,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"","keywords":"Plug-in; Computer science; Visualization; Software; Set (abstract data type); Cluster analysis; Data mining; Hierarchical clustering; Computational biology; Interactome; Biological network; Bioinformatics; Information retrieval; Biology; Machine learning; Programming language","score_opus":0.04813335152011109,"score_gpt":0.2986896161002469,"score_spread":0.2505562645801358,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2125339166","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.010518238,0.002037982,0.25835514,0.001112903,0.00050690334,0.0010611479,0.18738188,0.52702075,0.012005117],"genre_scores_gemma":[0.08729002,0.003819529,0.4919546,0.0019158765,0.00031161142,0.0060660685,0.33159855,0.054108806,0.022934971],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99952424,0.00007684951,0.00003539862,0.000091389855,0.00022122666,0.000050843326],"domain_scores_gemma":[0.9988158,0.00068567955,0.000082193554,0.00010582202,0.00016238821,0.00014816684],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00090726785,0.0017926523,0.001115663,0.0044843913,0.00073899416,0.0014426839,0.0025697665,0.0010043278,0.052107576],"category_scores_gemma":[0.003141093,0.0009892111,0.0010739481,0.002427519,0.00026359493,0.0017397942,0.0018132239,0.0021159633,0.012859982],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006273315,0.00018056268,0.0021254332,0.003627608,0.0004935379,0.00063756277,0.0002933937,0.008474918,0.026473053,0.0070170076,0.77004904,0.18000057],"study_design_scores_gemma":[0.0010426291,0.0001532654,0.008870427,0.00042450146,0.00023896527,0.0009859663,0.00011769704,0.12523217,0.07922077,0.027944516,0.7553531,0.00041594447],"about_ca_topic_score_codex":0.005009253,"about_ca_topic_score_gemma":0.008112609,"teacher_disagreement_score":0.052107576,"about_ca_system_score_codex":0.00065753335,"about_ca_system_score_gemma":0.00154651,"threshold_uncertainty_score":0.17431718},"labels":[],"label_agreement":null},{"id":"W2125353579","doi":"10.1093/bioinformatics/bth470","title":"SNP Chart: an integrated platform for visualization and interpretation of microarray genotyping data","year":2004,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":22,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"St. Paul's Hospital; University of British Columbia","funders":"Health Canada; Michael Smith Health Research BC","keywords":"SNP; Computer science; Replicate; Visualization; Chart; Genotyping; SNP genotyping; Data mining; SNP array; Single-nucleotide polymorphism; Data visualization; Computational biology; Biology; Genotype; Genetics; Statistics; Mathematics","score_opus":0.039674676826554726,"score_gpt":0.3162916687121121,"score_spread":0.2766169918855574,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2125353579","genre_codex":"software","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0025769738,0.00028147313,0.4446813,0.00039374112,0.00036259298,0.00063457486,0.08119469,0.46547148,0.0044032345],"genre_scores_gemma":[0.023328366,0.0006571077,0.7306972,0.0007569663,0.0003057878,0.0038511492,0.16069679,0.06314035,0.016566258],"study_design_codex":"not_applicable","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9967464,0.00076118374,0.00043870136,0.0005601372,0.0012746972,0.00021888521],"domain_scores_gemma":[0.99206686,0.0036638638,0.000611477,0.0010341378,0.0020063925,0.00061740767],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0055399043,0.0016197135,0.0017830501,0.004572796,0.0008891934,0.0028213717,0.0032262912,0.001291856,0.07847567],"category_scores_gemma":[0.009958924,0.0016450955,0.00112374,0.0032701772,0.00044437355,0.0019278381,0.0018689184,0.0024744067,0.043223534],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014251169,0.00019397073,0.0025327315,0.0012336264,0.00023334364,0.00048717728,0.0002781701,0.0041136905,0.02653384,0.0037851,0.728522,0.23066124],"study_design_scores_gemma":[0.0012907948,0.00045942568,0.023043463,0.000527735,0.0003241978,0.0015736561,0.00018472623,0.11877531,0.08944431,0.031472694,0.73190403,0.0009997855],"about_ca_topic_score_codex":0.0028178536,"about_ca_topic_score_gemma":0.0031796696,"teacher_disagreement_score":0.07847567,"about_ca_system_score_codex":0.0008604873,"about_ca_system_score_gemma":0.0021093234,"threshold_uncertainty_score":0.2625273},"labels":[],"label_agreement":null},{"id":"W2125598538","doi":"10.1093/bioinformatics/17.2.149","title":"An information-based sequence distance and its application to whole mitochondrial genome phylogeny","year":2001,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genome Rearrangement Algorithms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":542,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"Natural Sciences and Engineering Research Council of Canada; City University of Hong Kong","keywords":"Phylogenetics; Sequence (biology); Biology; Genome; Distance matrices in phylogeny; Evolutionary biology; Mitochondrial DNA; Edit distance; Computational biology; Computer science; Genetics; Algorithm; Bioinformatics; Gene","score_opus":0.009646601906384321,"score_gpt":0.2431462677942042,"score_spread":0.23349966588781987,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2125598538","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.016813979,0.00052387326,0.97951275,0.00038023732,0.000040936964,0.000037298527,0.00033813782,0.0005370465,0.0018157044],"genre_scores_gemma":[0.23173308,0.0007497746,0.76382846,0.00015527535,0.00016505002,0.0001599009,0.0012470225,0.0002835426,0.0016778458],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99853134,0.0004682345,0.000114778086,0.00033746357,0.00047456697,0.00007358803],"domain_scores_gemma":[0.9892484,0.0076123807,0.0009012698,0.0007398419,0.0011545961,0.0003433778],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0023484319,0.00064795377,0.00061299,0.003895644,0.0008934652,0.0018600471,0.0015234003,0.0012122246,0.0027341973],"category_scores_gemma":[0.02084346,0.00034995383,0.0007375583,0.0045451517,0.00159407,0.0035751623,0.0028260788,0.0019215387,0.00083513284],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00025067243,0.00012134966,0.007961324,0.00060892734,0.00012178226,0.0003974965,0.00050615496,0.3486495,0.009097769,0.3029218,0.005748019,0.3236152],"study_design_scores_gemma":[0.000029885381,0.00010100302,0.0024745709,0.00007038544,0.000026557549,0.0005368614,0.00008712811,0.6953413,0.00508429,0.28732187,0.008865397,0.000060692822],"about_ca_topic_score_codex":0.001218242,"about_ca_topic_score_gemma":0.001192133,"teacher_disagreement_score":0.003895644,"about_ca_system_score_codex":0.0014716135,"about_ca_system_score_gemma":0.0014804066,"threshold_uncertainty_score":0.012419879},"labels":[],"label_agreement":null},{"id":"W2125604134","doi":"10.1093/bioinformatics/16.1.41","title":"The early introduction of dynamic programming into computational biology","year":2000,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Algorithms and Data Compression","field":"Computer Science","cited_by":42,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"National Research Council Canada; American Mathematical Society","keywords":"Theme (computing); Field (mathematics); Sentence; Computer science; Mathematical economics; Mathematics; Artificial intelligence; World Wide Web","score_opus":0.0061490985847550295,"score_gpt":0.24222906463850982,"score_spread":0.23607996605375478,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2125604134","genre_codex":"methods","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"review","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0031999613,0.08598342,0.7514456,0.053439997,0.0067934594,0.000111242785,0.00036794678,0.00062010397,0.09803836],"genre_scores_gemma":[0.29277351,0.1232373,0.4626629,0.029434131,0.024240794,0.0011827598,0.00072444696,0.0016220801,0.06412205],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99335337,0.003170149,0.00033868363,0.0016890692,0.0011017076,0.0003469707],"domain_scores_gemma":[0.9819956,0.015403968,0.00032967643,0.0011160363,0.00076200324,0.00039282438],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0056943703,0.0014968286,0.001894557,0.0027534897,0.0018408947,0.005752147,0.002366635,0.0034553097,0.009683408],"category_scores_gemma":[0.018502446,0.0018415949,0.0018131853,0.003952179,0.015660973,0.01191199,0.0057966947,0.014727381,0.0027893996],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000029041894,0.000012798649,0.00008513401,0.00020734544,0.000021390066,0.00004219522,0.00029925446,0.0024920825,0.00010218244,0.9566984,0.005782496,0.034227725],"study_design_scores_gemma":[0.000021037888,0.000024197909,0.000095413605,0.0002656618,0.000007715022,0.000095129035,0.000068020345,0.006211637,0.0002797997,0.81269157,0.18021227,0.000027550093],"about_ca_topic_score_codex":0.0029476814,"about_ca_topic_score_gemma":0.0014179022,"teacher_disagreement_score":0.009683408,"about_ca_system_score_codex":0.005775729,"about_ca_system_score_gemma":0.002929491,"threshold_uncertainty_score":0.04190606},"labels":[],"label_agreement":null},{"id":"W2125642929","doi":"10.1093/bioinformatics/btn440","title":"Local coherence in genetic interaction patterns reveals prevalent functional versatility","year":2008,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":28,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"SickKids Foundation; Canada Research Chairs; Hospital for Sick Children; University of Toronto","funders":"","keywords":"Epistasis; Gene; Computational biology; In silico; Biology; Gene interaction; Cluster analysis; Saccharomyces cerevisiae; Genetics; Function (biology); Partition (number theory); Computer science; Artificial intelligence","score_opus":0.02005842178164167,"score_gpt":0.2350388322449486,"score_spread":0.21498041046330693,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2125642929","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.86057943,0.0005781853,0.13534886,0.00023095375,0.0000070321103,0.000050176848,0.0008591372,0.0007062974,0.0016398728],"genre_scores_gemma":[0.97733307,0.00013106178,0.021433586,0.00003544421,0.000017346436,0.0000306568,0.00076241134,0.000047337864,0.00020897244],"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99914575,0.00014550131,0.0000763928,0.00036564935,0.0001785658,0.00008823704],"domain_scores_gemma":[0.9953324,0.002794039,0.0008202733,0.00046040947,0.0003310214,0.00026187548],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010307707,0.00042178854,0.00058386,0.0038103035,0.00049291406,0.0011987488,0.0005221758,0.00056052505,0.0013429819],"category_scores_gemma":[0.0066563557,0.00029215,0.0004153306,0.0028880455,0.0010635579,0.0011596819,0.0014617522,0.00049005327,0.00027285377],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014115745,0.0004106855,0.36949223,0.00113595,0.00058292015,0.001579755,0.0014362205,0.07012188,0.256434,0.019564822,0.0026596854,0.2751703],"study_design_scores_gemma":[0.00014262533,0.00052187755,0.38196668,0.00009294166,0.00040434435,0.0032202834,0.00088449847,0.4712755,0.05772262,0.07677353,0.006846027,0.0001490331],"about_ca_topic_score_codex":0.0011154212,"about_ca_topic_score_gemma":0.001375631,"teacher_disagreement_score":0.0038103035,"about_ca_system_score_codex":0.00044022046,"about_ca_system_score_gemma":0.00041031605,"threshold_uncertainty_score":0.0054513216},"labels":[],"label_agreement":null},{"id":"W2125810118","doi":"10.1093/bioinformatics/btm224","title":"RankMotif++: a motif-search algorithm that accounts for relative ranks of K-mers in binding transcription factors","year":2007,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":67,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"Natural Sciences and Engineering Research Council of Canada; Ontario Genomics; Ontario Genomics Institute; Genome Canada","keywords":"Motif (music); DNA binding site; Computational biology; Computer science; Algorithm; Data mining; Biology; Genetics; Gene","score_opus":0.020642503547016827,"score_gpt":0.2618901333386715,"score_spread":0.24124762979165468,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2125810118","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.040324762,0.00085616997,0.9202634,0.0003643372,0.00011246306,0.00021172302,0.0016289733,0.034094788,0.0021433183],"genre_scores_gemma":[0.14478476,0.000276981,0.84581184,0.00029448437,0.000068832895,0.0003864708,0.0036091525,0.0016983756,0.0030690113],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9992968,0.00015861634,0.000030933094,0.00024919395,0.00020179147,0.00006257972],"domain_scores_gemma":[0.9991442,0.000382639,0.00009478046,0.00014150301,0.00016870085,0.0000682184],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011863874,0.001709107,0.0012696576,0.0010888636,0.0008046684,0.0008533994,0.0032759958,0.0014988754,0.0043539796],"category_scores_gemma":[0.0036016672,0.0006847193,0.0010258851,0.0010272205,0.00061305834,0.0017156434,0.0011944603,0.001539992,0.0022203054],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010273976,0.00048312082,0.008595895,0.0007332354,0.00027540576,0.00018690612,0.00019532806,0.35868534,0.030293668,0.017254129,0.049272966,0.53299654],"study_design_scores_gemma":[0.00008651289,0.000104762796,0.00031894847,0.000015243616,0.00002225773,0.000110556524,0.000019486879,0.9754283,0.0085225105,0.00940007,0.005942732,0.000028608341],"about_ca_topic_score_codex":0.0046082814,"about_ca_topic_score_gemma":0.009007965,"teacher_disagreement_score":0.0046082814,"about_ca_system_score_codex":0.0007559565,"about_ca_system_score_gemma":0.002050289,"threshold_uncertainty_score":0.014565468},"labels":[],"label_agreement":null},{"id":"W2126122334","doi":"10.1093/bioinformatics/bts542","title":"Application and evaluation of automated methods to extract neuroanatomical connectivity statements from free text","year":2012,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Biomedical Text Mining and Ontologies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":24,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"National Institute of General Medical Sciences; Natural Sciences and Engineering Research Council of Canada; Canadian Institutes of Health Research; National Institutes of Health; Michael Smith Health Research BC","keywords":"Computer science; Artificial intelligence; Natural language processing; Text messaging; Software; Machine learning; Data mining; Programming language; World Wide Web","score_opus":0.045433913833232746,"score_gpt":0.4067668540895707,"score_spread":0.36133294025633794,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2126122334","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6121725,0.008380027,0.23130447,0.0017228756,0.00078425417,0.003879971,0.051454697,0.07585865,0.014442537],"genre_scores_gemma":[0.545368,0.0014340085,0.3627898,0.00034198005,0.00026859937,0.0018439628,0.08114684,0.0013850927,0.0054217707],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.98933804,0.0040067476,0.0016276981,0.0023343265,0.0022569145,0.00043642247],"domain_scores_gemma":[0.92059803,0.05798049,0.0031490629,0.0051418566,0.011964372,0.0011661239],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.012044203,0.002793921,0.0010839867,0.012336367,0.0012099347,0.0030692199,0.0024078588,0.002136392,0.005069456],"category_scores_gemma":[0.04989831,0.0006449429,0.0012942725,0.0046275365,0.0007519549,0.004391008,0.0024821481,0.0011061161,0.0040469086],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0028410077,0.002073376,0.0303071,0.006221988,0.0012681687,0.0012231554,0.0023415869,0.02348241,0.03236118,0.002326249,0.05888581,0.8366679],"study_design_scores_gemma":[0.0014187525,0.0025622647,0.10373283,0.0010257419,0.0010923767,0.002906439,0.002846502,0.72529274,0.10060745,0.006951152,0.051148985,0.00041480132],"about_ca_topic_score_codex":0.011765667,"about_ca_topic_score_gemma":0.010795158,"teacher_disagreement_score":0.012336367,"about_ca_system_score_codex":0.0017169318,"about_ca_system_score_gemma":0.00315281,"threshold_uncertainty_score":0.06369662},"labels":[],"label_agreement":null},{"id":"W2126320151","doi":"10.1093/bioinformatics/btl410","title":"IndexToolkit: an open source toolbox to index protein databases for high-throughput proteomics","year":2006,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Biomedical Text Mining and Ontologies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"","keywords":"Computer science; Search engine indexing; Database; Toolbox; MIT License; Interface (matter); Application programming interface; Information retrieval; Source code; Software; Open source; Database index; License; Database search engine; Index (typography); World Wide Web; Search engine; Programming language; Operating system","score_opus":0.03160131351498109,"score_gpt":0.3029832167974074,"score_spread":0.27138190328242634,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2126320151","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.005852788,0.0032671513,0.3961022,0.0004530502,0.0004475997,0.00084704463,0.114334084,0.46922776,0.009468315],"genre_scores_gemma":[0.022698727,0.003420784,0.4775941,0.00066523725,0.00021743368,0.0025922605,0.4166763,0.06384082,0.01229431],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99883956,0.00013389217,0.00018083815,0.00022329297,0.0005003689,0.00012202024],"domain_scores_gemma":[0.9976501,0.0008386877,0.00033582497,0.00044287994,0.00048123221,0.0002512741],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0021038575,0.0035081794,0.0024019,0.007614188,0.0016228016,0.004559374,0.0049509285,0.0013340616,0.0353654],"category_scores_gemma":[0.007732161,0.0020236415,0.0017538471,0.008522292,0.0006652536,0.0056152577,0.0040719337,0.0022710778,0.04268356],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011526111,0.00021200845,0.0017909347,0.0057348646,0.0003548979,0.00082730665,0.0005089535,0.00240176,0.034075186,0.009018668,0.63958806,0.3043348],"study_design_scores_gemma":[0.00068860856,0.00025277855,0.005993571,0.00091395614,0.00024085521,0.0026375384,0.00024257695,0.0352215,0.07611455,0.032058354,0.8451453,0.0004904471],"about_ca_topic_score_codex":0.0022968757,"about_ca_topic_score_gemma":0.0023145988,"teacher_disagreement_score":0.0353654,"about_ca_system_score_codex":0.0012759792,"about_ca_system_score_gemma":0.0022113642,"threshold_uncertainty_score":0.11830908},"labels":[],"label_agreement":null},{"id":"W2126333123","doi":"10.1093/bioinformatics/bts214","title":"Dissect: detection and characterization of novel structural alterations in transcribed sequences","year":2012,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":18,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canada's Michael Smith Genome Sciences Centre; Simon Fraser University; University of British Columbia","funders":"SFU Community Trust Endowment Fund; Natural Sciences and Engineering Research Council of Canada; Canada's Michael Smith Genome Sciences Centre","keywords":"Transcriptome; Biology; Computational biology; Contig; Genome; Genetics; Reference genome; RNA-Seq; Gene; Gene expression","score_opus":0.014432972382788858,"score_gpt":0.23438011287422625,"score_spread":0.21994714049143738,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2126333123","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.15752804,0.0009074566,0.8221133,0.00053323276,0.000070253234,0.00019661001,0.009136442,0.0073282914,0.0021864001],"genre_scores_gemma":[0.2989386,0.0006286847,0.6794738,0.00024444776,0.000060606322,0.00022998084,0.0181975,0.0008124131,0.0014138299],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99936146,0.000101975435,0.00003916299,0.00030521248,0.00016031424,0.00003175998],"domain_scores_gemma":[0.9980421,0.001171014,0.00034825684,0.00016453234,0.00019466478,0.00007944641],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001167336,0.0009332894,0.0005402057,0.0012002696,0.00051708246,0.0009377819,0.0008253881,0.00063881656,0.0016906391],"category_scores_gemma":[0.0040142033,0.00036996207,0.00073151285,0.0012860633,0.0007479132,0.0008210376,0.00071374094,0.0007947398,0.0009227743],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0015449913,0.00025995015,0.03458393,0.0019549967,0.00022772841,0.001214984,0.0005542866,0.16827005,0.55938923,0.019557472,0.012211452,0.20023094],"study_design_scores_gemma":[0.00012146472,0.00031140697,0.015431734,0.00008838271,0.00012145383,0.0014416166,0.00032335252,0.6917692,0.24115789,0.030838927,0.018311143,0.0000835154],"about_ca_topic_score_codex":0.0012517596,"about_ca_topic_score_gemma":0.0018833912,"teacher_disagreement_score":0.0016906391,"about_ca_system_score_codex":0.0005882232,"about_ca_system_score_gemma":0.0008284177,"threshold_uncertainty_score":0.006173551},"labels":[],"label_agreement":null},{"id":"W2126399647","doi":"10.1093/bioinformatics/btt539","title":"Pattern search in BioPAX models","year":2013,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":28,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"National Institute of General Medical Sciences; National Human Genome Research Institute; National Institutes of Health","keywords":"Computer science; Java; Source code; Documentation; Software; License; Data mining; Programming language","score_opus":0.015280312212947483,"score_gpt":0.22755335150237607,"score_spread":0.2122730392894286,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2126399647","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.039678767,0.00030503573,0.9400653,0.0007458047,0.000052886833,0.00016304235,0.0049276333,0.0073971935,0.0066644424],"genre_scores_gemma":[0.25409076,0.0005589906,0.72705156,0.0003505758,0.000040839583,0.0005603821,0.012252513,0.0011765434,0.003917795],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99911076,0.000276454,0.0000898413,0.00023652836,0.0002283674,0.00005809288],"domain_scores_gemma":[0.99729246,0.0016896368,0.00025496882,0.0004458489,0.00022301811,0.00009396163],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001181273,0.0011012874,0.0007761669,0.0015102567,0.00079689507,0.001916803,0.0016707336,0.0013464054,0.0074121994],"category_scores_gemma":[0.0055449875,0.0005271806,0.0016115376,0.0020621773,0.00095326424,0.003102989,0.0019869118,0.0012642774,0.0017009419],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006634917,0.00024336293,0.0069800597,0.0010799834,0.00017887431,0.0012635429,0.0005321954,0.5556478,0.011535015,0.29001746,0.02433161,0.10752659],"study_design_scores_gemma":[0.00005204352,0.000043830394,0.0003013745,0.000038632515,0.000020596011,0.00024701044,0.000072260584,0.79482496,0.0041942406,0.18615687,0.014028924,0.00001921855],"about_ca_topic_score_codex":0.0032539957,"about_ca_topic_score_gemma":0.003930069,"teacher_disagreement_score":0.0074121994,"about_ca_system_score_codex":0.0009524939,"about_ca_system_score_gemma":0.0008078788,"threshold_uncertainty_score":0.024796247},"labels":[],"label_agreement":null},{"id":"W2126757300","doi":"10.1093/bioinformatics/bts662","title":"Visualization and Phospholipid Identification (VaLID): online integrated search engine capable of identifying and visualizing glycerophospholipids with given mass","year":2012,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Metabolomics and Mass Spectrometry Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":18,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Carleton University; University of Ottawa","funders":"Natural Sciences and Engineering Research Council of Canada; Mitacs; Canadian Institutes of Health Research; University of Ottawa","keywords":"Lipidomics; Glycerophospholipids; Visualization; Identification (biology); Computer science; Glycerophospholipid; Computational biology; Degree of unsaturation; Phospholipid; Data visualization; Bioinformatics; Data science; Information retrieval; Chemistry; Data mining; Biology; Biochemistry; Chromatography","score_opus":0.020670412357506687,"score_gpt":0.2879347834479615,"score_spread":0.2672643710904548,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2126757300","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.02416247,0.003610018,0.18829423,0.0011907308,0.0003418576,0.0007107598,0.17588133,0.5900282,0.015780348],"genre_scores_gemma":[0.12406843,0.0036646798,0.42575872,0.0011832439,0.00026922044,0.0016481312,0.39039406,0.03236801,0.020645507],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9994778,0.000076057855,0.00006172101,0.00014527923,0.00016663036,0.000072479874],"domain_scores_gemma":[0.99894494,0.00045495937,0.00015771564,0.00014877421,0.00016770244,0.00012597918],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012880487,0.0030443552,0.0011511501,0.0046232357,0.0006616678,0.002078918,0.00241742,0.0016989463,0.04178623],"category_scores_gemma":[0.0045238105,0.0007751033,0.001124817,0.002544702,0.000352986,0.0033542884,0.0030782772,0.0006896174,0.018697647],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00325012,0.0003401719,0.010272075,0.0047726696,0.00045667868,0.0013133443,0.0005935521,0.0043439097,0.033969257,0.01199678,0.6564876,0.2722039],"study_design_scores_gemma":[0.0018417934,0.00096513424,0.016583968,0.0012989307,0.00052500796,0.0033773636,0.0006259539,0.19919379,0.11680631,0.04582307,0.6122891,0.00066961464],"about_ca_topic_score_codex":0.0024975142,"about_ca_topic_score_gemma":0.0034811506,"teacher_disagreement_score":0.04178623,"about_ca_system_score_codex":0.00067883387,"about_ca_system_score_gemma":0.0010852457,"threshold_uncertainty_score":0.13978887},"labels":[],"label_agreement":null},{"id":"W2127143129","doi":"10.1093/bioinformatics/btl116","title":"PseudoPipe: an automated pseudogene identification pipeline","year":2006,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":212,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; University of Toronto","funders":"National Institutes of Health; National Human Genome Research Institute; University of Toronto","keywords":"Pseudogene; Genome; Biology; Genetics; Homology (biology); Computational biology; Gene; Retrotransposon; Intergenic region; Transposable element","score_opus":0.010396993762250138,"score_gpt":0.2506661242629637,"score_spread":0.24026913050071355,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2127143129","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.02157127,0.00043091856,0.78228486,0.00036792934,0.00012818209,0.0003961083,0.014078441,0.17799236,0.0027499131],"genre_scores_gemma":[0.083312325,0.00033166335,0.85537404,0.0002741805,0.00005328025,0.0005572363,0.050749492,0.00516409,0.004183701],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.999464,0.00009401933,0.000046787136,0.00016999035,0.00018144854,0.000043771735],"domain_scores_gemma":[0.999027,0.00039466954,0.000088092704,0.00013162126,0.00028060412,0.00007794034],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001136629,0.0011850264,0.0007770127,0.001853638,0.0009918333,0.001325961,0.0023566384,0.0008506719,0.011294344],"category_scores_gemma":[0.0032332237,0.0008791793,0.0010795734,0.0013080998,0.00042219367,0.0022514295,0.0013025104,0.00082778686,0.004578443],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0025392768,0.00042030754,0.0074612927,0.0018588292,0.00027928676,0.0011741155,0.00041480435,0.037412852,0.07250059,0.015740758,0.21946819,0.6407297],"study_design_scores_gemma":[0.0005122904,0.00026929541,0.0037939167,0.00007351975,0.00007615396,0.0014134287,0.00013252588,0.7993017,0.07063017,0.03316265,0.09044697,0.00018743478],"about_ca_topic_score_codex":0.0023686578,"about_ca_topic_score_gemma":0.0026170686,"teacher_disagreement_score":0.011294344,"about_ca_system_score_codex":0.0005525624,"about_ca_system_score_gemma":0.0015930003,"threshold_uncertainty_score":0.037783325},"labels":[],"label_agreement":null},{"id":"W2127166992","doi":"10.1093/bioinformatics/bth211","title":"A weighted least-squares approach for inferring phylogenies from incomplete distance matrices","year":2004,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":61,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal; Université du Québec à Montréal","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Distance matrices in phylogeny; Computer science; Least-squares function approximation; Mathematics; Algorithm; Statistics; Combinatorics","score_opus":0.01559600873354132,"score_gpt":0.2289389724992788,"score_spread":0.21334296376573747,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2127166992","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0014947734,0.000049191098,0.99821836,0.0000325833,0.0000070406672,0.0000074376176,0.000027363987,0.00006473402,0.00009839773],"genre_scores_gemma":[0.052676056,0.00016752064,0.9456771,0.000066750865,0.00005502361,0.0000975481,0.00041597267,0.000078609,0.00076540647],"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99752253,0.001105645,0.00013645882,0.00054181437,0.00060959207,0.0000838842],"domain_scores_gemma":[0.9932875,0.0043027415,0.0005862813,0.0007285952,0.0009343479,0.0001605815],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0033013348,0.0009825054,0.0010513915,0.0015255364,0.0005800273,0.00088807446,0.0025614873,0.0010289653,0.0026061558],"category_scores_gemma":[0.015487667,0.0005866625,0.0008841652,0.0024407627,0.0013490103,0.0025525799,0.0019263999,0.0022885816,0.0009709487],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00015070694,0.000076925586,0.0016557075,0.00038334623,0.00017973057,0.00013301465,0.00019260864,0.577485,0.008410903,0.07786005,0.0025039269,0.33096805],"study_design_scores_gemma":[0.000013797923,0.00004236208,0.00024295603,0.000015430629,0.0000135605915,0.00009899484,0.000020817797,0.95249283,0.0018415409,0.043723237,0.0014746271,0.000019757601],"about_ca_topic_score_codex":0.0022873408,"about_ca_topic_score_gemma":0.0023187017,"teacher_disagreement_score":0.0033013348,"about_ca_system_score_codex":0.00056010956,"about_ca_system_score_gemma":0.0012343451,"threshold_uncertainty_score":0.017459333},"labels":[],"label_agreement":null},{"id":"W2127303081","doi":"10.1093/bioinformatics/bts149","title":"Large-scale analysis of conserved rare codon clusters suggests an involvement in co-translational molecular recognition events","year":2012,"lang":"en","type":"article","venue":"Bioinformatics","topic":"RNA and protein synthesis mechanisms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":51,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Sherbrooke","funders":"National Human Genome Research Institute; Fonds de Recherche du Québec - Santé; Québec Consortium for Drug Discovery","keywords":"Codon usage bias; Computational biology; Biology; Genetics; Start codon; Conserved sequence; Function (biology); Genome; Gene; Peptide sequence; Messenger RNA","score_opus":0.022005665556037578,"score_gpt":0.26770326568454134,"score_spread":0.24569760012850375,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2127303081","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.92968965,0.003263458,0.055500373,0.0003451703,0.00008487633,0.00010325761,0.004839495,0.003005011,0.003168669],"genre_scores_gemma":[0.9210652,0.0009979071,0.065308034,0.00017228155,0.000066641,0.00008527627,0.010409867,0.0005303695,0.0013644523],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9995772,0.00005563134,0.000036882277,0.00015303411,0.00012520586,0.000051919127],"domain_scores_gemma":[0.9981944,0.0006762624,0.0005567316,0.0001753393,0.0001856888,0.00021164888],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00068588933,0.00070467754,0.00074602506,0.0013685146,0.0009618643,0.0007695639,0.0005227012,0.00068022363,0.0051333606],"category_scores_gemma":[0.0016799659,0.00019310266,0.0007392862,0.0018715932,0.0004298925,0.0005832879,0.00055029866,0.00074009853,0.0021936721],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0025912616,0.00024840632,0.07713676,0.00207769,0.00032919727,0.00433942,0.00038073587,0.004409221,0.8463062,0.0022674268,0.004943064,0.0549706],"study_design_scores_gemma":[0.0002356095,0.0011030491,0.31816986,0.00036893875,0.0010021916,0.017621178,0.0005373948,0.081491396,0.5340396,0.009127156,0.036087178,0.00021645728],"about_ca_topic_score_codex":0.00082255056,"about_ca_topic_score_gemma":0.0009998598,"teacher_disagreement_score":0.0051333606,"about_ca_system_score_codex":0.00035459775,"about_ca_system_score_gemma":0.0005334739,"threshold_uncertainty_score":0.017172813},"labels":[],"label_agreement":null},{"id":"W2127313795","doi":"10.1093/bioinformatics/bth224","title":"PedSplit: pedigree management for stratified analysis","year":2004,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Centre for Addiction and Mental Health","funders":"","keywords":"Executable; Documentation; Haplotype; Source code; Software; Computer science; Computational biology; Biology; Gene; Genetics; Programming language; Genotype","score_opus":0.016458840089778656,"score_gpt":0.2733222928392096,"score_spread":0.25686345274943095,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2127313795","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.006732319,0.00037319522,0.43216974,0.0006698952,0.00031732713,0.001020602,0.2519669,0.29834536,0.008404653],"genre_scores_gemma":[0.061608277,0.0007371044,0.46937585,0.0008828457,0.0003498779,0.0046648164,0.34928966,0.09719088,0.015900653],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9987483,0.00039418388,0.00021355,0.00030481353,0.00022748348,0.00011160802],"domain_scores_gemma":[0.99355996,0.003613494,0.0005770182,0.001117303,0.0008651741,0.00026697482],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003998749,0.0017922763,0.0014393535,0.0034202344,0.0007643543,0.0021990803,0.0029368722,0.00069134973,0.19848132],"category_scores_gemma":[0.020110607,0.0018377877,0.0011371082,0.0030676967,0.00031663757,0.0019603632,0.002670315,0.0016472929,0.05583077],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007835412,0.000089064124,0.00853865,0.00093250984,0.0002637884,0.00045972836,0.000349572,0.0059907585,0.0014158294,0.0049537364,0.7919353,0.1842875],"study_design_scores_gemma":[0.0027236482,0.0002598798,0.025353255,0.00095721585,0.000512931,0.0022918999,0.00032079764,0.1250845,0.016930979,0.07683938,0.7483162,0.000409311],"about_ca_topic_score_codex":0.0036343774,"about_ca_topic_score_gemma":0.0041242875,"teacher_disagreement_score":0.19848132,"about_ca_system_score_codex":0.00060704374,"about_ca_system_score_gemma":0.0016489264,"threshold_uncertainty_score":0.6639862},"labels":[],"label_agreement":null},{"id":"W2127341158","doi":"10.1093/bioinformatics/bti700","title":"An attempt to define allergen-specific molecular surface features: a bioinformatic approach","year":2005,"lang":"en","type":"article","venue":"Bioinformatics","topic":"thermodynamics and calorimetric analyses","field":"Chemistry","cited_by":39,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Institute of Infection and Immunity","funders":"Asthma and Lung UK","keywords":"Allergen; Computer science; Computational biology; Surface protein; Biology; Immunology; Allergy; Virology","score_opus":0.012981375310756502,"score_gpt":0.23812484423314886,"score_spread":0.22514346892239237,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2127341158","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.4981193,0.0027792505,0.4836525,0.00078050664,0.00011322547,0.00034802046,0.004674321,0.004890851,0.004641994],"genre_scores_gemma":[0.5769183,0.0012375453,0.41185686,0.0002786844,0.00006699373,0.00021012581,0.0074604247,0.00048078463,0.001490276],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9998671,0.000035718054,0.000012344277,0.00003800731,0.000034421075,0.000012390935],"domain_scores_gemma":[0.99972886,0.00009036878,0.00004317901,0.00003585467,0.000068039626,0.000033762102],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00041693382,0.0006867478,0.0006914697,0.001659078,0.00042315887,0.0007114564,0.00061154313,0.0004810594,0.0018828901],"category_scores_gemma":[0.0009837829,0.00018755281,0.00058111653,0.0011930122,0.00020872742,0.00065081153,0.00029311617,0.00068528426,0.00089748343],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005476486,0.00036025033,0.007788448,0.0009910509,0.00011665727,0.0004189699,0.0002255651,0.0076839137,0.8481246,0.00400661,0.0011383629,0.12859784],"study_design_scores_gemma":[0.00022465736,0.001351067,0.06297175,0.00015056675,0.00063415733,0.0070289937,0.0007221871,0.45664623,0.39992657,0.026145285,0.044010177,0.00018823636],"about_ca_topic_score_codex":0.00035993662,"about_ca_topic_score_gemma":0.00038943565,"teacher_disagreement_score":0.0018828901,"about_ca_system_score_codex":0.00026235735,"about_ca_system_score_gemma":0.00032717388,"threshold_uncertainty_score":0.00629884},"labels":[],"label_agreement":null},{"id":"W2128495797","doi":"10.1093/bioinformatics/btm068","title":"AMPer: a database and an automated discovery tool for antimicrobial peptides","year":2007,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Antimicrobial Peptides and Activities","field":"Immunology and Microbiology","cited_by":233,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Antimicrobial peptides; Hidden Markov model; Construct (python library); Computer science; Computational biology; Biological database; Database; Biology; Artificial intelligence; Bioinformatics; Peptide; Biochemistry","score_opus":0.014180595345511336,"score_gpt":0.2691789252803353,"score_spread":0.25499832993482396,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2128495797","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.046042126,0.0065720975,0.40660462,0.0014744932,0.00041834556,0.0011038316,0.2825943,0.24428073,0.010909467],"genre_scores_gemma":[0.065626375,0.0034160176,0.5875309,0.0006134403,0.00017423391,0.0018183247,0.33132073,0.0051203137,0.004379651],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99912506,0.00019504911,0.00014781719,0.00021491713,0.00024837677,0.00006884241],"domain_scores_gemma":[0.9977756,0.0011166495,0.00036566964,0.00025101347,0.00025871443,0.00023238255],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002097084,0.0017631869,0.0018009165,0.0041658524,0.00069071224,0.0016653793,0.0025846532,0.001991718,0.014312031],"category_scores_gemma":[0.007507848,0.00082689367,0.0011245549,0.0028323776,0.00043439755,0.0027197124,0.0018320306,0.0015905744,0.012984966],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.005637241,0.0008369335,0.00896295,0.008578924,0.00047250366,0.0034316957,0.00057714136,0.024341581,0.09235167,0.021064248,0.31940445,0.51434076],"study_design_scores_gemma":[0.0016570095,0.0010303806,0.012696171,0.0010613591,0.00041851352,0.0045924913,0.00033180855,0.18827425,0.09901717,0.032714102,0.6576998,0.0005069242],"about_ca_topic_score_codex":0.0006633938,"about_ca_topic_score_gemma":0.0006833894,"teacher_disagreement_score":0.014312031,"about_ca_system_score_codex":0.00048281127,"about_ca_system_score_gemma":0.0013721759,"threshold_uncertainty_score":0.047878504},"labels":[],"label_agreement":null},{"id":"W2128591967","doi":"10.1093/bioinformatics/18.3.440","title":"PatternHunter: faster and more sensitive homology search","year":2002,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genome Rearrangement Algorithms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":863,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Bioinformatics Solutions (Canada)","funders":"","keywords":"Computer science; Java; Homology (biology); Genomics; MIT License; Computational biology; Computation; License; Biology; Bioinformatics; Data mining; Algorithm; Genome; Genetics; Operating system; Gene","score_opus":0.015699973549434185,"score_gpt":0.23243666661569154,"score_spread":0.21673669306625737,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2128591967","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.013868901,0.0008828208,0.84301364,0.00060515996,0.00023861487,0.00027503248,0.0027233125,0.13461937,0.0037731477],"genre_scores_gemma":[0.049776576,0.00048304017,0.92336,0.00058328424,0.00014264949,0.00041134408,0.007873977,0.010428361,0.0069408193],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9968233,0.00044315564,0.00028114155,0.0010749275,0.0011805026,0.00019697892],"domain_scores_gemma":[0.9952891,0.0019940701,0.0004999339,0.0013654841,0.0005805859,0.00027086408],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0025891943,0.0018042019,0.0018030462,0.0024759066,0.00072605285,0.0022850828,0.004231278,0.0018649447,0.024220955],"category_scores_gemma":[0.008742313,0.0014584346,0.0011508913,0.0035315445,0.00074487284,0.0065333447,0.0028748726,0.0025251496,0.013868891],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0019022463,0.0003199568,0.0030606522,0.0010887478,0.00025255355,0.00073802174,0.00028983242,0.0073050223,0.14363696,0.012978177,0.10388612,0.7245418],"study_design_scores_gemma":[0.0010766997,0.0005488943,0.0043755076,0.00020372699,0.00022828017,0.0037689193,0.0001401783,0.37958035,0.3611255,0.044925768,0.20365265,0.0003735303],"about_ca_topic_score_codex":0.0014486891,"about_ca_topic_score_gemma":0.0012399931,"teacher_disagreement_score":0.024220955,"about_ca_system_score_codex":0.0006994978,"about_ca_system_score_gemma":0.00085813866,"threshold_uncertainty_score":0.08102715},"labels":[],"label_agreement":null},{"id":"W2129058089","doi":"10.1093/bioinformatics/btr533","title":"Unsupervised detection of genes of influence in lung cancer using biological networks","year":2011,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":15,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto; Ontario Institute for Cancer Research","funders":"Canadian Institutes of Health Research","keywords":"Gene; Computational biology; Biology; Cancer; Gene expression; Lung cancer; Genome; Set (abstract data type); Genetics; Computer science; Medicine; Pathology","score_opus":0.024400771024373887,"score_gpt":0.24892734094882135,"score_spread":0.22452656992444747,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2129058089","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.55514854,0.0012385986,0.43959075,0.0004626014,0.00002349519,0.00014540431,0.0010546884,0.00089049275,0.0014454193],"genre_scores_gemma":[0.94005734,0.0002171162,0.057699937,0.000078410296,0.000047733593,0.000096426265,0.0012219008,0.000042174397,0.00053904793],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9993394,0.00023769856,0.000032466265,0.00021273112,0.00012487333,0.000052784664],"domain_scores_gemma":[0.99732924,0.0018754365,0.00037744988,0.00015965506,0.0001801349,0.000078061996],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010047101,0.0004993608,0.00054253906,0.0022542418,0.00042359068,0.0006408179,0.0007545099,0.0006195127,0.00037847008],"category_scores_gemma":[0.0041744346,0.00025984805,0.0006131098,0.0013681581,0.0006015085,0.00043334637,0.00057060306,0.0005137317,0.00011472325],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00079828023,0.00025114702,0.13690059,0.00031061962,0.0004931435,0.0007744697,0.00035815625,0.6409981,0.046883833,0.005635116,0.002176226,0.16442026],"study_design_scores_gemma":[0.00001871692,0.000028622757,0.024404189,0.000009740548,0.00004741936,0.00015147578,0.000032412034,0.96332645,0.0045140185,0.006861464,0.00058904,0.000016409054],"about_ca_topic_score_codex":0.005554179,"about_ca_topic_score_gemma":0.007416239,"teacher_disagreement_score":0.005554179,"about_ca_system_score_codex":0.0008078403,"about_ca_system_score_gemma":0.0006429958,"threshold_uncertainty_score":0.011043668},"labels":[],"label_agreement":null},{"id":"W2129179457","doi":"10.1093/bioinformatics/btt716","title":"Sequence-based Gaussian network model for protein dynamics","year":2013,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Protein Structure and Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Natural Science Foundation of Zhejiang Province; Natural Sciences and Engineering Research Council of Canada; National Natural Science Foundation of China","keywords":"Sequence (biology); Protein structure prediction; Computer science; Gaussian; Protein function; Gaussian network model; Protein structure; Biological system; Algorithm; Statistical physics; Computational biology; Physics; Biology; Chemistry; Computational chemistry; Genetics","score_opus":0.011948010797763172,"score_gpt":0.23307142691906593,"score_spread":0.22112341612130276,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2129179457","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.043582045,0.0007800904,0.94940454,0.001011842,0.00009000394,0.00006543313,0.0008127709,0.00057898083,0.0036742908],"genre_scores_gemma":[0.84237987,0.0021749719,0.13460962,0.00052801595,0.00021416412,0.00057805236,0.0025914072,0.00034819636,0.01657568],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9993562,0.00025825092,0.000018145969,0.00016147972,0.00013400246,0.000071953524],"domain_scores_gemma":[0.9985453,0.00080939796,0.00020346342,0.00009296676,0.00025436035,0.00009458382],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013898384,0.0010125943,0.0010922737,0.0010658116,0.00056738436,0.0008104769,0.0024138235,0.0017537085,0.0030252696],"category_scores_gemma":[0.0045965114,0.00038595428,0.0008765691,0.001514426,0.0013336742,0.0022439235,0.0010008987,0.0015603869,0.0011030512],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00003608083,0.000018884504,0.00059948396,0.000035899604,0.000016501182,0.000053516684,0.000031803283,0.9641966,0.00063164777,0.02974224,0.0009175817,0.0037197536],"study_design_scores_gemma":[0.0000022220158,0.0000033918552,0.000050345618,0.000001642824,0.0000016209414,0.0000065554786,0.0000013692576,0.993068,0.00004783198,0.0066044126,0.00020964282,0.0000029550254],"about_ca_topic_score_codex":0.017198144,"about_ca_topic_score_gemma":0.009165014,"teacher_disagreement_score":0.017198144,"about_ca_system_score_codex":0.0019813331,"about_ca_system_score_gemma":0.0012510009,"threshold_uncertainty_score":0.03419608},"labels":[],"label_agreement":null},{"id":"W2129397798","doi":"10.1093/bioinformatics/btv279","title":"Data Safe Havens in health research and healthcare","year":2015,"lang":"en","type":"review","venue":"Bioinformatics","topic":"Electronic Health Records Systems","field":"Health Professions","cited_by":75,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; University of Ottawa; University of Toronto","funders":"Economic and Social Research Council; Medical Research Council; Canadian Institutes of Health Research; National Cancer Institute; Cancer Research UK; Wellcome Trust","keywords":"Context (archaeology); Health care; Interpretation (philosophy); Health informatics; Corporate governance; Data governance; Computer science; Biomedicine; Public relations; Data science; Work (physics); Trustworthiness; Knowledge management; Internet privacy; Political science; Business; Data quality; Law; Engineering; Service (business)","score_opus":0.7803968123818096,"score_gpt":0.6547543979116747,"score_spread":0.1256424144701349,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2129397798","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"review","genre_consensus":"review","domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.002397061,0.6736665,0.03807298,0.24370646,0.007227737,0.0005489327,0.00037299935,0.00024389099,0.033763513],"genre_scores_gemma":[0.16020373,0.60092956,0.086559296,0.12865402,0.010987845,0.003818356,0.0011202953,0.00038734186,0.0073395786],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.6048891,0.2902713,0.04157256,0.0145885395,0.044116598,0.004561931],"domain_scores_gemma":[0.37921664,0.51470464,0.035401132,0.038088698,0.026964786,0.005624187],"candidate_categories":["metaresearch","open_science"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.25689188,0.0014390567,0.0051466157,0.011139517,0.0067207413,0.030382115,0.00838425,0.020042371,0.0076669115],"category_scores_gemma":[0.3266755,0.0020552874,0.0037122457,0.016127007,0.07419931,0.056070898,0.0279277,0.018601874,0.002063531],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001002498,0.000026639433,0.0004454049,0.01306675,0.00013666217,0.00017419964,0.0053908164,0.00055431755,0.00011257606,0.8903157,0.01199654,0.07768021],"study_design_scores_gemma":[0.00006888408,0.0001070678,0.0005521787,0.06410526,0.00015074864,0.00054534554,0.0035983643,0.0002861408,0.00043007903,0.49076942,0.43928206,0.0001044562],"about_ca_topic_score_codex":0.0035235693,"about_ca_topic_score_gemma":0.0032673874,"teacher_disagreement_score":0.9916158,"about_ca_system_score_codex":0.016583407,"about_ca_system_score_gemma":0.04338473,"threshold_uncertainty_score":0.91638464},"labels":[],"label_agreement":null},{"id":"W2129668934","doi":"10.1093/bioinformatics/btt274","title":"A combinatorial approach to the peptide feature matching problem for label-free quantification","year":2013,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Advanced Proteomics Techniques and Applications","field":"Chemistry","cited_by":70,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Bipartite graph; Matching (statistics); Computer science; Feature (linguistics); Set (abstract data type); Function (biology); Software; Graph; Feature matching; Data mining; Pattern recognition (psychology); Algorithm; Artificial intelligence; Theoretical computer science; Feature extraction; Mathematics; Statistics","score_opus":0.019295810854236137,"score_gpt":0.26114302299562014,"score_spread":0.241847212141384,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2129668934","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0023626718,0.00020742504,0.99499065,0.0003119207,0.000049093713,0.00008964881,0.0001520392,0.0002790879,0.001557455],"genre_scores_gemma":[0.087012544,0.0005651887,0.9044867,0.000494783,0.0003054305,0.00070848776,0.0012961907,0.0004924749,0.004638202],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9961786,0.0013186408,0.00018843522,0.001169897,0.00083805673,0.00030638487],"domain_scores_gemma":[0.98895395,0.008335457,0.00070914894,0.0007745752,0.0008275885,0.00039924338],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0043644453,0.0026340054,0.0028127118,0.0026965777,0.001442563,0.0035726014,0.0048092916,0.0034442034,0.008818641],"category_scores_gemma":[0.013146674,0.001356644,0.0028402575,0.005406534,0.0023370497,0.005423835,0.0031546734,0.004294161,0.0018861773],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002972262,0.00055556814,0.0010933883,0.0009404342,0.00018899703,0.00041438994,0.0001823762,0.63193774,0.005757357,0.15378031,0.021786857,0.18306525],"study_design_scores_gemma":[0.00004646924,0.00005846814,0.0001714704,0.000028609733,0.000030613166,0.00014602789,0.000028026177,0.85331243,0.0011067388,0.14232394,0.0027146412,0.00003256959],"about_ca_topic_score_codex":0.002250864,"about_ca_topic_score_gemma":0.00212696,"teacher_disagreement_score":0.008818641,"about_ca_system_score_codex":0.0028264741,"about_ca_system_score_gemma":0.0025811938,"threshold_uncertainty_score":0.029501319},"labels":[],"label_agreement":null},{"id":"W2129907963","doi":"10.1093/bioinformatics/16.8.735","title":"<i>In silico</i> identification of transcripts and SNPs from a region of 4p linked with bipolar affective disorder","year":2000,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bipolar Disorder and Treatment","field":"Medicine","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Health Sciences Centre","funders":"Medical Research Council","keywords":"In silico; Identification (biology); Single-nucleotide polymorphism; Computational biology; Bipolar disorder; Genetics; Biology; Computer science; Gene; Neuroscience; Genotype","score_opus":0.008845474385124234,"score_gpt":0.22586827585500927,"score_spread":0.21702280146988503,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2129907963","genre_codex":"dataset","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.019073358,0.0003085372,0.023943389,0.00088093284,0.0005279302,0.00025992066,0.9133548,0.020628957,0.021022229],"genre_scores_gemma":[0.04954439,0.0004115477,0.045274492,0.00067442085,0.00017476929,0.0006912784,0.88335794,0.00633153,0.013539627],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9996557,0.00006544444,0.000028436298,0.000112938505,0.00007286777,0.00006461436],"domain_scores_gemma":[0.99863106,0.0008780672,0.00010260751,0.00016122074,0.00010135099,0.00012566091],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007097336,0.0007812842,0.0011464532,0.0016773854,0.00094461144,0.0008393504,0.0008997053,0.0007681813,0.2193284],"category_scores_gemma":[0.002546964,0.00055242283,0.0013195329,0.0020484661,0.00026741013,0.0003715209,0.0008150925,0.0007204435,0.06019611],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001881735,0.000122060505,0.00879064,0.002548154,0.00020936261,0.0014108451,0.00031910665,0.0019330541,0.12340883,0.0027225825,0.81803113,0.038622566],"study_design_scores_gemma":[0.002595292,0.0008008312,0.12477778,0.0006779256,0.00081215915,0.0027927973,0.00051236775,0.021891285,0.147184,0.009710359,0.68794674,0.0002984564],"about_ca_topic_score_codex":0.0026624785,"about_ca_topic_score_gemma":0.008283537,"teacher_disagreement_score":0.2193284,"about_ca_system_score_codex":0.0004145334,"about_ca_system_score_gemma":0.00063485175,"threshold_uncertainty_score":0.7337266},"labels":[],"label_agreement":null},{"id":"W2129908184","doi":"10.1093/bioinformatics/btq456","title":"A Bayesian method for 3D macromolecular structure inference using class average images from single particle electron microscopy","year":2010,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Advanced Electron Microscopy Techniques and Applications","field":"Biochemistry, Genetics and Molecular Biology","cited_by":26,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"National Center for Research Resources; Canadian Institutes of Health Research; National Institutes of Health","keywords":"Inference; Computer science; Software; Cryo-electron microscopy; Data mining; Algorithm; Artificial intelligence; Pattern recognition (psychology); Physics; Nuclear magnetic resonance","score_opus":0.007808718169959317,"score_gpt":0.33609011869290295,"score_spread":0.32828140052294363,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2129908184","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00074168947,0.00006693012,0.99804866,0.00004180195,0.000012449938,0.00003024601,0.00011092976,0.0008090745,0.00013834664],"genre_scores_gemma":[0.013135714,0.000098757824,0.985114,0.00007105414,0.000042229236,0.00014583717,0.00068126497,0.0003012623,0.00041000344],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99786514,0.00059156027,0.00011450063,0.00045730572,0.00087665557,0.000094733085],"domain_scores_gemma":[0.99360305,0.00369185,0.00044109006,0.0007097327,0.0013181616,0.00023604919],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005386417,0.0013175518,0.0016981038,0.0035810054,0.0014448893,0.002260772,0.004203978,0.0021919133,0.004082172],"category_scores_gemma":[0.014432929,0.0015864477,0.0019601907,0.0026611239,0.0013949177,0.0019715005,0.0019048824,0.00406432,0.0023512421],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000320842,0.0001936999,0.0026962007,0.00039843423,0.00040669038,0.00017197069,0.00021636912,0.26214078,0.011027846,0.030311046,0.017833695,0.67428243],"study_design_scores_gemma":[0.000042888823,0.000016736887,0.00040846935,0.000027490578,0.0000329492,0.00010495909,0.000014408064,0.9612595,0.0031459664,0.030864429,0.004042478,0.00003977845],"about_ca_topic_score_codex":0.0095005855,"about_ca_topic_score_gemma":0.012772727,"teacher_disagreement_score":0.0095005855,"about_ca_system_score_codex":0.0015365382,"about_ca_system_score_gemma":0.0035173313,"threshold_uncertainty_score":0.02848643},"labels":[],"label_agreement":null},{"id":"W2130062792","doi":"10.1093/bioinformatics/btu687","title":"E-MEM: efficient computation of maximal exact matches for very large genomes","year":2014,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genome Rearrangement Algorithms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":33,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"Natural Sciences and Engineering Research Council of Canada; Compute Canada","keywords":"Compressed suffix array; Computer science; Computation; Genome; Suffix; Suffix array; Source code; Code (set theory); Algorithm; Parallel computing; Theoretical computer science; Data structure; Suffix tree; Set (abstract data type); Programming language; Biology","score_opus":0.010695451651004901,"score_gpt":0.23836099971157,"score_spread":0.22766554806056508,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2130062792","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.10595517,0.0014203032,0.85305744,0.00087187276,0.00023906931,0.00025197683,0.0037489322,0.024115192,0.010339995],"genre_scores_gemma":[0.13230567,0.0002978556,0.8590088,0.00022279608,0.00008053609,0.00032364955,0.004065426,0.0010086126,0.002686661],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99949825,0.000112047084,0.000040835945,0.00014551406,0.00014137318,0.00006184306],"domain_scores_gemma":[0.99872607,0.0007467474,0.000095830605,0.00020392124,0.00015724648,0.00007012153],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000967714,0.0012462877,0.0009506467,0.001503265,0.0008022017,0.0011944369,0.002388109,0.0011919399,0.009440734],"category_scores_gemma":[0.0056436104,0.00062531687,0.000903337,0.0019313368,0.0007248738,0.0024922902,0.0018725708,0.001033653,0.0026307856],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0020551884,0.00044035897,0.0075942553,0.001791934,0.00039582126,0.00072421046,0.0005608489,0.21979387,0.04734785,0.04067432,0.08116866,0.5974527],"study_design_scores_gemma":[0.00038488724,0.00023899037,0.001940986,0.000116182375,0.00007519032,0.000531343,0.00024975135,0.8682296,0.043912265,0.060080294,0.024178399,0.00006201454],"about_ca_topic_score_codex":0.0016571885,"about_ca_topic_score_gemma":0.0026738462,"teacher_disagreement_score":0.009440734,"about_ca_system_score_codex":0.0006938914,"about_ca_system_score_gemma":0.0012252693,"threshold_uncertainty_score":0.031582355},"labels":[],"label_agreement":null},{"id":"W2130311175","doi":"10.1093/bioinformatics/btu671","title":"Novel function discovery with GeneMANIA: a new integrated resource for gene function prediction in <i>Escherichia coli</i>","year":2014,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":45,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Regina; University of Toronto","funders":"National Institute of General Medical Sciences; Canadian Institutes of Health Research","keywords":"Function (biology); Computational biology; Biology; Computer science; Escherichia coli; Biogenesis; Resource (disambiguation); Protein function; Gene; Genetics","score_opus":0.007439553823589437,"score_gpt":0.18784870305743182,"score_spread":0.18040914923384238,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2130311175","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.1196995,0.003183064,0.5525047,0.0011356022,0.00021830901,0.00032191907,0.12420665,0.19160782,0.0071224193],"genre_scores_gemma":[0.22939016,0.0015709396,0.53090435,0.00056512066,0.00008240642,0.0007238302,0.2287563,0.006032986,0.0019739252],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9993781,0.00013850372,0.000050577466,0.0002279693,0.00016583143,0.000038907932],"domain_scores_gemma":[0.9987122,0.0006897819,0.0001620802,0.00020913,0.00012995148,0.000096902855],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013618972,0.0016570875,0.0011227273,0.0028463155,0.00068138534,0.0015779779,0.001487186,0.0008992176,0.004603129],"category_scores_gemma":[0.0037683127,0.0006861672,0.0013427109,0.0021066833,0.00039435131,0.0011872206,0.001393643,0.0009658948,0.0023919472],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0041152486,0.00074293936,0.057253145,0.0052664177,0.0016924281,0.002876025,0.0008740489,0.11260128,0.21225174,0.024891132,0.24230656,0.33512896],"study_design_scores_gemma":[0.00041341913,0.0004000737,0.017777927,0.00034448903,0.0005551055,0.0018949272,0.0002189912,0.6927184,0.09652348,0.03209609,0.15685467,0.0002023797],"about_ca_topic_score_codex":0.0018794689,"about_ca_topic_score_gemma":0.0034287812,"teacher_disagreement_score":0.004603129,"about_ca_system_score_codex":0.0007169289,"about_ca_system_score_gemma":0.0013543101,"threshold_uncertainty_score":0.015398979},"labels":[],"label_agreement":null},{"id":"W2130324133","doi":"10.1093/bioinformatics/btt562","title":"INVEX—a web-based tool for integrative visualization of expression data","year":2013,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":89,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"National Institutes of Health; Canadian Institutes of Health Research; Killam Trusts","keywords":"Metadata; Visualization; Annotation; Context (archaeology); Computer science; Variety (cybernetics); Key (lock); Data visualization; Web application; Information retrieval; World Wide Web; Data mining; Biology; Artificial intelligence","score_opus":0.02978220196318498,"score_gpt":0.3035044129120842,"score_spread":0.2737222109488992,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2130324133","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.005870419,0.00031743065,0.39340538,0.00040572922,0.00023464381,0.00037083228,0.062068395,0.5307596,0.0065675727],"genre_scores_gemma":[0.065217845,0.0010275001,0.59135723,0.00095961156,0.0002506984,0.0042770677,0.19845633,0.12255531,0.015898466],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9987049,0.00024110102,0.0001675402,0.00027170972,0.0004976067,0.000117140735],"domain_scores_gemma":[0.9971426,0.0017770324,0.0002041719,0.00034094352,0.00031735725,0.00021783788],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003220223,0.0023205015,0.0015269346,0.0040561133,0.0007435257,0.0023689293,0.0028378195,0.0009219605,0.0643348],"category_scores_gemma":[0.00495509,0.0013851469,0.0017203965,0.0021425793,0.0006026206,0.0023492821,0.003905966,0.002418592,0.01800839],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.002190855,0.0003896828,0.0052645872,0.0022844097,0.00057232863,0.0018473003,0.0011080478,0.0062038265,0.06192971,0.012542914,0.69272226,0.21294414],"study_design_scores_gemma":[0.0012234056,0.00032688567,0.016833145,0.0008569662,0.0002629149,0.002757351,0.00038854888,0.13305102,0.11106362,0.036177915,0.6963789,0.0006793911],"about_ca_topic_score_codex":0.0017364763,"about_ca_topic_score_gemma":0.0018491888,"teacher_disagreement_score":0.0643348,"about_ca_system_score_codex":0.00068980874,"about_ca_system_score_gemma":0.001042536,"threshold_uncertainty_score":0.2152214},"labels":[],"label_agreement":null},{"id":"W2130549711","doi":"10.1093/bioinformatics/btr337","title":"Annotation concept synthesis and enrichment analysis: a logic-based approach to the interpretation of high-throughput experiments","year":2011,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Biomedical Text Mining and Ontologies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Ottawa","funders":"","keywords":"Interpretation (philosophy); Annotation; Throughput; Computer science; Programming language; Artificial intelligence","score_opus":0.029208102939729797,"score_gpt":0.26855426814801336,"score_spread":0.23934616520828356,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2130549711","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0031375445,0.00015612865,0.9948809,0.0002739111,0.000021417334,0.0001597132,0.00020678164,0.0005957086,0.00056780357],"genre_scores_gemma":[0.05806025,0.00021074228,0.9398205,0.0003061224,0.000057659734,0.00042212757,0.0005956658,0.00008678391,0.00044007492],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9896722,0.0043048677,0.0006462156,0.0016900084,0.0033417952,0.00034496066],"domain_scores_gemma":[0.96638507,0.026628673,0.0017580842,0.0017602483,0.0030601674,0.0004077687],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.014292174,0.0026031458,0.0014545323,0.0064649805,0.0013659456,0.0044668107,0.0038490614,0.001262193,0.0034645677],"category_scores_gemma":[0.024324542,0.00070522306,0.0037797454,0.0036106606,0.0035214636,0.0040008426,0.0028981378,0.0030749603,0.00077595754],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011617348,0.0008274314,0.008717907,0.0031990423,0.0010412447,0.0016316104,0.0017281964,0.18871275,0.039579567,0.24280253,0.005141851,0.50545615],"study_design_scores_gemma":[0.000105674015,0.00018796707,0.0011845588,0.0002776196,0.0003001051,0.00034473566,0.00026651012,0.6198054,0.028993236,0.34006312,0.008369319,0.00010172384],"about_ca_topic_score_codex":0.0027111715,"about_ca_topic_score_gemma":0.0022537229,"teacher_disagreement_score":0.014292174,"about_ca_system_score_codex":0.002817674,"about_ca_system_score_gemma":0.004319505,"threshold_uncertainty_score":0.07558519},"labels":[],"label_agreement":null},{"id":"W2130614771","doi":"10.1093/bioinformatics/btt265","title":"DAPPLE: a pipeline for the homology-based prediction of phosphorylation sites","year":2013,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Machine Learning in Bioinformatics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":36,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Saskatchewan","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Phosphorylation; Pipeline (software); Computer science; Computational biology; Homology (biology); Model organism; Organism; Software; Sequence homology; Biology; World Wide Web; Bioinformatics; Programming language; Cell biology; Biochemistry; Genetics; Peptide sequence; Gene","score_opus":0.011620847573372952,"score_gpt":0.23903340122398387,"score_spread":0.22741255365061092,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2130614771","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.022962641,0.0010056649,0.5371435,0.0003873682,0.00023716317,0.000503463,0.117916495,0.31336832,0.0064753043],"genre_scores_gemma":[0.10612171,0.0009541041,0.5735604,0.0004119486,0.00008979477,0.0015520763,0.29534692,0.015901955,0.0060610846],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9996013,0.000054520442,0.000039267652,0.00015167319,0.000111781446,0.000041375195],"domain_scores_gemma":[0.9993467,0.00030254447,0.000069518166,0.00009694509,0.00010546056,0.00007878552],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010830939,0.0020072584,0.0011228261,0.0022121093,0.0008199126,0.0011157767,0.0020810626,0.00092783314,0.024218164],"category_scores_gemma":[0.0029421567,0.0012025286,0.00118034,0.0013366522,0.00029081941,0.0013408068,0.0014689582,0.0020556874,0.014667402],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.002983672,0.0003125667,0.007333388,0.0044866255,0.00054250367,0.0009868463,0.0003729846,0.018813377,0.18198702,0.0084548695,0.5270959,0.2466303],"study_design_scores_gemma":[0.001489318,0.00052990863,0.021135619,0.00029688547,0.00034316178,0.0030334403,0.00022737637,0.37430534,0.21809478,0.03987335,0.34024283,0.00042787567],"about_ca_topic_score_codex":0.001387557,"about_ca_topic_score_gemma":0.0022805007,"teacher_disagreement_score":0.024218164,"about_ca_system_score_codex":0.0006095402,"about_ca_system_score_gemma":0.0011069704,"threshold_uncertainty_score":0.08101785},"labels":[],"label_agreement":null},{"id":"W2130706354","doi":"10.1093/bioinformatics/bth092","title":"TANDEM: matching proteins with tandem mass spectra","year":2004,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Advanced Proteomics Techniques and Applications","field":"Chemistry","cited_by":2606,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Manitoba","funders":"Canadian Institutes of Health Research; University of Chicago","keywords":"Tandem; Matching (statistics); Tandem repeat; Tandem mass spectrometry; Computer science; Computational biology; Chemistry; Biology; Genetics; Mass spectrometry; Chromatography; Mathematics; Genome; Statistics; Materials science","score_opus":0.01073187936705139,"score_gpt":0.2430020231674446,"score_spread":0.2322701438003932,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2130706354","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.02611585,0.0016281242,0.6271328,0.000838981,0.0008390458,0.0007290201,0.04551481,0.2853704,0.011831006],"genre_scores_gemma":[0.087710336,0.001183666,0.7995979,0.001263774,0.0004385813,0.0011153532,0.0802474,0.015586866,0.012856131],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99833196,0.00020287264,0.00024000522,0.0004909745,0.00059507665,0.00013914407],"domain_scores_gemma":[0.99820054,0.0005406424,0.00033369823,0.00044482044,0.00032584846,0.00015448462],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002460954,0.0020850187,0.0008498613,0.0033443752,0.0008748036,0.0020044826,0.0025733146,0.0016887512,0.02797272],"category_scores_gemma":[0.0068813693,0.0010298437,0.001107301,0.0033761472,0.00041795295,0.0032962286,0.0030577907,0.0011339546,0.025352666],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0037927676,0.00041858514,0.0103151025,0.0027887744,0.00065929315,0.0018323284,0.00044673742,0.0045022275,0.11863001,0.01600228,0.40533236,0.4352795],"study_design_scores_gemma":[0.0007285708,0.0005401066,0.012211297,0.0003223288,0.00040671678,0.007221999,0.00026618375,0.08118228,0.29569462,0.050659712,0.55037105,0.00039517038],"about_ca_topic_score_codex":0.0005225064,"about_ca_topic_score_gemma":0.0005285595,"teacher_disagreement_score":0.02797272,"about_ca_system_score_codex":0.0003906845,"about_ca_system_score_gemma":0.00069643767,"threshold_uncertainty_score":0.0935781},"labels":[],"label_agreement":null},{"id":"W2131675589","doi":"10.1093/bioinformatics/btt769","title":"SNPdryad: predicting deleterious non-synonymous human SNPs using only orthologous protein sequences","year":2014,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":61,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Classifier (UML); Biology; Computational biology; Protein sequencing; Genome; Sequence alignment; Human proteome project; Proteome; Computer science; Artificial intelligence; Machine learning; Genetics; Gene; Peptide sequence; Proteomics","score_opus":0.015261747681945676,"score_gpt":0.24683134273027246,"score_spread":0.23156959504832678,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2131675589","genre_codex":"empirical","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.48473474,0.0038426758,0.40598544,0.0010693397,0.0006168363,0.000558085,0.034062177,0.063817784,0.005312907],"genre_scores_gemma":[0.50111073,0.00084957865,0.44930872,0.0005621185,0.00016746052,0.0003058305,0.04405542,0.0015467381,0.002093439],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9989318,0.00022567158,0.00007718365,0.00038668767,0.00029203153,0.000086683795],"domain_scores_gemma":[0.998618,0.00070824864,0.00016900756,0.00014116033,0.00023026009,0.00013331555],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0020292404,0.0013630242,0.0013708241,0.0018747385,0.00077413075,0.0009205745,0.0019771361,0.0012365574,0.0029604605],"category_scores_gemma":[0.0032944367,0.0006593035,0.0008577045,0.0015748017,0.00046284395,0.0010351248,0.0012514705,0.0011413748,0.0020947761],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0049743857,0.00082574703,0.16108334,0.0021914428,0.00097223354,0.001464239,0.0004256723,0.21802886,0.06041318,0.005364705,0.13342565,0.41083056],"study_design_scores_gemma":[0.0005347473,0.0004454193,0.016810775,0.000115038616,0.00017283246,0.0013328921,0.00012410746,0.92354894,0.03094081,0.0068917563,0.018953236,0.00012947319],"about_ca_topic_score_codex":0.003596105,"about_ca_topic_score_gemma":0.0070634177,"teacher_disagreement_score":0.003596105,"about_ca_system_score_codex":0.00056842173,"about_ca_system_score_gemma":0.0013572708,"threshold_uncertainty_score":0.010731757},"labels":[],"label_agreement":null},{"id":"W2131776976","doi":"10.1093/bioinformatics/bts727","title":"Assessing identity, redundancy and confounds in Gene Ontology annotations over time","year":2013,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":66,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"National Institute of General Medical Sciences; Canadian Institutes of Health Research; National Institutes of Health; Michael Smith Health Research BC","keywords":"Redundancy (engineering); Computer science; Gene ontology; Identity (music); Ontology; Gene; Annotation; Computational biology; Artificial intelligence; Biology; Genetics","score_opus":0.011262456417417537,"score_gpt":0.2673509985252674,"score_spread":0.25608854210784987,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2131776976","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"evaluation","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"evaluation","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.95651144,0.0016242535,0.03413396,0.0008112833,0.000070326416,0.00013190416,0.0039298297,0.0003265731,0.0024603305],"genre_scores_gemma":[0.9789128,0.00024117985,0.016607823,0.00011504253,0.00005572412,0.00013096388,0.003453645,0.00007394874,0.00040875914],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.98221004,0.006872364,0.0020166682,0.0038416912,0.0043566762,0.00070271094],"domain_scores_gemma":[0.8143813,0.117266156,0.03996773,0.015381508,0.011146722,0.0018566127],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.026484864,0.00045841074,0.0008310114,0.008298721,0.0014020195,0.002199752,0.0013229039,0.0007956591,0.0012263405],"category_scores_gemma":[0.119480945,0.00022386627,0.0008500257,0.012939467,0.0020503015,0.0030647102,0.0034190577,0.0011205352,0.0002332835],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008402292,0.000086255954,0.9319843,0.00060335896,0.0006501345,0.00035757534,0.0026418408,0.0025237931,0.008211303,0.002398876,0.0013238355,0.048378367],"study_design_scores_gemma":[0.000025685338,0.00035950178,0.9488233,0.00013416378,0.00048474385,0.000965458,0.002291991,0.014254726,0.00752473,0.018538347,0.006508774,0.00008873235],"about_ca_topic_score_codex":0.0035510713,"about_ca_topic_score_gemma":0.0055494593,"teacher_disagreement_score":0.97351515,"about_ca_system_score_codex":0.0014202211,"about_ca_system_score_gemma":0.0011691689,"threshold_uncertainty_score":0.14006704},"labels":[],"label_agreement":null},{"id":"W2131813051","doi":"10.1093/bioinformatics/btu762","title":"EPGA: <i>de novo</i> assembly using the distributions of reads and insert size","year":2014,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genome Rearrangement Algorithms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":25,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Saskatchewan","funders":"","keywords":"Contig; Sequence assembly; Hybrid genome assembly; De Bruijn graph; Computer science; Sequence (biology); Insert (composites); Genome; De Bruijn sequence; Graph; Extension (predicate logic); Reference genome; Algorithm; Computational biology; Biology; Theoretical computer science; Genetics; Mathematics; Combinatorics","score_opus":0.011039391652501954,"score_gpt":0.24156995759420027,"score_spread":0.23053056594169832,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2131813051","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.022279605,0.00030731133,0.9642116,0.00012466773,0.000033643486,0.00021767043,0.00052315264,0.011220331,0.001082084],"genre_scores_gemma":[0.099484645,0.00027512075,0.89437985,0.00013274618,0.000023676923,0.0002645084,0.002799868,0.0011001673,0.0015394026],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99884206,0.00027017097,0.00007412604,0.00043108774,0.00029580484,0.00008675589],"domain_scores_gemma":[0.99720097,0.0012874332,0.00047601468,0.00045497293,0.00049544696,0.00008518219],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015970005,0.0023227457,0.0010191158,0.0021811053,0.0009747374,0.0014875578,0.0015329557,0.001122784,0.002697368],"category_scores_gemma":[0.0053905547,0.00079899945,0.0011908045,0.0018084412,0.0009048019,0.0017368022,0.0014250034,0.0013253178,0.0015182246],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006576119,0.00027629192,0.019677859,0.0008840068,0.00036554175,0.000468098,0.00045604605,0.22771935,0.07156665,0.010725784,0.014163904,0.65303886],"study_design_scores_gemma":[0.000083989435,0.00029831927,0.0059443717,0.000069330454,0.00012729672,0.0005725258,0.00015456922,0.91406316,0.0510115,0.016759608,0.010849918,0.00006531466],"about_ca_topic_score_codex":0.0043203454,"about_ca_topic_score_gemma":0.0045482377,"teacher_disagreement_score":0.0043203454,"about_ca_system_score_codex":0.0009644153,"about_ca_system_score_gemma":0.0015699144,"threshold_uncertainty_score":0.009023607},"labels":[],"label_agreement":null},{"id":"W2131848438","doi":"10.1093/bioinformatics/bti671","title":"Using information theory to search for co-evolving residues in proteins","year":2005,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":287,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"","keywords":"Multiple sequence alignment; Phylogenetic tree; Computational biology; Computer science; Normalization (sociology); In silico; Sequence alignment; Protein sequencing; Conserved sequence; Mutual information; Biology; Genetics; Peptide sequence; Artificial intelligence; Gene","score_opus":0.029206417434222313,"score_gpt":0.2963293837242249,"score_spread":0.2671229662900026,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2131848438","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.15213136,0.001106166,0.843144,0.0006513158,0.000029987224,0.000098399265,0.00026871875,0.00088566187,0.0016843864],"genre_scores_gemma":[0.7353771,0.00052759633,0.26242664,0.00017478605,0.0000858431,0.00024082573,0.000631579,0.00008645922,0.00044921556],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9980661,0.0009962957,0.000109759414,0.00025660056,0.00047972405,0.000091426926],"domain_scores_gemma":[0.9831955,0.014746291,0.0009922134,0.0003849978,0.0004401065,0.00024086621],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0047703185,0.0008417631,0.0013299431,0.005749028,0.0007243569,0.001648716,0.0011920792,0.0014753563,0.00094483595],"category_scores_gemma":[0.020132296,0.0006017356,0.0008842857,0.00259428,0.0017005872,0.0027297582,0.0016435847,0.001032528,0.00025313513],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000546952,0.00028877566,0.02114863,0.0006656888,0.00059398805,0.00060894684,0.000332766,0.7665965,0.010253777,0.053533774,0.0016348232,0.14379527],"study_design_scores_gemma":[0.000017582726,0.000041793093,0.0011563303,0.000012888257,0.000022931694,0.0000657045,0.000019248735,0.96222407,0.0016233688,0.034565873,0.00023068707,0.000019624284],"about_ca_topic_score_codex":0.0012447863,"about_ca_topic_score_gemma":0.0009953898,"teacher_disagreement_score":0.005749028,"about_ca_system_score_codex":0.0015435383,"about_ca_system_score_gemma":0.0009468352,"threshold_uncertainty_score":0.025228202},"labels":[],"label_agreement":null},{"id":"W2131994815","doi":"10.1093/bioinformatics/btu604","title":"SignalSpider: probabilistic pattern discovery on multiple normalized ChIP-Seq signal profiles","year":2014,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":33,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of Toronto","funders":"","keywords":"Chromatin immunoprecipitation; Computational biology; Cluster analysis; DECIPHER; Computer science; ENCODE; Biology; Tiling array; Probabilistic logic; Transcription factor; Genome; Enhancer; ChIP-sequencing; DNA microarray; Gene; Genetics; Promoter; Artificial intelligence; Chromatin remodeling; Gene expression","score_opus":0.007426637499481214,"score_gpt":0.20666372978038328,"score_spread":0.19923709228090206,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2131994815","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.013762226,0.00014089656,0.9262245,0.00031507382,0.00006282101,0.00019128755,0.008128377,0.050302967,0.00087179797],"genre_scores_gemma":[0.15522721,0.00025165066,0.8133892,0.00041472257,0.00006966421,0.0009238619,0.023551669,0.003932542,0.0022395106],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99874866,0.00028316618,0.00007321313,0.00044014474,0.00037411816,0.00008073962],"domain_scores_gemma":[0.99792224,0.0013207138,0.00017636314,0.0002873373,0.00020672023,0.00008657901],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0029028503,0.002035707,0.0012390552,0.0016518444,0.0006377952,0.0017260015,0.0036476192,0.0013267058,0.010170262],"category_scores_gemma":[0.007206959,0.001160864,0.002619574,0.0015305273,0.00074986217,0.0018247892,0.0018307632,0.002204535,0.0026791464],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011851842,0.00035261005,0.011970176,0.0013076836,0.000972808,0.0008149242,0.0002279185,0.6998149,0.023930892,0.022532182,0.04356114,0.19332957],"study_design_scores_gemma":[0.000045847613,0.00002367743,0.00043588143,0.00000676046,0.0000182581,0.000054721648,0.000009753072,0.98642343,0.003910127,0.007153191,0.0018983303,0.000020095944],"about_ca_topic_score_codex":0.006388481,"about_ca_topic_score_gemma":0.008966803,"teacher_disagreement_score":0.010170262,"about_ca_system_score_codex":0.0012519252,"about_ca_system_score_gemma":0.0022514863,"threshold_uncertainty_score":0.034022868},"labels":[],"label_agreement":null},{"id":"W2132028676","doi":"10.1093/bioinformatics/btp567","title":"Adaptive multi-agent architecture for functional sequence motifs recognition","year":2009,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Machine Learning in Bioinformatics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"Calgary Laboratory Services","keywords":"Computer science; Artificial intelligence; Machine learning; Robustness (evolution); Classifier (UML); Solver; Distributed computing; Biology","score_opus":0.041040487647807274,"score_gpt":0.2765442407598718,"score_spread":0.2355037531120645,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2132028676","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.10774736,0.000399931,0.88451177,0.00032539194,0.00008332659,0.0001492355,0.00011457264,0.002404986,0.0042633563],"genre_scores_gemma":[0.83725935,0.00018793794,0.15759295,0.00010084028,0.000025804342,0.00025290612,0.0002371118,0.000048132737,0.0042948974],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9998048,0.000048267968,0.000012092185,0.000058379293,0.00004546788,0.000031139334],"domain_scores_gemma":[0.99971956,0.00009496146,0.000032759956,0.000037714533,0.000082840284,0.00003212004],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00051490456,0.00054969423,0.00042473277,0.00024692828,0.0003829366,0.0005191052,0.0012285713,0.00066965265,0.0022100832],"category_scores_gemma":[0.0009226548,0.00021841067,0.0003495522,0.00019733341,0.00035467633,0.00048088635,0.00066185894,0.00074515824,0.0005889963],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00013415898,0.00014330618,0.0019331878,0.00008570443,0.00007414285,0.0001405029,0.00007961562,0.9084694,0.01512375,0.0041157748,0.001333717,0.06836672],"study_design_scores_gemma":[0.000007212268,0.00002259144,0.00010887349,0.000002249908,0.0000071709806,0.000010295473,0.0000049248215,0.997191,0.0014098175,0.0007601925,0.0004728257,0.0000027076098],"about_ca_topic_score_codex":0.0036416245,"about_ca_topic_score_gemma":0.004061918,"teacher_disagreement_score":0.0036416245,"about_ca_system_score_codex":0.00047543854,"about_ca_system_score_gemma":0.000645637,"threshold_uncertainty_score":0.0073934793},"labels":[],"label_agreement":null},{"id":"W2132034930","doi":"10.1093/bioinformatics/btm057","title":"Cerebral: a Cytoscape plugin for layout of and interaction with biological networks using subcellular localization annotation","year":2007,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Cell Image Analysis Techniques","field":"Biochemistry, Genetics and Molecular Biology","cited_by":179,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"Genome Prairie; Canada Research Chairs; Genome British Columbia; Michael Smith Health Research BC","keywords":"Plug-in; Annotation; Computer science; Java; Subcellular localization; Rendering (computer graphics); Software; Computational biology; Biological network; Function (biology); Artificial intelligence; Biology; Programming language; Gene","score_opus":0.017316755269191048,"score_gpt":0.2790989649097475,"score_spread":0.26178220964055643,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2132034930","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0027608308,0.0003280754,0.497066,0.0005075626,0.00040343034,0.00030151603,0.042663213,0.4497412,0.0062281876],"genre_scores_gemma":[0.04519802,0.0013275035,0.6695596,0.0009184454,0.00036180977,0.0034130828,0.123483405,0.12936486,0.026373236],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.999413,0.00007210414,0.000035086105,0.000128024,0.00028923436,0.00006262229],"domain_scores_gemma":[0.9977939,0.0010738799,0.00012961736,0.00033223335,0.00036126823,0.00030913405],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010983412,0.0019257839,0.0010782095,0.0033559194,0.0008818493,0.0016513484,0.004225979,0.0014073793,0.07849215],"category_scores_gemma":[0.0045161294,0.001329888,0.0012660483,0.0019002715,0.00042777392,0.0017192668,0.0022271825,0.0030619113,0.03122631],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00041224263,0.00014783502,0.0013018947,0.0018791871,0.00024505117,0.0007836113,0.0003798233,0.007440992,0.03893162,0.009990386,0.71743864,0.2210488],"study_design_scores_gemma":[0.0005122923,0.0000869146,0.0049210787,0.00037973316,0.00015582924,0.0011253331,0.00012023545,0.09747359,0.08483964,0.02603788,0.7839138,0.00043359792],"about_ca_topic_score_codex":0.005998163,"about_ca_topic_score_gemma":0.0124727795,"teacher_disagreement_score":0.07849215,"about_ca_system_score_codex":0.0007850047,"about_ca_system_score_gemma":0.0018584251,"threshold_uncertainty_score":0.26258242},"labels":[],"label_agreement":null},{"id":"W2132091595","doi":"10.1093/bioinformatics/btv376","title":"GenomeD3Plot: a library for rich, interactive visualizations of genomic data in web applications","year":2015,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":14,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Dalhousie University; Simon Fraser University","funders":"Natural Sciences and Engineering Research Council of Canada; Genome Canada","keywords":"Visualization; Computer science; JSON; JavaScript; Annotation; World Wide Web; Scalable Vector Graphics; Metadata; Web application; Data visualization; Interactive visualization; Genome browser; Genome; Information retrieval; Genomics; Data mining; Biology","score_opus":0.03629197658024201,"score_gpt":0.29539592336120235,"score_spread":0.2591039467809603,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2132091595","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0017643467,0.000666728,0.29765233,0.00066445273,0.0002949962,0.000273007,0.038589124,0.65281963,0.0072753867],"genre_scores_gemma":[0.025611531,0.001996214,0.53086174,0.0013484035,0.00027338238,0.001713541,0.1153811,0.30412838,0.018685756],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99837816,0.0003093918,0.0001433174,0.00030047054,0.0006981074,0.0001705306],"domain_scores_gemma":[0.99556595,0.0021476827,0.00035089295,0.00071246974,0.0007638707,0.00045913964],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003047212,0.0027066104,0.0019222741,0.004240616,0.0011445715,0.0037970536,0.0039269957,0.0013622293,0.11382587],"category_scores_gemma":[0.008638475,0.0019642734,0.0023793764,0.004138008,0.00068282953,0.0040710038,0.004872723,0.0042631486,0.05944509],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006819559,0.00013769175,0.0017019508,0.0023724718,0.0002713915,0.0006236301,0.000644565,0.0027845618,0.016701553,0.008395404,0.8107384,0.15494646],"study_design_scores_gemma":[0.00056221,0.00015330165,0.005309283,0.0008046831,0.00016583496,0.0012050786,0.00022467383,0.044103626,0.051563703,0.030231476,0.8652207,0.0004554911],"about_ca_topic_score_codex":0.0030476246,"about_ca_topic_score_gemma":0.004226005,"teacher_disagreement_score":0.11382587,"about_ca_system_score_codex":0.0010628084,"about_ca_system_score_gemma":0.002153998,"threshold_uncertainty_score":0.38078547},"labels":[],"label_agreement":null},{"id":"W2132186599","doi":"10.1093/bioinformatics/btp225","title":"PICKY: a novel SVD-based NMR spectra peak picking method","year":2009,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Protein Structure and Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":72,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"York University; University of Waterloo","funders":"National Institutes of Health; Natural Sciences and Engineering Research Council of Canada; Mitacs","keywords":"Computer science; Singular value decomposition; Process (computing); Pruning; NMR spectra database; Algorithm; Artificial intelligence; Pattern recognition (psychology); Spectral line; Physics","score_opus":0.009756982460976367,"score_gpt":0.26225474069432003,"score_spread":0.25249775823334364,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2132186599","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.004194379,0.0005314699,0.9702261,0.00016216801,0.00014905079,0.00016081275,0.000922942,0.022348065,0.0013048964],"genre_scores_gemma":[0.020297827,0.00033027361,0.968856,0.00023780529,0.0001211727,0.00017538326,0.003799721,0.0029733884,0.0032084587],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9976793,0.00019841445,0.00015574797,0.0006022269,0.0011766694,0.00018757969],"domain_scores_gemma":[0.99750346,0.0005457931,0.00021350117,0.0004544623,0.0010604376,0.00022241933],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0020078504,0.003443688,0.0025278258,0.0049596373,0.001580359,0.002805207,0.0036795016,0.0016563255,0.012217186],"category_scores_gemma":[0.0058020703,0.0014208381,0.0021326342,0.0032236804,0.0010450453,0.0037026745,0.0037833198,0.002179992,0.011653625],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001100222,0.00020954995,0.0016608137,0.00082380674,0.00025916297,0.00048589075,0.00021719397,0.018413918,0.061587725,0.0063071535,0.049689837,0.85924464],"study_design_scores_gemma":[0.00034972306,0.00027397246,0.0014916466,0.00010844645,0.00014621594,0.0014071747,0.00020490571,0.84535235,0.08179735,0.020347383,0.048293598,0.00022727741],"about_ca_topic_score_codex":0.0037933926,"about_ca_topic_score_gemma":0.0048119463,"teacher_disagreement_score":0.012217186,"about_ca_system_score_codex":0.0007960875,"about_ca_system_score_gemma":0.002705412,"threshold_uncertainty_score":0.040870607},"labels":[],"label_agreement":null},{"id":"W2132239943","doi":"10.1093/bioinformatics/bts308","title":"BATMAN—an R package for the automated quantification of metabolites from nuclear magnetic resonance spectra using a Bayesian model","year":2012,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Metabolomics and Mass Spectrometry Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":178,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"Biotechnology and Biological Sciences Research Council; Directorate for Biological Sciences","keywords":"Bayesian probability; Deconvolution; Markov chain Monte Carlo; Computer science; Spectral line; Algorithm; Monte Carlo method; Biological system; NMR spectra database; Artificial intelligence; Nuclear magnetic resonance; Pattern recognition (psychology); Mathematics; Statistics; Physics; Biology","score_opus":0.027805285930278055,"score_gpt":0.27322679553577145,"score_spread":0.2454215096054934,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2132239943","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.001933067,0.00091914996,0.7373555,0.00058280537,0.00022514259,0.00040085622,0.029614432,0.226657,0.0023120122],"genre_scores_gemma":[0.018529596,0.0011297491,0.84573495,0.00062547385,0.00019849205,0.0023305123,0.0389614,0.08750347,0.004986363],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9969561,0.0014031698,0.00027255324,0.00053426065,0.0006840672,0.00014988291],"domain_scores_gemma":[0.9883343,0.008029767,0.0011037289,0.0011462509,0.0010942621,0.00029170906],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.006911627,0.0044575394,0.002746897,0.002865973,0.0007834668,0.0025767917,0.0050756596,0.0016847862,0.05482594],"category_scores_gemma":[0.025884487,0.0028231433,0.0034329386,0.0021109781,0.0010415512,0.002330466,0.0032011166,0.004000284,0.0434035],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011844166,0.00018461532,0.005480982,0.0049640564,0.003103836,0.00097385136,0.00045006882,0.054914236,0.009467357,0.020113155,0.65105104,0.2481123],"study_design_scores_gemma":[0.0010450408,0.00026905112,0.0052238354,0.0010150019,0.00095289975,0.0013189825,0.000097877215,0.3819994,0.012954457,0.0649043,0.52976435,0.00045479322],"about_ca_topic_score_codex":0.005066196,"about_ca_topic_score_gemma":0.0048924834,"teacher_disagreement_score":0.05482594,"about_ca_system_score_codex":0.00095011014,"about_ca_system_score_gemma":0.003658958,"threshold_uncertainty_score":0.183411},"labels":[],"label_agreement":null},{"id":"W2132620282","doi":"10.1093/bioinformatics/btp035","title":"The protein–small-molecule database, a non-redundant structural resource for the analysis of protein-ligand binding","year":2009,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Computational Drug Discovery Methods","field":"Computer Science","cited_by":58,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of Toronto","funders":"","keywords":"Computer science; Database; Protein structure database; Redundancy (engineering); Protein Data Bank (RCSB PDB); Protein Data Bank; Annotation; Small molecule; Function (biology); Protein structure; Data mining; Computational biology; Chemistry; Biology; Artificial intelligence; Biochemistry; Genetics","score_opus":0.025551784374906905,"score_gpt":0.295789393044796,"score_spread":0.2702376086698891,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2132620282","genre_codex":"dataset","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.029559294,0.015818022,0.09615268,0.0017228639,0.0003710174,0.0012768761,0.7988522,0.0400393,0.016207656],"genre_scores_gemma":[0.030782036,0.0047623743,0.09158276,0.00022757477,0.00011189466,0.00066717586,0.8673781,0.0014374747,0.0030507315],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99928916,0.00012750081,0.00009015761,0.00011532786,0.00031790335,0.000059847396],"domain_scores_gemma":[0.9984042,0.00036809553,0.00028439303,0.00032103472,0.00028925532,0.0003331589],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001629039,0.002305237,0.002900902,0.0040901513,0.0011095479,0.001990148,0.0044210134,0.0012534729,0.021106785],"category_scores_gemma":[0.0049988823,0.0011664146,0.00081013754,0.0052496125,0.0004884623,0.0019923209,0.0018482455,0.00174137,0.016675554],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0029901145,0.00035033358,0.0064505585,0.009616439,0.0006465426,0.001388144,0.0002854032,0.007941814,0.08730657,0.020347573,0.700414,0.16226257],"study_design_scores_gemma":[0.0024235481,0.00059434946,0.014254944,0.00069374224,0.0008118771,0.0017295482,0.00014533849,0.04573495,0.050959654,0.015975641,0.86642534,0.00025112194],"about_ca_topic_score_codex":0.0046113534,"about_ca_topic_score_gemma":0.0081533445,"teacher_disagreement_score":0.021106785,"about_ca_system_score_codex":0.0013172626,"about_ca_system_score_gemma":0.0046285954,"threshold_uncertainty_score":0.07060921},"labels":[],"label_agreement":null},{"id":"W2132840768","doi":"10.1093/bioinformatics/btu719","title":"deML: robust demultiplexing of Illumina sequences using a likelihood-based approach","year":2014,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":215,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Natural Sciences and Engineering Research Council of Canada; Max-Planck-Gesellschaft","keywords":"Pooling; Quality Score; Multiplexing; Computer science; Sample (material); Set (abstract data type); Illumina dye sequencing; Index (typography); Software; DNA sequencing; Sequence (biology); Data mining; Computational biology; Algorithm; Biology; Genetics; Artificial intelligence; DNA; Metric (unit); Engineering","score_opus":0.027482901823111562,"score_gpt":0.23428084675742514,"score_spread":0.20679794493431358,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2132840768","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0038958313,0.00034272316,0.9562114,0.0002608218,0.0000977005,0.00020467432,0.0036429844,0.033365656,0.0019782912],"genre_scores_gemma":[0.018490493,0.0001718784,0.96570677,0.0003224094,0.00006369881,0.00054703956,0.007570422,0.0048112706,0.0023160651],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9968172,0.00111008,0.00020327658,0.0008365283,0.0008756527,0.00015725053],"domain_scores_gemma":[0.9957775,0.0021191186,0.00042787983,0.00090474583,0.0006413043,0.0001294581],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00753489,0.0024060374,0.0015532655,0.0029825496,0.0010971632,0.0024200317,0.0031376593,0.0014871471,0.019758958],"category_scores_gemma":[0.01625907,0.0020014457,0.0025096543,0.0024400163,0.0011643847,0.0021470145,0.003005823,0.0036030635,0.013715932],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014263181,0.0003001292,0.005236229,0.0022755957,0.0009152804,0.0005484423,0.00057021313,0.06660495,0.07855006,0.029179057,0.13831118,0.6760825],"study_design_scores_gemma":[0.00035154467,0.00021884216,0.0030452714,0.00021589873,0.00013189559,0.0007088167,0.00009804911,0.66883534,0.15937407,0.0599673,0.106761605,0.00029145577],"about_ca_topic_score_codex":0.0015208842,"about_ca_topic_score_gemma":0.0028004195,"teacher_disagreement_score":0.019758958,"about_ca_system_score_codex":0.0017592391,"about_ca_system_score_gemma":0.001461517,"threshold_uncertainty_score":0.06610036},"labels":[],"label_agreement":null},{"id":"W2133001645","doi":"10.1093/bioinformatics/btu644","title":"PINBPA: Cytoscape app for network analysis of GWAS data","year":2014,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":33,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Queen's University","funders":"National Multiple Sclerosis Society","keywords":"Genome-wide association study; Computer science; Interface (matter); Data mining; Computational biology; Biology; Gene; Genetics; Operating system; Single-nucleotide polymorphism","score_opus":0.024696411385221217,"score_gpt":0.2854148186385947,"score_spread":0.2607184072533735,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2133001645","genre_codex":"dataset","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0059446543,0.0013677485,0.1968196,0.00093648245,0.0011804482,0.0009747559,0.4073831,0.37045342,0.014939775],"genre_scores_gemma":[0.07699845,0.0027935388,0.35741636,0.0026483047,0.00064468954,0.014127387,0.42187262,0.097763896,0.025734685],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99900913,0.00019492384,0.00007689998,0.00025205853,0.00037171884,0.0000952554],"domain_scores_gemma":[0.9958385,0.0026008685,0.0002109684,0.00046779372,0.00053260016,0.0003493002],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0017577918,0.0023393866,0.0017090873,0.0057811257,0.0011765519,0.0019045903,0.004161744,0.0014802085,0.18336958],"category_scores_gemma":[0.01040603,0.0015511195,0.0017132774,0.003258089,0.000423789,0.002065304,0.00235006,0.003314325,0.051653944],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00045661503,0.0001218796,0.002427841,0.0064104693,0.00083837635,0.00054904434,0.00039100388,0.0057980632,0.007741378,0.008646895,0.89550996,0.07110843],"study_design_scores_gemma":[0.00080504804,0.00009983832,0.00766679,0.0010200957,0.00050897105,0.001108762,0.00014985277,0.06531422,0.02995647,0.06045158,0.8324513,0.00046709637],"about_ca_topic_score_codex":0.006136739,"about_ca_topic_score_gemma":0.010617721,"teacher_disagreement_score":0.18336958,"about_ca_system_score_codex":0.00076199864,"about_ca_system_score_gemma":0.0030821972,"threshold_uncertainty_score":0.61343235},"labels":[],"label_agreement":null},{"id":"W2133360663","doi":"10.1093/bioinformatics/btq166","title":"Picante: R tools for integrating phylogenies and ecology","year":2010,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Ecology and Vegetation Dynamics Studies","field":"Environmental Science","cited_by":6383,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"Natural Sciences and Engineering Research Council of Canada; Gordon and Betty Moore Foundation; Chinese Academy of Sciences; National Science Foundation","keywords":"R package; Phylogenetic tree; Trait; Phylogenetic diversity; Set (abstract data type); License; Open source; Diversity (politics); Software package; Computer science; Software; Biology; Data science; Ecology; Evolutionary biology; Sociology; Programming language; Anthropology","score_opus":0.01060361882652631,"score_gpt":0.23638697334989223,"score_spread":0.22578335452336593,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2133360663","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0023421696,0.0015219189,0.65556514,0.0006567493,0.0003631524,0.00028995963,0.091420405,0.24229506,0.0055453773],"genre_scores_gemma":[0.025629805,0.0013850499,0.6811977,0.0008589585,0.0002654141,0.0026922647,0.13910745,0.14281666,0.006046737],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99586785,0.0017506223,0.00045390258,0.0008722167,0.00084221165,0.0002133506],"domain_scores_gemma":[0.9901716,0.0058882693,0.00095896365,0.0016270168,0.00096814556,0.000386094],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007217632,0.0038644886,0.0041063945,0.005318498,0.001168773,0.0036092848,0.006948134,0.0012586011,0.05849132],"category_scores_gemma":[0.027236896,0.0031783022,0.0037118206,0.004425957,0.0011518697,0.003895859,0.004695412,0.005781225,0.04842932],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008800777,0.00010729968,0.006233744,0.0053448207,0.002937778,0.00071203534,0.00091849826,0.020380724,0.0071379803,0.03320583,0.7608152,0.16132599],"study_design_scores_gemma":[0.0005660763,0.00013756343,0.007926461,0.0007962855,0.00097310153,0.0013408222,0.00013923942,0.09050055,0.007998968,0.111418985,0.7777919,0.00040999867],"about_ca_topic_score_codex":0.0034708225,"about_ca_topic_score_gemma":0.0026415172,"teacher_disagreement_score":0.05849132,"about_ca_system_score_codex":0.00077875506,"about_ca_system_score_gemma":0.002179768,"threshold_uncertainty_score":0.19567293},"labels":[],"label_agreement":null},{"id":"W2133525847","doi":"10.1093/bioinformatics/btl235","title":"Integrating image data into biomedical text categorization","year":2006,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Biomedical Text Mining and Ontologies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":118,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Queen's University","funders":"Natural Sciences and Engineering Research Council of Canada; Xerox Foundation","keywords":"Computer science; Annotation; Categorization; Information retrieval; Task (project management); Text categorization; Feature (linguistics); Artificial intelligence; Natural language processing","score_opus":0.015740115506897037,"score_gpt":0.27755073193468244,"score_spread":0.2618106164277854,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2133525847","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.10884064,0.006781992,0.84506637,0.0029286987,0.0005825154,0.000953729,0.009363782,0.01357989,0.011902402],"genre_scores_gemma":[0.16869038,0.0019982695,0.8128728,0.00040445698,0.00048508693,0.00033421352,0.011146557,0.0003189728,0.0037492805],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9990363,0.00016104116,0.00010479038,0.00025203679,0.00036534684,0.00008048799],"domain_scores_gemma":[0.9963206,0.0014579683,0.00031800507,0.0006578663,0.0010842286,0.00016135351],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011219992,0.00093128055,0.0009563782,0.009778846,0.0005494352,0.0022753077,0.001083211,0.001498062,0.0031567556],"category_scores_gemma":[0.0054279445,0.00033390019,0.00092079956,0.005108669,0.00067501154,0.0030697025,0.001324425,0.00091150875,0.0030247173],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00028680908,0.00040307594,0.0060221893,0.00072129176,0.00010846376,0.0003440718,0.00019563979,0.00362822,0.093587935,0.0035191628,0.011696801,0.8794864],"study_design_scores_gemma":[0.00017079165,0.0009899054,0.053649984,0.00051828666,0.0006003064,0.0026293898,0.0015677648,0.39424908,0.32659242,0.09884266,0.1198579,0.00033149472],"about_ca_topic_score_codex":0.00205477,"about_ca_topic_score_gemma":0.0027658117,"teacher_disagreement_score":0.009778846,"about_ca_system_score_codex":0.0005417443,"about_ca_system_score_gemma":0.0005725572,"threshold_uncertainty_score":0.010560393},"labels":[],"label_agreement":null},{"id":"W2133550211","doi":"10.1093/bioinformatics/btr058","title":"A common layer of interoperability for biomedical ontologies based on OWL EL","year":2011,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Biomedical Text Mining and Ontologies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":45,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Carleton University","funders":"National Human Genome Research Institute; Biotechnology and Biological Sciences Research Council; European Commission; Natural Sciences and Engineering Research Council of Canada; National Institutes of Health; European Bioinformatics Institute","keywords":"Computer science; Interoperability; Ontology; Open Biomedical Ontologies; Web Ontology Language; IDEF5; Semantics (computer science); Knowledge representation and reasoning; Ontology components; Upper ontology; Description logic; Software; Information retrieval; Semantic Web; Software engineering; World Wide Web; Programming language; Artificial intelligence; Ontology alignment","score_opus":0.057446294203414114,"score_gpt":0.30675205372400544,"score_spread":0.2493057595205913,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2133550211","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.008901391,0.00018316887,0.9679,0.0010196663,0.00010845165,0.00044482073,0.0010966402,0.012653535,0.0076924236],"genre_scores_gemma":[0.10507289,0.00026415143,0.882709,0.0007974674,0.000067759975,0.00051459676,0.0052888207,0.0016064357,0.003678855],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9899768,0.0028571754,0.0016661291,0.0011734875,0.003747214,0.0005791947],"domain_scores_gemma":[0.98283374,0.004107384,0.0012809249,0.00843246,0.0027932154,0.00055229454],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.012057998,0.00085046026,0.0010164928,0.0028144857,0.0017316607,0.0069727106,0.0035462186,0.001769876,0.004038912],"category_scores_gemma":[0.025064815,0.0012012081,0.0030847038,0.002545581,0.0024118815,0.00940431,0.007785522,0.0042343494,0.0022202653],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007052248,0.0007761928,0.005324319,0.0010849007,0.00052348466,0.0017860516,0.0042556343,0.02643102,0.031701475,0.5418651,0.03558814,0.34995848],"study_design_scores_gemma":[0.00015819514,0.00021335408,0.0032946651,0.00075911224,0.00037491904,0.0016555097,0.0009466902,0.24097832,0.054735668,0.387515,0.30915844,0.00021013748],"about_ca_topic_score_codex":0.006744383,"about_ca_topic_score_gemma":0.0048336126,"teacher_disagreement_score":0.012057998,"about_ca_system_score_codex":0.0018468738,"about_ca_system_score_gemma":0.00468078,"threshold_uncertainty_score":0.06376958},"labels":[],"label_agreement":null},{"id":"W2133594999","doi":"10.1093/bioinformatics/btr146","title":"<i>anota</i>: analysis of differential translation in genome-wide studies","year":2011,"lang":"en","type":"article","venue":"Bioinformatics","topic":"RNA and protein synthesis mechanisms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":65,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University and Génome Québec Innovation Centre","funders":"Knut och Alice Wallenbergs Stiftelse","keywords":"Bioconductor; Translation (biology); Computational biology; Polysome; Biology; Microarray analysis techniques; Gene expression profiling; Eukaryotic translation; Messenger RNA; DNA microarray; Gene expression; Computer science; Gene; Ribosome; Genetics; RNA","score_opus":0.04230599683304403,"score_gpt":0.2562835172677359,"score_spread":0.21397752043469187,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2133594999","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.018093174,0.001578308,0.72632235,0.0014894468,0.0009857913,0.0003923994,0.11533708,0.12890126,0.0069001527],"genre_scores_gemma":[0.104143895,0.00081328605,0.7610988,0.0009903039,0.00041744654,0.0032805405,0.10085698,0.024310958,0.004087802],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99618345,0.001523707,0.00037882657,0.00093998906,0.00079313165,0.00018078847],"domain_scores_gemma":[0.98616034,0.008470028,0.0014319873,0.0019669968,0.0015332799,0.00043729137],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0055611026,0.001683409,0.002052038,0.003478677,0.0008792325,0.002210198,0.0024344595,0.0007591307,0.025979936],"category_scores_gemma":[0.024055634,0.00065513636,0.0017411286,0.0059285224,0.0008015694,0.0010861006,0.0018297902,0.0015232293,0.010930681],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0022222886,0.00016852845,0.0328968,0.00781275,0.0028461767,0.000880049,0.00090174313,0.007141418,0.07277188,0.02245319,0.5825211,0.26738408],"study_design_scores_gemma":[0.0007191963,0.0007025218,0.14165261,0.00091944775,0.0019196006,0.002032065,0.00025649165,0.1602004,0.067722455,0.06416788,0.55918664,0.00052069576],"about_ca_topic_score_codex":0.0011381633,"about_ca_topic_score_gemma":0.002406454,"teacher_disagreement_score":0.025979936,"about_ca_system_score_codex":0.00045112934,"about_ca_system_score_gemma":0.0012770803,"threshold_uncertainty_score":0.08691156},"labels":[],"label_agreement":null},{"id":"W2133601033","doi":"10.1093/bioinformatics/btp602","title":"Evaluation of linguistic features useful in extraction of interactions from PubMed; Application to annotating known, high-throughput and predicted interactions in I2D","year":2009,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Biomedical Text Mining and Ontologies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":77,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"U.S. National Library of Medicine; Ontario Genomics; Ontario Genomics Institute; Genome Canada","keywords":"Computer science; Sentence; Identification (biology); Recall; Precision and recall; Software; Natural language processing; Information retrieval; Process (computing); Artificial intelligence; Data mining; Machine learning; Programming language; Biology","score_opus":0.02690545087677414,"score_gpt":0.33233209653783674,"score_spread":0.3054266456610626,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2133601033","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8493561,0.0052263965,0.07697894,0.0020166857,0.00021675602,0.0015794833,0.04192663,0.015820256,0.0068787336],"genre_scores_gemma":[0.56926537,0.0010610692,0.3923007,0.0003391572,0.00010223134,0.0007924326,0.03459542,0.00029715532,0.0012464212],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.996698,0.0012184442,0.0008625767,0.00052888534,0.0005683697,0.00012361039],"domain_scores_gemma":[0.9751209,0.01913846,0.001765548,0.0007416039,0.0026775203,0.0005560324],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0059854006,0.001101822,0.0008591355,0.009662319,0.0010664073,0.0016143203,0.0011033688,0.001078992,0.0018751792],"category_scores_gemma":[0.018291546,0.00026435906,0.00071305357,0.005032147,0.00038706244,0.0015085334,0.0012107014,0.00046340618,0.0007690819],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.005479754,0.0021151102,0.108122535,0.010525835,0.00093403563,0.003622109,0.0027997443,0.009720945,0.1188782,0.0017934073,0.041388907,0.6946194],"study_design_scores_gemma":[0.0018667963,0.0029654664,0.28063488,0.0010558692,0.0022572582,0.007181465,0.004946539,0.4013218,0.22121535,0.0047447262,0.07128467,0.00052522606],"about_ca_topic_score_codex":0.00444016,"about_ca_topic_score_gemma":0.0085008275,"teacher_disagreement_score":0.009662319,"about_ca_system_score_codex":0.0008938867,"about_ca_system_score_gemma":0.0014336987,"threshold_uncertainty_score":0.03165418},"labels":[],"label_agreement":null},{"id":"W2134335091","doi":"10.1093/bioinformatics/btn211","title":"A correction for estimating error when using the Local Pooled Error Statistical Test","year":2008,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University and Génome Québec Innovation Centre","funders":"","keywords":"Bioconductor; Sample size determination; Software; Statistics; Computer science; Type I and type II errors; Sample (material); R package; Standard error; Statistical hypothesis testing; Multiple comparisons problem; Data mining; Mathematics; Programming language; Biology","score_opus":0.03975235909724276,"score_gpt":0.3027381671937055,"score_spread":0.2629858080964627,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2134335091","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0021739523,0.00042039843,0.99266243,0.0007099005,0.0013093967,0.00038833288,0.0003890537,0.0012360639,0.00071040203],"genre_scores_gemma":[0.07702819,0.00035857613,0.91056657,0.0017543503,0.0011070971,0.0036064244,0.0008858698,0.0022039302,0.0024890096],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.7396227,0.19693789,0.015344542,0.018815443,0.027324477,0.0019549113],"domain_scores_gemma":[0.48467392,0.38632843,0.01918013,0.07177409,0.036728237,0.0013151546],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.16900942,0.00268417,0.0036990494,0.005112155,0.0018679891,0.0038603244,0.007042673,0.0045056758,0.010628733],"category_scores_gemma":[0.5387313,0.0013207821,0.0050947308,0.007664786,0.0050891237,0.004035139,0.005789623,0.0086570345,0.00412409],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0016978752,0.00037323413,0.027433654,0.0042125136,0.0053117913,0.0014405104,0.0023301465,0.02755008,0.008059487,0.09179918,0.073658414,0.7561331],"study_design_scores_gemma":[0.00085985794,0.0026867962,0.047075372,0.003930188,0.0037449254,0.0037737698,0.0011314963,0.2986503,0.054115336,0.31725326,0.2656152,0.0011633955],"about_ca_topic_score_codex":0.0016807857,"about_ca_topic_score_gemma":0.0017665034,"teacher_disagreement_score":0.16900942,"about_ca_system_score_codex":0.0021960575,"about_ca_system_score_gemma":0.0040873187,"threshold_uncertainty_score":0.8938179},"labels":[],"label_agreement":null},{"id":"W2134384386","doi":"10.1093/bioinformatics/btl109","title":"Construction of null statistics in permutation-based multiple testing for multi-factorial microarray experiments","year":2006,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Statistical Methods in Clinical Trials","field":"Mathematics","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"York University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Null distribution; Statistics; Permutation (music); Test statistic; Mathematics; Null (SQL); Statistical hypothesis testing; Statistic; False discovery rate; Null hypothesis; Asymptotic distribution; Ancillary statistic; p-value; Data mining; Computer science; Biology; Genetics","score_opus":0.4215325324378103,"score_gpt":0.49919503487273753,"score_spread":0.07766250243492723,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2134384386","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0021448839,0.00012461668,0.9968605,0.00007153268,0.000058264137,0.0001130179,0.00005474312,0.00029449447,0.00027786774],"genre_scores_gemma":[0.082865395,0.00032014828,0.9138648,0.00021370918,0.00013670605,0.0015779658,0.000499967,0.0002532673,0.00026807154],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9623488,0.023059556,0.0020723103,0.004427743,0.0073570544,0.0007344821],"domain_scores_gemma":[0.9069499,0.07118377,0.005292782,0.0075524063,0.00817546,0.00084575027],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.04196006,0.0019926215,0.0025054205,0.003946194,0.0013314956,0.002281471,0.0036561203,0.0021550593,0.0027346993],"category_scores_gemma":[0.10771829,0.0009531716,0.0025365767,0.0038219811,0.0052475315,0.0036224273,0.0030771845,0.004232819,0.001143363],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0018474269,0.00044597907,0.014545569,0.0031055447,0.0009654319,0.0016630719,0.0014687732,0.11777854,0.028365713,0.2831029,0.009749577,0.5369615],"study_design_scores_gemma":[0.00028707713,0.0012439537,0.003884293,0.000376515,0.00028210497,0.0014826796,0.00025939022,0.5661048,0.025345897,0.38637644,0.014106363,0.00025050793],"about_ca_topic_score_codex":0.00057859946,"about_ca_topic_score_gemma":0.000306803,"teacher_disagreement_score":0.04196006,"about_ca_system_score_codex":0.0014211134,"about_ca_system_score_gemma":0.0041194567,"threshold_uncertainty_score":0.22190869},"labels":[],"label_agreement":null},{"id":"W2134425821","doi":"10.1093/bioinformatics/btt031","title":"Computational phosphorylation site prediction in plants using random forests and organism-specific instance weights","year":2013,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":44,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Saskatchewan","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Phosphorylation; Organism; Random forest; Computer science; Arabidopsis thaliana; Arabidopsis; Computational biology; Interface (matter); Biology; Artificial intelligence; Biochemistry; Mutant; Genetics; Gene","score_opus":0.009811674771357713,"score_gpt":0.2033164158562933,"score_spread":0.19350474108493557,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2134425821","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.56252354,0.0019572696,0.41981712,0.00070532796,0.00013938623,0.00015347585,0.0038029314,0.008998629,0.0019022229],"genre_scores_gemma":[0.77326155,0.0002664797,0.22019538,0.00017256303,0.00007341689,0.00011509459,0.004859155,0.00020958153,0.00084681786],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99961424,0.00015356393,0.000017394757,0.00012882044,0.000041141615,0.000044860408],"domain_scores_gemma":[0.9976241,0.0019371674,0.00012118819,0.000113300746,0.000112270536,0.00009193602],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016731523,0.0010525655,0.00090476655,0.0013161503,0.00033104437,0.0004972974,0.0012002782,0.0012147351,0.001613519],"category_scores_gemma":[0.0031604127,0.00058081094,0.0013205021,0.00094362505,0.0003171237,0.00076130545,0.0003976547,0.0009440838,0.00043090383],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00033609086,0.00010624698,0.00390664,0.000119482786,0.00017711909,0.00014715455,0.00001910529,0.943923,0.0014465404,0.0007825805,0.0021432047,0.046892744],"study_design_scores_gemma":[0.000016006245,0.000012808195,0.00023446715,0.000002744903,0.000008720214,0.000012346005,0.0000024999085,0.99859947,0.00024040126,0.0007945883,0.000073258205,0.0000027159758],"about_ca_topic_score_codex":0.0064180754,"about_ca_topic_score_gemma":0.010732322,"teacher_disagreement_score":0.0064180754,"about_ca_system_score_codex":0.0007989659,"about_ca_system_score_gemma":0.0007877211,"threshold_uncertainty_score":0.012761414},"labels":[],"label_agreement":null},{"id":"W2134459098","doi":"10.1093/bioinformatics/btt574","title":"PLIDA: cross-platform gene expression normalization using perturbed topic models","year":2013,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":23,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Normalization (sociology); Computer science; Data mining; Multiplicative function; Database normalization; Machine learning; MATLAB; Artificial intelligence; Pattern recognition (psychology); Programming language; Mathematics","score_opus":0.033973699461407575,"score_gpt":0.27516427195463017,"score_spread":0.2411905724932226,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2134459098","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0064785574,0.00045567672,0.97772855,0.00020638548,0.00014504064,0.000107934306,0.0013166025,0.012567421,0.0009938326],"genre_scores_gemma":[0.15077129,0.0007478164,0.8236093,0.0004370063,0.00026879486,0.001270811,0.012180196,0.003995208,0.006719603],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9967073,0.0006825042,0.00014876011,0.0014324072,0.0008704489,0.00015847759],"domain_scores_gemma":[0.99643975,0.0015141178,0.00027304856,0.0009337115,0.000732233,0.000107131986],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005119747,0.0020884164,0.0015528877,0.0017137496,0.00085135136,0.0020861092,0.0027174538,0.0010970404,0.0037717223],"category_scores_gemma":[0.01285905,0.00094638113,0.0020529435,0.002835838,0.0009147231,0.0020445094,0.002520881,0.0033666813,0.0036388694],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009299163,0.00028722626,0.0077563343,0.00059332023,0.0009142538,0.00020888826,0.00042621125,0.13298711,0.063574225,0.011140409,0.037536804,0.7436452],"study_design_scores_gemma":[0.00009741984,0.00017438145,0.006627377,0.000044560813,0.00019986756,0.00027389324,0.00009373833,0.8824061,0.061385468,0.023618283,0.024971407,0.00010756165],"about_ca_topic_score_codex":0.0039895186,"about_ca_topic_score_gemma":0.0051214425,"teacher_disagreement_score":0.005119747,"about_ca_system_score_codex":0.0014212615,"about_ca_system_score_gemma":0.0019456827,"threshold_uncertainty_score":0.027076125},"labels":[],"label_agreement":null},{"id":"W2134594748","doi":"10.1093/bioinformatics/18.10.1305","title":"Short inversions and conserved gene cluster","year":2002,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":32,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"","keywords":"Cluster (spacecraft); Gene; Computational biology; Genetics; Gene cluster; Conserved sequence; Biology; Computer science; Base sequence; Programming language","score_opus":0.012863111426107332,"score_gpt":0.2023517259428525,"score_spread":0.18948861451674517,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2134594748","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9919292,0.0005964392,0.006171363,0.00026144265,0.000009196491,0.0000076015795,0.000120022276,0.000057206664,0.0008475494],"genre_scores_gemma":[0.99786514,0.000120765384,0.001500751,0.00007139524,0.000020998388,0.0000067534465,0.00021626626,0.000013525952,0.00018442835],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9996051,0.00006717164,0.0000329768,0.00016373153,0.000087628054,0.00004337266],"domain_scores_gemma":[0.9926508,0.003784893,0.0022105782,0.00041387274,0.00041622153,0.0005236931],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007319165,0.00018153404,0.0003310166,0.00061752135,0.00061470096,0.0005631056,0.00056520564,0.0005337967,0.0020737792],"category_scores_gemma":[0.004620575,0.00014252723,0.00016778675,0.0008345529,0.0017845315,0.00088490185,0.000694019,0.0007709273,0.00021064714],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.004436661,0.00021281281,0.4287664,0.0014240429,0.00031156858,0.0021600164,0.0015427361,0.027717665,0.3986066,0.04649592,0.0012853547,0.08704027],"study_design_scores_gemma":[0.00031821235,0.0017094901,0.48379397,0.000174289,0.00044059954,0.008996643,0.001484081,0.07580636,0.18465891,0.23195061,0.010420461,0.00024632708],"about_ca_topic_score_codex":0.00062147365,"about_ca_topic_score_gemma":0.00052306126,"teacher_disagreement_score":0.0020737792,"about_ca_system_score_codex":0.0004279212,"about_ca_system_score_gemma":0.00028170782,"threshold_uncertainty_score":0.006937444},"labels":[],"label_agreement":null},{"id":"W2134773344","doi":"10.1093/bioinformatics/btn385","title":"An integrated approach for automating validation of extracted ion chromatographic peaks","year":2008,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Advanced Proteomics Techniques and Applications","field":"Chemistry","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Office of Experimental Program to Stimulate Competitive Research; National Center for Research Resources; National Institutes of Health; Canadian Institute for Theoretical Astrophysics","keywords":"Computer science; Data mining; Software; Chromatography; Chemistry; Programming language","score_opus":0.023520908674632343,"score_gpt":0.27449157440997285,"score_spread":0.2509706657353405,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2134773344","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.006847557,0.00018173011,0.9794037,0.000103181475,0.000080889025,0.00021995217,0.0005494291,0.012164064,0.00044942295],"genre_scores_gemma":[0.01771493,0.00008896882,0.97875345,0.00012930797,0.000027026548,0.00030072546,0.0015094779,0.00080519874,0.00067096984],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9926528,0.00096623634,0.00079527055,0.001523062,0.0037964399,0.00026611262],"domain_scores_gemma":[0.98371994,0.003949617,0.0011286101,0.0026991758,0.008081531,0.00042110766],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008882326,0.0019270821,0.0013501542,0.004414927,0.001096531,0.0027007202,0.003199527,0.0016354729,0.0050739674],"category_scores_gemma":[0.01730298,0.0012262868,0.0013640053,0.0024265102,0.0006734754,0.0019263393,0.0027307267,0.0026513617,0.0050536576],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005990603,0.00045619856,0.0060620806,0.0005124812,0.00032604721,0.0002144206,0.00018971466,0.0056789205,0.559783,0.0014922126,0.009249381,0.41543657],"study_design_scores_gemma":[0.00011414858,0.0003238977,0.0075193215,0.000069906644,0.00026908508,0.0010447621,0.00005394224,0.11437701,0.8515468,0.0022589017,0.02218343,0.00023882286],"about_ca_topic_score_codex":0.0017760855,"about_ca_topic_score_gemma":0.0031224801,"teacher_disagreement_score":0.008882326,"about_ca_system_score_codex":0.0008941219,"about_ca_system_score_gemma":0.0023814712,"threshold_uncertainty_score":0.04697478},"labels":[],"label_agreement":null},{"id":"W2134880148","doi":"10.1093/bioinformatics/btk028","title":"Inferring global levels of alternative splicing isoforms using a generative model of microarray data","year":2006,"lang":"en","type":"article","venue":"Bioinformatics","topic":"RNA Research and Splicing","field":"Biochemistry, Genetics and Molecular Biology","cited_by":61,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of Toronto","funders":"","keywords":"Alternative splicing; DNA microarray; Exon; Microarray; Microarray analysis techniques; Computational biology; Context (archaeology); RNA splicing; Biology; Exon skipping; Gene chip analysis; Gene; Genetics; Gene expression; RNA","score_opus":0.06752830136639101,"score_gpt":0.32512590210927544,"score_spread":0.2575976007428844,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2134880148","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.080675706,0.00011230444,0.9176581,0.00015238892,0.0000063303173,0.000030304176,0.00037820268,0.0007183181,0.00026842393],"genre_scores_gemma":[0.74389213,0.00032914084,0.25167048,0.00017317513,0.000048161743,0.00019589375,0.002359408,0.00016370002,0.001167961],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9988926,0.00033964592,0.000048675534,0.00041179784,0.00022112163,0.00008613766],"domain_scores_gemma":[0.99471843,0.004281436,0.0003467219,0.00036871346,0.0001988705,0.00008581961],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0031577742,0.0009864282,0.0011008689,0.0012565958,0.000335392,0.001126185,0.0013250969,0.0010210979,0.0007585095],"category_scores_gemma":[0.0059526577,0.00091996475,0.0013456611,0.00109228,0.0011510042,0.0011962639,0.0009597638,0.0015051794,0.00040220158],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00030647425,0.00013140941,0.02821321,0.000116490686,0.00018447604,0.00023882858,0.00018102083,0.87914455,0.022481931,0.011091943,0.00084117346,0.057068516],"study_design_scores_gemma":[0.000008258994,0.000016074706,0.0016866997,0.000003869193,0.000014793495,0.00006232109,0.000008342255,0.98546845,0.0025323187,0.010016776,0.00017198398,0.000010157341],"about_ca_topic_score_codex":0.0025903275,"about_ca_topic_score_gemma":0.0041798595,"teacher_disagreement_score":0.0031577742,"about_ca_system_score_codex":0.001098763,"about_ca_system_score_gemma":0.0007271875,"threshold_uncertainty_score":0.016700149},"labels":[],"label_agreement":null},{"id":"W2135061646","doi":"10.1093/bioinformatics/bts188","title":"NanoStringNorm: an extensible R package for the pre-processing of NanoString mRNA and miRNA data","year":2012,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Molecular Biology Techniques and Applications","field":"Biochemistry, Genetics and Molecular Biology","cited_by":271,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Princess Margaret Cancer Centre; University Health Network; University of Toronto; Ontario Institute for Cancer Research","funders":"","keywords":"Computer science; R package; Sample (material); Extensibility; Process (computing); Data mining; Data extraction; Computational science; Programming language","score_opus":0.03881180580201618,"score_gpt":0.31692176082460377,"score_spread":0.2781099550225876,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2135061646","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0059551983,0.00042235566,0.62560064,0.0005384845,0.00034829145,0.000699474,0.0716966,0.29179907,0.002939842],"genre_scores_gemma":[0.037044503,0.0005894147,0.75072235,0.00093712617,0.00014818681,0.0052590454,0.08532991,0.114479445,0.00549005],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.996473,0.0012837915,0.00046169024,0.00066798343,0.000891212,0.00022225802],"domain_scores_gemma":[0.9897503,0.0062972526,0.0009825391,0.0014549065,0.0011910945,0.00032402063],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008235097,0.003047858,0.0020670788,0.0029806276,0.0006829085,0.0017879835,0.00405127,0.00093548297,0.030470077],"category_scores_gemma":[0.024888212,0.0016105988,0.0029257846,0.0020163648,0.0010081067,0.0017661965,0.0028080575,0.0030144334,0.023862014],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0022429998,0.00025872522,0.015170598,0.0050799814,0.0023197,0.0015419325,0.00069227006,0.054360442,0.038115963,0.018007085,0.65390515,0.20830512],"study_design_scores_gemma":[0.0010120573,0.00052820164,0.01664,0.0005823177,0.0008299784,0.0019670296,0.00015655957,0.26594728,0.10133087,0.04860985,0.56167966,0.00071626157],"about_ca_topic_score_codex":0.0023097873,"about_ca_topic_score_gemma":0.003075465,"teacher_disagreement_score":0.030470077,"about_ca_system_score_codex":0.000815762,"about_ca_system_score_gemma":0.0030172493,"threshold_uncertainty_score":0.101932526},"labels":[],"label_agreement":null},{"id":"W2135338058","doi":"10.1093/bioinformatics/bts154","title":"Bayesian integration of networks without gold standards","year":2012,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":14,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"Biotechnology and Biological Sciences Research Council; Directorate for Biological Sciences","keywords":"Computer science; Bayesian network; Nexus (standard); Data mining; Data integration; Bayesian probability; Reference data; Gold standard (test); Machine learning; Artificial intelligence; Statistics; Mathematics","score_opus":0.008798978695646387,"score_gpt":0.2461073266602108,"score_spread":0.2373083479645644,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2135338058","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.007126176,0.0005977734,0.9872279,0.00025077665,0.000048596085,0.000120461395,0.0011753985,0.0017132148,0.0017396963],"genre_scores_gemma":[0.18789537,0.0009422018,0.7926225,0.00040758797,0.0001908511,0.0006481518,0.012569736,0.0007254319,0.003998183],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9871207,0.0046174363,0.00082945346,0.0037416916,0.0031670218,0.00052360044],"domain_scores_gemma":[0.97582597,0.014524836,0.0017506608,0.003638131,0.003834714,0.0004256827],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.017971342,0.0021893554,0.0028065238,0.007232414,0.0016064522,0.005438267,0.0048448993,0.0027163776,0.0051002707],"category_scores_gemma":[0.05893977,0.0018833885,0.0020225297,0.007033772,0.0020118237,0.0070120944,0.005525645,0.0038409084,0.002860277],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005351325,0.00019216382,0.011940845,0.00084318884,0.0005938229,0.0003618761,0.00037902812,0.48848298,0.0036249852,0.09594413,0.014883335,0.38221848],"study_design_scores_gemma":[0.000042263167,0.00003725399,0.0019154168,0.00013266287,0.00008919876,0.00015896666,0.000049112983,0.79399896,0.0041271136,0.19043587,0.0089649,0.000048310812],"about_ca_topic_score_codex":0.013385391,"about_ca_topic_score_gemma":0.013618972,"teacher_disagreement_score":0.017971342,"about_ca_system_score_codex":0.0033996645,"about_ca_system_score_gemma":0.003040336,"threshold_uncertainty_score":0.095042646},"labels":[],"label_agreement":null},{"id":"W2135511965","doi":"10.1093/bioinformatics/btq023","title":"rMAT - an R/Bioconductor package for analyzing ChIP-chip experiments","year":2010,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":29,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Montreal Clinical Research Institute","funders":"","keywords":"Bioconductor; Tiling array; Chromatin immunoprecipitation; DNA microarray; Computer science; Chip; Mac OS; R package; ChIP-on-chip; Histone; Computational biology; Chromatin; ChIP-sequencing; DNA; Biology; Software; Genetics; Computational science; Gene; Operating system; Nucleosome; Gene expression","score_opus":0.01463581004984009,"score_gpt":0.2662241722125298,"score_spread":0.2515883621626897,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2135511965","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0023935451,0.0018841609,0.4178253,0.0011168363,0.001064631,0.0006362613,0.2040129,0.36202946,0.009036898],"genre_scores_gemma":[0.016554613,0.0013354588,0.63733554,0.001369442,0.0004116126,0.0067808297,0.182664,0.14098571,0.012562818],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99391127,0.0020079156,0.0006487558,0.0016661849,0.0013416226,0.00042432902],"domain_scores_gemma":[0.9883493,0.006050176,0.0009937383,0.0021655317,0.0020318034,0.00040942538],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008397196,0.005980761,0.0072529074,0.0061594304,0.0016104693,0.0035994207,0.010631164,0.0019883143,0.12163061],"category_scores_gemma":[0.023794167,0.0028978018,0.0044553936,0.007970448,0.0016303057,0.0031740374,0.004110841,0.006302898,0.14267872],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00082175154,0.000112996924,0.0019562524,0.006683795,0.001972295,0.0003596538,0.00033715868,0.006142284,0.0069049867,0.012948802,0.9027015,0.05905853],"study_design_scores_gemma":[0.0010815345,0.00024880085,0.010993655,0.0011719139,0.0021141355,0.0011741279,0.00015880958,0.072893396,0.02205417,0.0756196,0.8120303,0.0004595167],"about_ca_topic_score_codex":0.004225568,"about_ca_topic_score_gemma":0.0039249426,"teacher_disagreement_score":0.12163061,"about_ca_system_score_codex":0.0012578873,"about_ca_system_score_gemma":0.004542589,"threshold_uncertainty_score":0.40689492},"labels":[],"label_agreement":null},{"id":"W2135833397","doi":"10.1093/bioinformatics/16.11.1054","title":"BLAST Search Updater: a notification system for new database matches","year":2000,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Victoria","funders":"Medical Research Council; Natural Sciences and Engineering Research Council of Canada","keywords":"Unix; Computer science; Information retrieval; Database; Software; World Wide Web; Programming language","score_opus":0.02132106413948929,"score_gpt":0.24514800362597097,"score_spread":0.2238269394864817,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2135833397","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.005503046,0.0012733316,0.08666296,0.0035976388,0.0036981942,0.001217621,0.100566424,0.77309334,0.024387386],"genre_scores_gemma":[0.049334586,0.0021654854,0.24560802,0.006410253,0.0035948607,0.002809128,0.50081927,0.10221258,0.08704579],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9965922,0.0007246497,0.0005107436,0.0005540016,0.0012753312,0.00034300858],"domain_scores_gemma":[0.98875636,0.002626604,0.001090135,0.0025120757,0.0027861022,0.0022287415],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0065273093,0.0028548695,0.0029488197,0.006776318,0.00230534,0.0030164893,0.005470092,0.004460111,0.18043298],"category_scores_gemma":[0.018640114,0.0021038323,0.00097300415,0.0050369813,0.00057511113,0.0042059426,0.004874622,0.0038950597,0.19351786],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0012001602,0.00009782689,0.0010260475,0.00061548286,0.000030597388,0.00039304933,0.00013085654,0.0001320874,0.008470569,0.0008010944,0.9191793,0.06792292],"study_design_scores_gemma":[0.00072842935,0.00040483061,0.0057454854,0.0004945727,0.00010970398,0.0014814524,0.00016189681,0.012164746,0.032850616,0.0037324743,0.9417709,0.00035481298],"about_ca_topic_score_codex":0.0017147241,"about_ca_topic_score_gemma":0.0020551074,"teacher_disagreement_score":0.18043298,"about_ca_system_score_codex":0.0010691059,"about_ca_system_score_gemma":0.0016288412,"threshold_uncertainty_score":0.6036084},"labels":[],"label_agreement":null},{"id":"W2135884466","doi":"10.1093/bioinformatics/btm060","title":"Semantic Web Service provision: a realistic framework for Bioinformatics programmers","year":2007,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Scientific Computing and Data Management","field":"Decision Sciences","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"Genome Alberta; Genome Canada","keywords":"Computer science; Java; Interoperability; Web service; World Wide Web; Documentation; Software engineering; Semantic Web; Software deployment; Service (business); Programming language","score_opus":0.12229443389299481,"score_gpt":0.3935934427364254,"score_spread":0.27129900884343056,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2135884466","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.004203466,0.00046149158,0.9613832,0.014919837,0.000400022,0.00026693396,0.00026398117,0.005704329,0.01239679],"genre_scores_gemma":[0.054281276,0.00075508194,0.93034744,0.0023903665,0.00033631717,0.0006823907,0.0009960383,0.0017645698,0.008446442],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9903736,0.0048053674,0.00089377817,0.0010889918,0.002022786,0.0008155735],"domain_scores_gemma":[0.9899691,0.0035645408,0.00044835938,0.002710216,0.0018080327,0.001499694],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.022626186,0.0011924828,0.00094036973,0.0018648622,0.0047499514,0.0096435975,0.005178983,0.0065934225,0.007850124],"category_scores_gemma":[0.016278056,0.0014279288,0.0021982172,0.0025228674,0.007108894,0.016042119,0.0068893307,0.0063119116,0.008228954],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001537537,0.00015927483,0.00093844446,0.00035921836,0.0000368903,0.0005369378,0.0022762765,0.006246797,0.0039317934,0.9047832,0.02833676,0.05224065],"study_design_scores_gemma":[0.00007800942,0.000087360415,0.00017937181,0.0003630771,0.000039670424,0.00066903187,0.0008628654,0.047988646,0.0030194898,0.4612063,0.4854297,0.00007635787],"about_ca_topic_score_codex":0.0036737008,"about_ca_topic_score_gemma":0.0026481485,"teacher_disagreement_score":0.022626186,"about_ca_system_score_codex":0.0033751687,"about_ca_system_score_gemma":0.007803547,"threshold_uncertainty_score":0.11966014},"labels":[],"label_agreement":null},{"id":"W2136104033","doi":"10.1093/bioinformatics/btm225","title":"Homology search for genes","year":2007,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":18,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"U.S. Public Health Service","keywords":"Gene; Homology (biology); Gene prediction; Genome; Annotation; Gene Annotation; Exon; Genetics; Computational biology; Biology; Computer science; Human genome; Ensembl; Genomics","score_opus":0.01585314769633712,"score_gpt":0.2663310844368851,"score_spread":0.25047793674054797,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2136104033","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.19243525,0.00823915,0.7452167,0.0022905292,0.00025060127,0.00050369254,0.018687941,0.010640914,0.02173518],"genre_scores_gemma":[0.4971395,0.0026862072,0.43086937,0.00083343644,0.00021170266,0.0004922262,0.05803893,0.00086499937,0.008863504],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99850583,0.00029331344,0.000111727124,0.0006379509,0.00034629923,0.00010486747],"domain_scores_gemma":[0.997081,0.0017219025,0.0002829596,0.00038959912,0.00040978126,0.00011463193],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013210286,0.0007899179,0.0011864009,0.0022763438,0.00081620476,0.001209551,0.0015341786,0.0011587357,0.01371421],"category_scores_gemma":[0.00616682,0.0003926909,0.0009271177,0.0034254817,0.00067270274,0.0021369988,0.0014728282,0.0008865842,0.0071679717],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0021299121,0.000661975,0.03548399,0.0043519293,0.00052273576,0.002935428,0.00094728364,0.040757153,0.19687344,0.087450966,0.04759019,0.58029497],"study_design_scores_gemma":[0.0006627296,0.0013493607,0.028624717,0.00060659245,0.0006272838,0.0102016255,0.0010385296,0.29975802,0.22748792,0.21633308,0.21312372,0.00018644398],"about_ca_topic_score_codex":0.00083422335,"about_ca_topic_score_gemma":0.0008311135,"teacher_disagreement_score":0.01371421,"about_ca_system_score_codex":0.0008107925,"about_ca_system_score_gemma":0.0008313091,"threshold_uncertainty_score":0.04587865},"labels":[],"label_agreement":null},{"id":"W2136180457","doi":"10.1093/bioinformatics/btr023","title":"PhyloPro: a web-based tool for the generation and visualization of phylogenetic profiles across Eukarya","year":2011,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":12,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"SickKids Foundation; Hospital for Sick Children; University of Toronto","funders":"Canadian Institutes of Health Research; Hospital for Sick Children","keywords":"Phylogenetic tree; Visualization; Computer science; Web application; World Wide Web; Data science; Computational biology; Data mining; Biology; Genetics","score_opus":0.029211893887241266,"score_gpt":0.2650650657588872,"score_spread":0.23585317187164595,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2136180457","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.006638151,0.00091286807,0.5947519,0.00041345382,0.00022262275,0.00021371385,0.050744776,0.34235325,0.0037492837],"genre_scores_gemma":[0.04986229,0.0018298855,0.77851063,0.00059550855,0.00013267889,0.0014008031,0.11305029,0.048812896,0.0058049443],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99932003,0.00014161797,0.00007940019,0.0001564775,0.00023384042,0.000068618676],"domain_scores_gemma":[0.9980405,0.0009962269,0.0002322592,0.00025001136,0.00021347661,0.00026757357],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002084092,0.0022221538,0.0015364498,0.0059069623,0.0013352978,0.002073124,0.0024238613,0.0015130874,0.045809213],"category_scores_gemma":[0.0043371855,0.0016102485,0.0015232952,0.0032728408,0.00075438013,0.0029865636,0.0034565888,0.0029804404,0.0168046],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0025202287,0.00034773847,0.006536726,0.0069180797,0.00062184973,0.002562271,0.0016459536,0.007225498,0.13328439,0.013821203,0.4374799,0.38703614],"study_design_scores_gemma":[0.0009856262,0.0003549656,0.015433044,0.0012608431,0.00032504115,0.0050419755,0.0004933838,0.08330774,0.09089034,0.032461476,0.76885617,0.00058948074],"about_ca_topic_score_codex":0.0015075936,"about_ca_topic_score_gemma":0.0018724396,"teacher_disagreement_score":0.045809213,"about_ca_system_score_codex":0.0005304587,"about_ca_system_score_gemma":0.0011422689,"threshold_uncertainty_score":0.15324706},"labels":[],"label_agreement":null},{"id":"W2136207878","doi":"10.1093/bioinformatics/btu595","title":"BigDataScript: a scripting language for data pipelines","year":2014,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":39,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University and Génome Québec Innovation Centre","funders":"National Institute of Diabetes and Digestive and Kidney Diseases; Canadian Institutes of Health Research; National Institutes of Health; Natural Sciences and Engineering Research Council of Canada; McGill University","keywords":"Computer science; Pipeline transport; Software portability; Serialization; Scripting language; Programming language; Debugging; Robustness (evolution); Python (programming language); Operating system","score_opus":0.03272703732389456,"score_gpt":0.27702773098020317,"score_spread":0.2443006936563086,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2136207878","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0017293158,0.0001129804,0.79052806,0.000325912,0.00012880405,0.00031258859,0.0050761052,0.19961683,0.0021693264],"genre_scores_gemma":[0.035973992,0.00038479315,0.85286695,0.0012596125,0.00014109377,0.00140988,0.020828951,0.08168312,0.0054516364],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9967451,0.0005960265,0.00065666606,0.00085083756,0.00093805656,0.00021316376],"domain_scores_gemma":[0.9910945,0.0038608806,0.0008044061,0.0018806938,0.0016090263,0.0007505745],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0058256737,0.0022045162,0.00088241964,0.0010563526,0.00085535244,0.0024658754,0.00393031,0.0012879733,0.011532073],"category_scores_gemma":[0.013771163,0.0021875072,0.0017443956,0.0013794341,0.0022692806,0.0041765203,0.0035998758,0.0048416452,0.0128729725],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0030424695,0.00046550337,0.0094590355,0.0039983164,0.00029263957,0.0017981258,0.002539568,0.028704021,0.1155103,0.07432544,0.44290227,0.3169623],"study_design_scores_gemma":[0.0005479055,0.00025838433,0.0028402675,0.0005745577,0.00010465139,0.0015789665,0.00016961023,0.18689384,0.16936,0.094956,0.5423629,0.00035293365],"about_ca_topic_score_codex":0.0013441548,"about_ca_topic_score_gemma":0.0011498935,"teacher_disagreement_score":0.011532073,"about_ca_system_score_codex":0.0008496265,"about_ca_system_score_gemma":0.0031699287,"threshold_uncertainty_score":0.03857863},"labels":[],"label_agreement":null},{"id":"W2136265964","doi":"10.1093/bioinformatics/btt613","title":"Mouse model phenotypes provide information about human drug targets","year":2013,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Biomedical Text Mining and Ontologies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":49,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Carleton University","funders":"","keywords":"Drug repositioning; Phenotype; Computational biology; Drug; Identification (biology); Similarity (geometry); Drug development; Drug discovery; Biology; Phenotypic screening; Repurposing; Computer science; Bioinformatics; Genetics; Pharmacology; Artificial intelligence; Gene","score_opus":0.010892537661136478,"score_gpt":0.24047240124778707,"score_spread":0.22957986358665058,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2136265964","genre_codex":"dataset","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.039561134,0.0031560871,0.26067916,0.0034572892,0.00016372619,0.00059102016,0.6182119,0.050079953,0.024099655],"genre_scores_gemma":[0.20537248,0.005850348,0.27834496,0.0017370422,0.000207575,0.0011495829,0.49585822,0.0063005704,0.00517924],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99843735,0.00039108368,0.00024379372,0.0003503063,0.0005283123,0.000049106686],"domain_scores_gemma":[0.9889031,0.0061430554,0.0016440115,0.0019547227,0.0009799794,0.00037503408],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0028329513,0.001700542,0.0009246326,0.005006981,0.00034655991,0.0013435956,0.0016593949,0.0010476122,0.018533397],"category_scores_gemma":[0.013982685,0.0005717518,0.00096865447,0.0048229424,0.0004931373,0.0024270813,0.0017138877,0.0010875122,0.008899318],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0025570479,0.00083093444,0.07391763,0.009130728,0.0008682726,0.0020140156,0.00069943175,0.023827381,0.054499194,0.047509477,0.35210028,0.43204567],"study_design_scores_gemma":[0.0006044877,0.0005851174,0.08817182,0.0019092538,0.0012175185,0.005553112,0.00041117377,0.050856233,0.057897642,0.12926872,0.6631454,0.00037951383],"about_ca_topic_score_codex":0.0015508182,"about_ca_topic_score_gemma":0.0018617361,"teacher_disagreement_score":0.018533397,"about_ca_system_score_codex":0.00052935863,"about_ca_system_score_gemma":0.0010988032,"threshold_uncertainty_score":0.062000453},"labels":[],"label_agreement":null},{"id":"W2136359150","doi":"10.1093/bioinformatics/btp610","title":"Methods for combining peptide intensities to estimate relative protein abundance","year":2009,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Advanced Proteomics Techniques and Applications","field":"Chemistry","cited_by":64,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University and Génome Québec Innovation Centre","funders":"Canadian Institutes of Health Research; Genome Canada; Natural Sciences and Engineering Research Council of Canada; McGill University","keywords":"Intensity (physics); Peptide; Mass spectrometry; Abundance (ecology); Biological system; Detector; Chemistry; Algorithm; Analytical Chemistry (journal); Mathematics; Physics; Biology; Chromatography; Optics","score_opus":0.025978251458796814,"score_gpt":0.37441554503384433,"score_spread":0.3484372935750475,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2136359150","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0026308594,0.00026395358,0.99542266,0.000056772282,0.000048225178,0.00010266367,0.000053986925,0.0011969593,0.00022392371],"genre_scores_gemma":[0.014760224,0.00021500138,0.9839372,0.00005767588,0.000057781413,0.00023653469,0.00019249914,0.00025266327,0.00029046883],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.991284,0.0022878353,0.0008766527,0.0017113963,0.0036427637,0.00019736381],"domain_scores_gemma":[0.9742819,0.01312776,0.003124704,0.0022786101,0.00691661,0.00027039056],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.014625288,0.004134716,0.0019946243,0.0070852796,0.0010157189,0.0030818633,0.003703025,0.002047205,0.0027269672],"category_scores_gemma":[0.045278594,0.0015258343,0.0021547824,0.005238178,0.0012959822,0.0037066122,0.0022206455,0.0030411282,0.0031630758],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00038777408,0.00030211112,0.0067329053,0.00074654474,0.00065595895,0.00012872714,0.00024967833,0.08408881,0.04916426,0.007976373,0.0046606925,0.8449061],"study_design_scores_gemma":[0.00015672794,0.0002597854,0.0048553627,0.00015101212,0.0003127382,0.0008350397,0.00007996052,0.8622197,0.08622075,0.03157552,0.01303935,0.0002940816],"about_ca_topic_score_codex":0.0012312575,"about_ca_topic_score_gemma":0.0011753864,"teacher_disagreement_score":0.014625288,"about_ca_system_score_codex":0.00088420603,"about_ca_system_score_gemma":0.0012291243,"threshold_uncertainty_score":0.0773468},"labels":[],"label_agreement":null},{"id":"W2136499634","doi":"10.1093/bioinformatics/bts572","title":"FeatureStack: Perl module for comparative visualization of gene features","year":2012,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetics, Bioinformatics, and Biomedical Research","field":"Biochemistry, Genetics and Molecular Biology","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"Natural Sciences and Engineering Research Council of Canada; SFU Community Trust Endowment Fund; Simon Fraser University; Michael Smith Health Research BC; Canadian Institutes of Health Research; Western Canada Research Grid","keywords":"Perl; Computer science; Visualization; Programming language; Computational biology; Biology; Artificial intelligence","score_opus":0.03286352512112129,"score_gpt":0.3336641445107511,"score_spread":0.3008006193896298,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2136499634","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0019125629,0.00030536723,0.18302529,0.0002457432,0.00025896722,0.0002569381,0.17176418,0.63658303,0.0056479475],"genre_scores_gemma":[0.027356377,0.0005037002,0.262923,0.0009267293,0.0001415954,0.0031433583,0.36988488,0.3219617,0.013158656],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9984675,0.00021724502,0.00017101526,0.00042288753,0.0005056028,0.00021566954],"domain_scores_gemma":[0.9979691,0.0007991994,0.00020894728,0.00040993706,0.00042821735,0.0001847028],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0023641346,0.004176447,0.0025913848,0.004143309,0.0011954005,0.0032782913,0.004902139,0.0016693526,0.20418593],"category_scores_gemma":[0.004777065,0.002473736,0.0021716107,0.0027701384,0.00063353556,0.003725296,0.0023559805,0.003646408,0.1046707],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010112933,0.000108230735,0.0010816671,0.002955415,0.00020252279,0.00041260937,0.00046395828,0.001422897,0.036843523,0.0044946685,0.8655035,0.08549971],"study_design_scores_gemma":[0.00056389085,0.00014308433,0.005544192,0.00059054076,0.00014611054,0.0009906447,0.00022197989,0.030513115,0.12567021,0.017914115,0.8172754,0.00042681178],"about_ca_topic_score_codex":0.0028083592,"about_ca_topic_score_gemma":0.0028358137,"teacher_disagreement_score":0.20418593,"about_ca_system_score_codex":0.0018042447,"about_ca_system_score_gemma":0.0018309576,"threshold_uncertainty_score":0.68307},"labels":[],"label_agreement":null},{"id":"W2137112190","doi":"10.1093/bioinformatics/btm004","title":"SGN Sim, a Stochastic Genetic Networks Simulator","year":2007,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene Regulatory Network Analysis","field":"Biochemistry, Genetics and Molecular Biology","cited_by":65,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"","keywords":"Computer science; Set (abstract data type); Gene regulatory network; Stochastic simulation; Master regulator; Translation (biology); Stochastic modelling; Stochastic process; Simulation; Algorithm; Theoretical computer science; Gene; Transcription factor; Genetics; Mathematics; Gene expression; Programming language; Biology; Messenger RNA","score_opus":0.0066315848812003135,"score_gpt":0.2278340997158765,"score_spread":0.2212025148346762,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2137112190","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.055402055,0.0005808136,0.8550736,0.0009914485,0.0004078619,0.00026486273,0.008730149,0.049449615,0.029099582],"genre_scores_gemma":[0.48996767,0.00080273947,0.4793264,0.00048551313,0.000087096865,0.0010619267,0.010324965,0.0048400653,0.01310375],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99971265,0.00009701333,0.00001622643,0.00004289299,0.000099839184,0.000031364903],"domain_scores_gemma":[0.99871576,0.00076313445,0.00007811471,0.00011572515,0.00022301397,0.000104247294],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000809639,0.0007452417,0.00057492673,0.0004891679,0.0004427322,0.00066387065,0.0019311209,0.0009419919,0.012423539],"category_scores_gemma":[0.00293424,0.00037211674,0.0006925856,0.0006873973,0.00045129406,0.0007594122,0.00068179495,0.000985796,0.0020449406],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00016759062,0.000079284335,0.0018621626,0.00019511025,0.000060543236,0.00010533351,0.00006133213,0.9421071,0.0022540465,0.018577857,0.013816923,0.020712698],"study_design_scores_gemma":[0.000025473017,0.00001162859,0.000052598713,0.0000051352854,0.0000049700793,0.000013161759,0.0000030294686,0.9907047,0.0007435413,0.003156791,0.005274453,0.000004556784],"about_ca_topic_score_codex":0.006077577,"about_ca_topic_score_gemma":0.0053203395,"teacher_disagreement_score":0.012423539,"about_ca_system_score_codex":0.0008881345,"about_ca_system_score_gemma":0.0013208707,"threshold_uncertainty_score":0.04156089},"labels":[],"label_agreement":null},{"id":"W2138003801","doi":"10.1093/bioinformatics/bth351","title":"Protein complex prediction via cost-based clustering","year":2004,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":668,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ontario Institute for Cancer Research; University of Toronto","funders":"National Institute of General Medical Sciences; National Institutes of Health; University of Toronto","keywords":"Cluster analysis; Computer science; Scalability; Data mining; Partition (number theory); Identification (biology); Biological network; Protein function prediction; Caenorhabditis elegans; Computational biology; Protein function; Machine learning; Biology; Mathematics; Genetics","score_opus":0.015073963754215663,"score_gpt":0.22761972047215107,"score_spread":0.2125457567179354,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2138003801","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.11059562,0.00037677767,0.88469505,0.00017797631,0.00001533091,0.00013471897,0.00063164724,0.0016948447,0.0016779555],"genre_scores_gemma":[0.54058737,0.00024504587,0.4550457,0.00005427165,0.000023798442,0.00016505174,0.0022310086,0.0002029246,0.0014448604],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9992943,0.00014151698,0.00004410314,0.00017607874,0.0003010722,0.000042946816],"domain_scores_gemma":[0.9978301,0.00084331405,0.0002770473,0.00025302763,0.00071571063,0.00008085097],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006912088,0.00067943387,0.00072346145,0.0033700576,0.00048680961,0.000934415,0.0010625032,0.0008368455,0.0013178084],"category_scores_gemma":[0.0048491126,0.0002929804,0.0005140748,0.0018994631,0.00053438486,0.0010371498,0.0005828492,0.00048332123,0.00059902493],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00026285264,0.000108505956,0.011087108,0.0002625625,0.00015164132,0.00013209876,0.00008148745,0.7944101,0.016875096,0.010313916,0.00455302,0.16176158],"study_design_scores_gemma":[0.000005085113,0.000011255262,0.0010481243,0.0000034250581,0.000008017572,0.000034353416,0.0000067609703,0.9922506,0.0031708016,0.0030163864,0.00043782208,0.0000072900643],"about_ca_topic_score_codex":0.009970807,"about_ca_topic_score_gemma":0.0074321004,"teacher_disagreement_score":0.009970807,"about_ca_system_score_codex":0.0012498737,"about_ca_system_score_gemma":0.0009410035,"threshold_uncertainty_score":0.019825578},"labels":[],"label_agreement":null},{"id":"W2138044641","doi":"10.1093/bioinformatics/btn395","title":"Site-specific evolutionary rates in proteins are better modeled as non-independent and strictly relative","year":2008,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Protein Structure and Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"National Institute of General Medical Sciences; Natural Sciences and Engineering Research Council of Canada; National Institutes of Health; Canadian Institutes of Health Research; North Carolina State University","keywords":"Bayes' theorem; Prior probability; A priori and a posteriori; Bayesian probability; Statistics; Posterior probability; Mathematics; Maximum a posteriori estimation; Bayes factor; Sequence (biology); Computer science; Algorithm; Biology; Maximum likelihood; Genetics","score_opus":0.008864132810185162,"score_gpt":0.21585627109962052,"score_spread":0.20699213828943536,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2138044641","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.16938154,0.0013255744,0.8242954,0.0006574964,0.000084792475,0.00006758608,0.00043486693,0.00055864174,0.0031940325],"genre_scores_gemma":[0.89387953,0.0012190036,0.10080066,0.00031350373,0.000104287436,0.0001739342,0.00074917375,0.000361382,0.002398482],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99687093,0.001228203,0.00021140455,0.0011678933,0.00037519285,0.00014637673],"domain_scores_gemma":[0.9883292,0.0069658603,0.0022241294,0.0018661147,0.00040875262,0.00020590531],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005935128,0.0015984055,0.0011138273,0.0014200689,0.0006541724,0.0029981863,0.0022234926,0.002312448,0.0034671286],"category_scores_gemma":[0.024797963,0.0009802133,0.0014535431,0.0022620833,0.0022863431,0.0067075063,0.0010738872,0.0029648226,0.0012253989],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00042612897,0.0001757296,0.047707323,0.0005082829,0.00035701974,0.0003148126,0.0008744554,0.64263076,0.019697165,0.2347257,0.0014094446,0.05117326],"study_design_scores_gemma":[0.000040299525,0.0001501002,0.012028342,0.00013534867,0.000086155436,0.00057111605,0.00011441269,0.75060743,0.005683838,0.22453882,0.005913081,0.00013099203],"about_ca_topic_score_codex":0.0023548498,"about_ca_topic_score_gemma":0.0021414154,"teacher_disagreement_score":0.005935128,"about_ca_system_score_codex":0.0013981196,"about_ca_system_score_gemma":0.000776724,"threshold_uncertainty_score":0.031388342},"labels":[],"label_agreement":null},{"id":"W2138334294","doi":"10.1093/bioinformatics/18.11.1544","title":"Poxvirus Orthologous Clusters (POCs)","year":2002,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":66,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Victoria","funders":"National Institute of Allergy and Infectious Diseases; Natural Sciences and Engineering Research Council of Canada; National Institutes of Health","keywords":"Genome; Computer science; Gene; Computational biology; Interface (matter); Biology; Java; Genetics; Programming language; Operating system","score_opus":0.016935666276901438,"score_gpt":0.21779803493254102,"score_spread":0.20086236865563958,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2138334294","genre_codex":"dataset","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.1085027,0.0038585376,0.17251909,0.0012320178,0.001271199,0.0013439335,0.43670106,0.21394548,0.06062598],"genre_scores_gemma":[0.14554021,0.0015715496,0.21326362,0.000364093,0.00017567827,0.0012998532,0.5981453,0.01935096,0.02028874],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9994066,0.000050904386,0.00004577562,0.00023095544,0.00019425135,0.000071594295],"domain_scores_gemma":[0.99939704,0.00014637935,0.0001116868,0.00010976446,0.000109699795,0.00012545244],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008012516,0.0010349236,0.00093908276,0.0024512971,0.0013672172,0.0016795509,0.0015488083,0.00097071746,0.04797145],"category_scores_gemma":[0.0019865884,0.0010027512,0.0009168015,0.0030591746,0.00033386456,0.0014046156,0.0017812458,0.0012141585,0.022132615],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0027743182,0.0002713707,0.013292553,0.005084812,0.00025959883,0.0011919399,0.00086558535,0.0017510778,0.13047394,0.013493185,0.6455118,0.18502985],"study_design_scores_gemma":[0.0004620116,0.00030862758,0.02397253,0.00047675657,0.0002293825,0.0022294577,0.0002939053,0.014724071,0.07496851,0.009418635,0.87276363,0.00015246167],"about_ca_topic_score_codex":0.0015293333,"about_ca_topic_score_gemma":0.0015203463,"teacher_disagreement_score":0.04797145,"about_ca_system_score_codex":0.00063289347,"about_ca_system_score_gemma":0.0012117814,"threshold_uncertainty_score":0.1604805},"labels":[],"label_agreement":null},{"id":"W2138379687","doi":"10.1093/bioinformatics/18.8.1034","title":"Improving gene recognition accuracy by combiningpredictions from two gene-finding programs","year":2002,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":50,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia; Children's & Women's Health Centre of British Columbia","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Exon; Correctness; Computer science; Gene prediction; Gene; Consistency (knowledge bases); Computational biology; Genetics; Data mining; Artificial intelligence; Biology; Algorithm; Genome","score_opus":0.01937588661777488,"score_gpt":0.22736097331829,"score_spread":0.20798508670051513,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2138379687","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.2428911,0.0007228842,0.6950913,0.000495422,0.00009412434,0.000128774,0.0019210641,0.056496933,0.002158356],"genre_scores_gemma":[0.29481634,0.00025269826,0.6966791,0.0001548908,0.00005412516,0.00012533419,0.004812656,0.001742131,0.0013627069],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9975656,0.0005318616,0.00024080323,0.00076757534,0.000703222,0.00019096231],"domain_scores_gemma":[0.9905702,0.006008807,0.00063294475,0.0013271989,0.0012217561,0.00023917003],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004135847,0.0019306733,0.0014912977,0.001686562,0.000610994,0.001418497,0.0016826125,0.0009226394,0.0030470018],"category_scores_gemma":[0.010705503,0.0009155902,0.0009957692,0.0020595589,0.00062075234,0.0023097596,0.001871872,0.001334141,0.0021631029],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0024064058,0.0004741034,0.07735962,0.00074316026,0.00057702645,0.000519069,0.00031324997,0.059325587,0.15631458,0.0016087749,0.008538095,0.6918203],"study_design_scores_gemma":[0.0001419414,0.0003302626,0.019060671,0.00005493663,0.00045704294,0.0011198616,0.00012612386,0.6457377,0.31850484,0.004285444,0.010024197,0.00015700712],"about_ca_topic_score_codex":0.0024461383,"about_ca_topic_score_gemma":0.0038370262,"teacher_disagreement_score":0.004135847,"about_ca_system_score_codex":0.00054848136,"about_ca_system_score_gemma":0.001125831,"threshold_uncertainty_score":0.0218727},"labels":[],"label_agreement":null},{"id":"W2138453085","doi":"10.1093/bioinformatics/bts430","title":"Gemma: a resource for the reuse, sharing and meta-analysis of expression profiling data","year":2012,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":158,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"National Institute of General Medical Sciences; Canadian Institutes of Health Research; National Institutes of Health; Michael Smith Health Research BC","keywords":"Computer science; License; Gemma; Profiling (computer programming); Plug-in; Software; World Wide Web; Data science; Database; Information retrieval; Biology; Operating system","score_opus":0.15434453868251755,"score_gpt":0.3402158842362292,"score_spread":0.18587134555371168,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2138453085","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.008264328,0.0023685421,0.42033514,0.001226072,0.00055969367,0.0005592767,0.12883465,0.43186188,0.0059904344],"genre_scores_gemma":[0.063783415,0.0019139193,0.5187754,0.0010202061,0.00036557845,0.0017315944,0.339159,0.0685764,0.0046744603],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99636275,0.0008385554,0.000407427,0.0010086776,0.0011254678,0.00025713624],"domain_scores_gemma":[0.9898178,0.0033086636,0.00090381707,0.003933143,0.00094929687,0.0010872673],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.010501526,0.0027131722,0.002839214,0.008761039,0.001396443,0.0041315444,0.006556887,0.0016881524,0.020882037],"category_scores_gemma":[0.0180162,0.0018298468,0.0026613595,0.007835286,0.00092681654,0.004561177,0.0063938466,0.0034311004,0.016680824],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0030689135,0.00040701087,0.013891736,0.0056552263,0.0032561633,0.0011268549,0.001309393,0.01162424,0.07371938,0.018661158,0.55077326,0.31650665],"study_design_scores_gemma":[0.0007222832,0.00028859536,0.01834659,0.00093233446,0.0016630053,0.0013295646,0.0003218986,0.08076994,0.07359286,0.04891517,0.77250886,0.00060896465],"about_ca_topic_score_codex":0.0032425805,"about_ca_topic_score_gemma":0.0035915985,"teacher_disagreement_score":0.99344313,"about_ca_system_score_codex":0.0012761543,"about_ca_system_score_gemma":0.0037898996,"threshold_uncertainty_score":0.06985736},"labels":[],"label_agreement":null},{"id":"W2138839185","doi":"10.1093/bioinformatics/btt770","title":"Enhanced flowType/RchyOptimyx: a Bioconductor pipeline for discovery in high-dimensional cytometry data","year":2014,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Single-cell and spatial transcriptomics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":36,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia; BC Cancer Agency","funders":"National Institute of Biomedical Imaging and Bioengineering; Terry Fox Research Institute; Canadian Institutes of Health Research; Terry Fox Foundation; National Institutes of Health; Michael Smith Health Research BC","keywords":"Bioconductor; Pipeline (software); Flow cytometry; Mass cytometry; Computer science; Throughput; Cytometry; Software; License; Computational biology; Data mining; Biology; Molecular biology; Operating system; Gene","score_opus":0.024644175101572536,"score_gpt":0.2594918384563199,"score_spread":0.23484766335474735,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2138839185","genre_codex":"software","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0023431436,0.00055456226,0.3794596,0.00092040846,0.0006305114,0.0007601155,0.07294943,0.5370906,0.0052915215],"genre_scores_gemma":[0.022341909,0.0008873081,0.61671335,0.0018529721,0.00024268642,0.0066971523,0.15105747,0.18281588,0.01739124],"study_design_codex":"not_applicable","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.996376,0.0006855665,0.00035541327,0.0009647867,0.0011962275,0.00042207498],"domain_scores_gemma":[0.99253416,0.0026628152,0.0005209161,0.0015928108,0.0021849538,0.00050424505],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0084995655,0.0049747885,0.0031508908,0.0041233925,0.0016285147,0.0047377897,0.009193624,0.0023520407,0.10730311],"category_scores_gemma":[0.016843887,0.0029866404,0.0033875047,0.0037621295,0.0013027269,0.003473002,0.0038201062,0.006487891,0.091667145],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001576372,0.0001359149,0.0017795336,0.0023708171,0.0007972487,0.00038182872,0.00036806375,0.0027672588,0.029860884,0.0094800815,0.87574524,0.07473668],"study_design_scores_gemma":[0.0018328412,0.0002712322,0.0073651075,0.00075831,0.000599078,0.00094805035,0.00013778215,0.1090602,0.14661768,0.040286783,0.6914209,0.0007019712],"about_ca_topic_score_codex":0.007380018,"about_ca_topic_score_gemma":0.008591279,"teacher_disagreement_score":0.10730311,"about_ca_system_score_codex":0.0029413442,"about_ca_system_score_gemma":0.0076840804,"threshold_uncertainty_score":0.35896468},"labels":[],"label_agreement":null},{"id":"W2139020869","doi":"10.1093/bioinformatics/btq111","title":"Mutual information is critically dependent on prior assumptions: would the <i>correct</i> estimate of mutual information please identify itself?","year":2010,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":32,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Mutual information; Computer science; Prior information; Critically ill; Information theory; Econometrics; Statistics; Artificial intelligence; Mathematics; Medicine; Intensive care medicine","score_opus":0.008144480487158157,"score_gpt":0.2617045430190412,"score_spread":0.25356006253188307,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2139020869","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.1193069,0.0031104253,0.8608538,0.006361924,0.00031280072,0.00010516394,0.0012143011,0.0008345594,0.007900043],"genre_scores_gemma":[0.88568705,0.0016592026,0.107819855,0.0016169414,0.0005785031,0.00022702706,0.001346674,0.00033023013,0.0007345827],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.98839545,0.0055521154,0.00084719126,0.002588397,0.0022430574,0.00037377764],"domain_scores_gemma":[0.8026752,0.16444455,0.013049432,0.0135436235,0.0050967345,0.0011905433],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.020208612,0.0010155273,0.0016599748,0.0024787607,0.0012287094,0.0048287115,0.002161678,0.0017659824,0.002510723],"category_scores_gemma":[0.16612378,0.0008208933,0.0009995059,0.0019686709,0.0069752014,0.008117506,0.0029817172,0.0041364403,0.0008015235],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013297104,0.0001762226,0.12792802,0.002939204,0.0013586911,0.0010035275,0.0019205089,0.16269577,0.018604109,0.42136142,0.017935213,0.24274762],"study_design_scores_gemma":[0.00004673842,0.00016802025,0.028593903,0.0004821026,0.0001638283,0.0011394519,0.00040766769,0.23465928,0.0136671895,0.7144459,0.0059219613,0.0003039203],"about_ca_topic_score_codex":0.001823907,"about_ca_topic_score_gemma":0.0012235304,"teacher_disagreement_score":0.020208612,"about_ca_system_score_codex":0.0021232653,"about_ca_system_score_gemma":0.0015828445,"threshold_uncertainty_score":0.106874645},"labels":[],"label_agreement":null},{"id":"W2139168653","doi":"10.1093/bioinformatics/btt312","title":"NGS++: a library for rapid prototyping of epigenomics software tools","year":2013,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Rare Diseases","field":"Biochemistry, Genetics and Molecular Biology","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université Laval; Université de Sherbrooke","funders":"Natural Sciences and Engineering Research Council of Canada; Canadian Institutes of Health Research; Ministère du Développement Économique, de l’Innovation et de l’Exportation","keywords":"Computer science; Documentation; Interface (matter); Software; Source code; Application programming interface; File format; Rapid prototyping; Software engineering; World Wide Web; Programming language; Operating system","score_opus":0.013074813625190692,"score_gpt":0.2167227787658998,"score_spread":0.2036479651407091,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2139168653","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.003026469,0.00043020656,0.65112495,0.00058014534,0.00031453941,0.000730822,0.014481918,0.3170046,0.012306445],"genre_scores_gemma":[0.018941505,0.0009693794,0.75304836,0.000961964,0.00013119048,0.0026790015,0.052958556,0.15367496,0.016635004],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99623317,0.0007961184,0.00042270488,0.0006581317,0.0015441115,0.00034581462],"domain_scores_gemma":[0.99371934,0.0031797723,0.00044018836,0.0008094261,0.0013371424,0.00051413715],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0059706406,0.003554721,0.0013928927,0.0026675265,0.0013308085,0.0035931075,0.0060168994,0.0014976534,0.058560852],"category_scores_gemma":[0.013501693,0.0026175042,0.002532943,0.0018302012,0.0013626835,0.0036928128,0.004788377,0.0048188567,0.045665655],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014843293,0.00030241135,0.0031602879,0.0032023133,0.00022862612,0.0016076548,0.0017333223,0.021037284,0.030547766,0.029099071,0.5312092,0.37638772],"study_design_scores_gemma":[0.0005287606,0.00020238337,0.0019425998,0.00088584784,0.00010803038,0.00097480067,0.00018516867,0.07822772,0.0573154,0.03262184,0.8265776,0.00042983054],"about_ca_topic_score_codex":0.0029666908,"about_ca_topic_score_gemma":0.0033377588,"teacher_disagreement_score":0.058560852,"about_ca_system_score_codex":0.0012102085,"about_ca_system_score_gemma":0.003241676,"threshold_uncertainty_score":0.19590557},"labels":[],"label_agreement":null},{"id":"W2139203357","doi":"10.1093/bioinformatics/btu196","title":"Association analysis using next-generation sequence data from publicly available control groups: the robust variance score statistic","year":2014,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":32,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"Ontario Ministry of Research and Innovation; European Commission; Epilepsy Research UK; Canadian Institutes of Health Research; National Institute for Health and Care Research; Cancer Research UK; Wellcome Trust; South London and Maudsley NHS Foundation Trust","keywords":"Type I and type II errors; Minor allele frequency; Statistic; Computer science; Confounding; Statistics; Selection (genetic algorithm); Variance (accounting); Allele frequency; Statistical power; Sequence (biology); Data mining; Genotype; Biology; Mathematics; Artificial intelligence; Genetics","score_opus":0.11260539496948144,"score_gpt":0.2841004375756698,"score_spread":0.17149504260618834,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2139203357","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.018188186,0.0016879475,0.9679947,0.0017434348,0.00023180533,0.00021988935,0.005037371,0.003696262,0.0012003944],"genre_scores_gemma":[0.30590647,0.0013574008,0.6724825,0.0011562663,0.0008025103,0.0012659548,0.0135902325,0.0022351793,0.0012034688],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9662673,0.021457601,0.0016788074,0.005963415,0.0042228787,0.0004100536],"domain_scores_gemma":[0.8991927,0.076845504,0.008062224,0.012167084,0.0028599086,0.00087260845],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.039327007,0.0008730003,0.0020335265,0.0037554726,0.0008072847,0.0028788147,0.0037398823,0.0021155311,0.005453176],"category_scores_gemma":[0.13189101,0.0008350031,0.0020713974,0.005308845,0.0023024662,0.0014707806,0.0022498586,0.0024810312,0.0022648436],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0022134772,0.00036003688,0.1294254,0.004484778,0.007326112,0.0017205253,0.0009225096,0.062865175,0.02037631,0.12490646,0.081162676,0.5642365],"study_design_scores_gemma":[0.0014557655,0.0007866762,0.065878965,0.0011591347,0.0024889219,0.0047259596,0.0002334011,0.44270995,0.018983185,0.3989147,0.062118474,0.00054493494],"about_ca_topic_score_codex":0.0028213328,"about_ca_topic_score_gemma":0.0024260369,"teacher_disagreement_score":0.039327007,"about_ca_system_score_codex":0.00089114753,"about_ca_system_score_gemma":0.0021970524,"threshold_uncertainty_score":0.20798361},"labels":[],"label_agreement":null},{"id":"W2139249401","doi":"10.1093/bioinformatics/btt673","title":"The International Society of Computational Biology presents: the Great Lakes Bioinformatics Conference, May 16–18, 2014, Cincinnati, Ohio","year":2013,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetics, Bioinformatics, and Biomedical Research","field":"Biochemistry, Genetics and Molecular Biology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Data science; Computational biology; Bioinformatics; Library science; Biology; Computer science","score_opus":0.025539242809749486,"score_gpt":0.2896759633183911,"score_spread":0.2641367205086416,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2139249401","genre_codex":"commentary","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0053784386,0.066473365,0.062282104,0.31618094,0.21847709,0.0012201874,0.018354394,0.016128276,0.29550514],"genre_scores_gemma":[0.01337786,0.049351323,0.04104403,0.03199929,0.033188313,0.0016027151,0.0240779,0.0078079794,0.7975506],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99450195,0.0013024489,0.0003398853,0.0012881929,0.0021091443,0.00045829106],"domain_scores_gemma":[0.9876067,0.0018942633,0.0003819045,0.0011856689,0.0047738845,0.0041574812],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.012082592,0.0023345994,0.0017127335,0.0021021124,0.0024681322,0.01075314,0.0026585765,0.0032514404,0.2222697],"category_scores_gemma":[0.016934192,0.0009995569,0.0012445954,0.0019403823,0.0014910578,0.005924323,0.0062798806,0.0070739733,0.12122417],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000037228623,0.000021268424,0.00021880354,0.000072888586,0.000012284428,0.000026285497,0.000043569195,0.00009242036,0.00022870376,0.0010779017,0.95864,0.03952861],"study_design_scores_gemma":[0.000017289216,0.000015587899,0.0005174651,0.00012829503,0.0000086705795,0.000035878827,0.00009656383,0.00038592046,0.00015609611,0.0011969954,0.9974245,0.000016836671],"about_ca_topic_score_codex":0.0035340704,"about_ca_topic_score_gemma":0.005423938,"teacher_disagreement_score":0.2222697,"about_ca_system_score_codex":0.0023425997,"about_ca_system_score_gemma":0.0054798955,"threshold_uncertainty_score":0.7435662},"labels":[],"label_agreement":null},{"id":"W2139417287","doi":"10.1093/bioinformatics/btg064","title":"Phylogenetic Web Profiler","year":2003,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canada Research Chairs; University of Toronto","funders":"","keywords":"Phylogenetic tree; Genome; Annotation; Similarity (geometry); Web server; Biology; Selection (genetic algorithm); Web service; Computer science; Computational biology; World Wide Web; Genetics; Artificial intelligence; The Internet; Gene","score_opus":0.011792514245245885,"score_gpt":0.22211089526825645,"score_spread":0.21031838102301056,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2139417287","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.008807184,0.000691507,0.29354084,0.0006226531,0.0002959988,0.0007265695,0.27612743,0.38635194,0.032835945],"genre_scores_gemma":[0.046561692,0.00087349874,0.25675282,0.00059933466,0.00014918248,0.001716159,0.6185829,0.044296812,0.030467588],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.998458,0.000232105,0.00021789945,0.0003390287,0.0005959105,0.00015691416],"domain_scores_gemma":[0.997603,0.0006454927,0.0002583063,0.00062252145,0.0006668287,0.000203959],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0020754752,0.0016365048,0.0015009458,0.00440154,0.0009708669,0.0021991346,0.0019904824,0.0010017868,0.057152774],"category_scores_gemma":[0.004751823,0.0012643322,0.0009954419,0.0043099197,0.00025595567,0.002022389,0.0017433745,0.0022106594,0.06389347],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001847767,0.0002605905,0.0072715166,0.0017412557,0.00024617376,0.0009086718,0.0005128082,0.0033894395,0.032691196,0.012227533,0.6447614,0.29414168],"study_design_scores_gemma":[0.00023068121,0.0001265108,0.007157861,0.00018964807,0.000111436784,0.0011203102,0.00014811136,0.033385046,0.033418115,0.015725324,0.9082475,0.00013953398],"about_ca_topic_score_codex":0.0014404064,"about_ca_topic_score_gemma":0.001438973,"teacher_disagreement_score":0.057152774,"about_ca_system_score_codex":0.0007379263,"about_ca_system_score_gemma":0.0012253253,"threshold_uncertainty_score":0.19119507},"labels":[],"label_agreement":null},{"id":"W2139627484","doi":"10.1093/bioinformatics/btp030","title":"IslandViewer: an integrated interface for computational identification and visualization of genomic islands","year":2009,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Salmonella and Campylobacter epidemiology","field":"Agricultural and Biological Sciences","cited_by":392,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"SFU Community Trust Endowment Fund; Canadian Institutes of Health Research; Michael Smith Health Research BC; Genome Canada","keywords":"Upload; Computer science; Graphical user interface; Interface (matter); Visualization; MIT License; License; Source code; Identification (biology); Web service; User interface; Genomics; Data mining; Comparative genomics; Chromosome; Hidden Markov model; Computational biology; Genome; World Wide Web; Biology; Artificial intelligence; Gene; Operating system; Genetics","score_opus":0.026748915155144534,"score_gpt":0.2853578140212567,"score_spread":0.25860889886611216,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2139627484","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0038561553,0.00055983104,0.4207924,0.0002873071,0.00016598693,0.00035130218,0.04737673,0.52160627,0.005003936],"genre_scores_gemma":[0.04414479,0.0014250524,0.6938098,0.0005800774,0.00016818794,0.0029512246,0.1519078,0.089513,0.015500134],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99943095,0.000118711476,0.0000635803,0.0001184062,0.00020656042,0.000061720115],"domain_scores_gemma":[0.9981969,0.0010745199,0.000088472276,0.00018119845,0.00024185186,0.00021719401],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0019432656,0.0025277296,0.0021561095,0.002475863,0.0005405461,0.0019665146,0.004688155,0.0016986545,0.10397225],"category_scores_gemma":[0.004748511,0.0016030746,0.001476369,0.0016528529,0.00043337303,0.0028822126,0.0032202015,0.0027766835,0.030388877],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0022181978,0.00033132805,0.003196452,0.002750595,0.00056290353,0.00083664584,0.00085171603,0.015365048,0.030161241,0.008643794,0.6683619,0.26672018],"study_design_scores_gemma":[0.0019313148,0.00033691048,0.008712871,0.00073832646,0.00027925958,0.0012590368,0.00024372722,0.40138042,0.04882584,0.029911747,0.5057373,0.00064318365],"about_ca_topic_score_codex":0.0026014503,"about_ca_topic_score_gemma":0.003085962,"teacher_disagreement_score":0.10397225,"about_ca_system_score_codex":0.00052812084,"about_ca_system_score_gemma":0.0010119139,"threshold_uncertainty_score":0.34782183},"labels":[],"label_agreement":null},{"id":"W2139856280","doi":"10.1093/bioinformatics/bts392","title":"Fractionation, rearrangement and subgenome dominance","year":2012,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genome Rearrangement Algorithms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":32,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Ottawa","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Fractionation; Polyploid; Biology; Gene duplication; Genome; Computer science; Gene; Genetics; Chemistry; Chromatography","score_opus":0.010925370706434448,"score_gpt":0.23500801163734025,"score_spread":0.2240826409309058,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2139856280","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.43928915,0.0008898785,0.55441624,0.00039403455,0.000044373388,0.0000970744,0.00015284274,0.0008665591,0.0038498205],"genre_scores_gemma":[0.868574,0.00030975847,0.12879683,0.00017672464,0.00002379123,0.000063954285,0.00028891835,0.00021391617,0.0015520661],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9994553,0.00010304862,0.000032986827,0.0002212215,0.00012663851,0.00006080651],"domain_scores_gemma":[0.99647456,0.0020446521,0.0006263584,0.0004311534,0.0002555189,0.00016774799],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012606075,0.00037916887,0.0004222198,0.00064976973,0.00048699038,0.0008663206,0.0007126887,0.0005221979,0.0025781302],"category_scores_gemma":[0.006871547,0.00024136416,0.00043345278,0.0008543159,0.001203255,0.0012984588,0.0011588543,0.00078174815,0.00034215493],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009633574,0.00012510404,0.07159021,0.00065925537,0.00011932286,0.00048610766,0.0010303695,0.2193747,0.23786938,0.105002046,0.0023457098,0.36043438],"study_design_scores_gemma":[0.00010718375,0.00035822732,0.033187106,0.00006859394,0.000090399306,0.0011302446,0.00032866205,0.67253625,0.17249417,0.10834675,0.011279929,0.000072500916],"about_ca_topic_score_codex":0.002677566,"about_ca_topic_score_gemma":0.0026490316,"teacher_disagreement_score":0.002677566,"about_ca_system_score_codex":0.0007126056,"about_ca_system_score_gemma":0.0006925169,"threshold_uncertainty_score":0.0086247325},"labels":[],"label_agreement":null},{"id":"W2139886875","doi":"10.1093/bioinformatics/btm064","title":"Clustering by common friends finds locally significant proteins mediating modules","year":2007,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":32,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"York University","funders":"Natural Sciences and Engineering Research Council of Canada; Ontario Innovation Trust","keywords":"Cluster analysis; Computer science; Precision and recall; Cluster (spacecraft); Domain (mathematical analysis); Computational biology; Function (biology); Cytoplasm; Set (abstract data type); Data mining; Theoretical computer science; Biology; Genetics; Artificial intelligence; Mathematics; Computer network","score_opus":0.008097790721321434,"score_gpt":0.22604835040608212,"score_spread":0.21795055968476068,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2139886875","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.84445107,0.002928011,0.14340453,0.0005025033,0.00005086674,0.00022655429,0.0020077897,0.0017880342,0.00464058],"genre_scores_gemma":[0.94707876,0.0004175018,0.04821347,0.00004095279,0.00005311474,0.000059828413,0.0027477734,0.000107542844,0.0012810386],"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9987865,0.00025536882,0.000049572685,0.0004950463,0.00027968996,0.00013382419],"domain_scores_gemma":[0.9961593,0.0017276552,0.00071383355,0.0006366213,0.00047218156,0.00029039706],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010177996,0.00084922503,0.0010102566,0.005346511,0.0015488112,0.0014731578,0.0012941449,0.0010327705,0.0032821237],"category_scores_gemma":[0.007488484,0.00035846554,0.0008388522,0.0030909304,0.0008164825,0.0015892786,0.0013656659,0.00036452318,0.0013990772],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0023620066,0.00070211437,0.47077987,0.0026598717,0.0021523838,0.0024687925,0.0028886874,0.074238606,0.07915938,0.01833694,0.02343135,0.32082006],"study_design_scores_gemma":[0.00029831144,0.00043079365,0.22809567,0.00021715795,0.0015385572,0.005865359,0.002409372,0.61208665,0.071809545,0.052531786,0.024517087,0.00019971025],"about_ca_topic_score_codex":0.003983395,"about_ca_topic_score_gemma":0.0042384528,"teacher_disagreement_score":0.005346511,"about_ca_system_score_codex":0.0006435648,"about_ca_system_score_gemma":0.00079676486,"threshold_uncertainty_score":0.010979831},"labels":[],"label_agreement":null},{"id":"W2139909337","doi":"10.1093/bioinformatics/btg191","title":"Greedy method for inferring tandem duplication history","year":2003,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomic variations and chromosomal abnormalities","field":"Biochemistry, Genetics and Molecular Biology","cited_by":36,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Natural Sciences and Engineering Research Council of Canada; City University of Hong Kong","keywords":"Computer science; Gene duplication; Tandem exon duplication; Tandem; Computational biology; Biology; Genetics; Gene; Engineering","score_opus":0.02016407766159523,"score_gpt":0.2511239886199717,"score_spread":0.23095991095837648,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2139909337","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.011502744,0.00022686122,0.9854145,0.00013972889,0.000037805912,0.00010892677,0.00034623724,0.0015680527,0.000655271],"genre_scores_gemma":[0.11782595,0.00017581678,0.8777952,0.00018417482,0.00010379652,0.000391353,0.0017073272,0.0002696144,0.001546743],"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99844056,0.0005915606,0.00008811642,0.00038494205,0.00034661626,0.00014825928],"domain_scores_gemma":[0.99421716,0.004338533,0.0002883322,0.00044515575,0.0005163233,0.00019445061],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002987995,0.0013107518,0.0017801167,0.0031177592,0.0008878828,0.0011652252,0.0033947935,0.001729193,0.0053495443],"category_scores_gemma":[0.009805686,0.00081965287,0.0010384549,0.0029874372,0.0010662788,0.0018015219,0.0014999543,0.0014037404,0.0017267739],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00076987396,0.00022246003,0.004995751,0.00036996033,0.00027582646,0.00034477818,0.0001395805,0.50266963,0.0064847665,0.015270839,0.0131468875,0.45530963],"study_design_scores_gemma":[0.00009097169,0.00004516518,0.0003936382,0.00001296163,0.000029751749,0.00013447631,0.000021217991,0.978005,0.0013794288,0.018602503,0.0012635977,0.000021346183],"about_ca_topic_score_codex":0.006391382,"about_ca_topic_score_gemma":0.007256231,"teacher_disagreement_score":0.006391382,"about_ca_system_score_codex":0.0011015114,"about_ca_system_score_gemma":0.003238665,"threshold_uncertainty_score":0.017895997},"labels":[],"label_agreement":null},{"id":"W2139911807","doi":"10.1093/bioinformatics/btr372","title":"In-depth annotation of SNPs arising from resequencing projects using NGS-SNP","year":2011,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic diversity and population structure","field":"Biochemistry, Genetics and Molecular Biology","cited_by":121,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"SNP; Annotation; Single-nucleotide polymorphism; Computational biology; Computer science; Biology; Genetics; Artificial intelligence; Genotype; Gene","score_opus":0.08122098224689127,"score_gpt":0.2782390111609927,"score_spread":0.19701802891410142,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2139911807","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.10272954,0.0013582268,0.60747236,0.00051463617,0.0006138763,0.0010760383,0.21350764,0.05478519,0.017942527],"genre_scores_gemma":[0.11350776,0.000537778,0.66272706,0.00053651223,0.00014795132,0.00096846215,0.20331034,0.01034664,0.007917553],"study_design_codex":"bench_or_experimental","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9978115,0.0003578111,0.00035771282,0.00072227535,0.0006103237,0.00014045599],"domain_scores_gemma":[0.99340034,0.0032733318,0.00073919713,0.0012881577,0.0009811458,0.00031786683],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0035366642,0.0010933793,0.0013259181,0.0042008054,0.0015830698,0.001723901,0.0013137745,0.00090890523,0.013935194],"category_scores_gemma":[0.010054841,0.00069240277,0.0013400345,0.0038491697,0.00033850005,0.0010796938,0.0015935549,0.001508005,0.010423034],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0026783545,0.00034058717,0.044164784,0.0074130883,0.00077617844,0.004485507,0.004989215,0.012517033,0.4203702,0.012998314,0.15536188,0.33390495],"study_design_scores_gemma":[0.00030380004,0.00030441198,0.10570776,0.0009373805,0.0008887602,0.0044089016,0.0010839781,0.06716431,0.26435506,0.02701159,0.52742875,0.00040536217],"about_ca_topic_score_codex":0.0017584624,"about_ca_topic_score_gemma":0.0049811564,"teacher_disagreement_score":0.013935194,"about_ca_system_score_codex":0.0006023509,"about_ca_system_score_gemma":0.0014609284,"threshold_uncertainty_score":0.046617925},"labels":[],"label_agreement":null},{"id":"W2139993758","doi":"10.1093/bioinformatics/bts484","title":"PRISM: Pair-read informed split-read mapping for base-pair level detection of insertion, deletion and structural variants","year":2012,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":128,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Hospital for Sick Children; University of Toronto","funders":"Natural Sciences and Engineering Research Council of Canada; Canadian Institutes of Health Research; China Scholarship Council","keywords":"Base (topology); Prism; Computer science; Software; Algorithm; Mathematics; Physics; Optics; Programming language","score_opus":0.03396231251202535,"score_gpt":0.24663264072529348,"score_spread":0.21267032821326814,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2139993758","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.03215245,0.00078282325,0.8380463,0.00028903756,0.00022006208,0.00048128498,0.025653942,0.09814701,0.004227064],"genre_scores_gemma":[0.10377699,0.00025813936,0.860676,0.00029433498,0.000051297324,0.00077829877,0.026776578,0.005232539,0.0021557915],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9952319,0.001183723,0.0002930873,0.0015755084,0.0014363423,0.00027935504],"domain_scores_gemma":[0.9942767,0.0028790054,0.0008251474,0.0010624545,0.0006307981,0.000325887],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0052463943,0.002368961,0.0015731354,0.0021299173,0.00084002886,0.0015829976,0.0043077716,0.0014104634,0.010748773],"category_scores_gemma":[0.013888874,0.0014958391,0.0018792945,0.0024017263,0.0009861538,0.0018113831,0.0032744296,0.0026703354,0.007628159],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0055063325,0.00061861787,0.029768953,0.005745579,0.0021296246,0.001712434,0.0012648312,0.0653063,0.34825474,0.020383442,0.124813065,0.3944961],"study_design_scores_gemma":[0.0009247146,0.001145536,0.01667048,0.00022212164,0.0004822742,0.0023737256,0.0002761555,0.5568639,0.3227861,0.026202908,0.07151961,0.0005325429],"about_ca_topic_score_codex":0.0023838729,"about_ca_topic_score_gemma":0.0046975184,"teacher_disagreement_score":0.010748773,"about_ca_system_score_codex":0.00088132266,"about_ca_system_score_gemma":0.002441513,"threshold_uncertainty_score":0.03595823},"labels":[],"label_agreement":null},{"id":"W2140001356","doi":"10.1093/bioinformatics/btp259","title":"Application and evaluation of automated semantic annotation of gene expression experiments","year":2009,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Biomedical Text Mining and Ontologies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"National Institute of General Medical Sciences; Natural Sciences and Engineering Research Council of Canada; Canadian Institutes of Health Research; National Institutes of Health; Michael Smith Health Research BC","keywords":"Computer science; Unified Medical Language System; Annotation; Natural language processing; Documentation; Information retrieval; Controlled vocabulary; Software; Expression (computer science); Pipeline (software); Context (archaeology); Artificial intelligence; Source code; Programming language","score_opus":0.023997876135531407,"score_gpt":0.326307267117323,"score_spread":0.3023093909817916,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2140001356","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.46944362,0.0040061567,0.36867893,0.002153545,0.0009998892,0.0025848874,0.02159773,0.11808798,0.012447293],"genre_scores_gemma":[0.40712982,0.0005999284,0.532751,0.00074894744,0.00013561497,0.0012580232,0.05063585,0.0035414076,0.0031994076],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9731358,0.013207208,0.0020542552,0.005260631,0.005697173,0.0006448645],"domain_scores_gemma":[0.9398437,0.035362393,0.0020940257,0.009431545,0.0123783285,0.00089008396],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.029345382,0.0022710834,0.0013606638,0.005925643,0.001903368,0.00335349,0.0033679425,0.0029623345,0.0042124866],"category_scores_gemma":[0.07339972,0.0005874386,0.0019505899,0.0032050977,0.0017656646,0.0037620165,0.004196067,0.0015755616,0.0024192077],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.007853603,0.0028359804,0.040833563,0.009749242,0.0019020897,0.0015108454,0.0052680513,0.06268299,0.15610409,0.012067199,0.06640964,0.63278264],"study_design_scores_gemma":[0.00092501414,0.0021414254,0.049518757,0.0008201449,0.0010867455,0.0013500873,0.0023681961,0.61190253,0.2207765,0.016221464,0.0924864,0.00040277367],"about_ca_topic_score_codex":0.007700226,"about_ca_topic_score_gemma":0.0063476106,"teacher_disagreement_score":0.029345382,"about_ca_system_score_codex":0.0031486598,"about_ca_system_score_gemma":0.004118334,"threshold_uncertainty_score":0.15519506},"labels":[],"label_agreement":null},{"id":"W2140028978","doi":"10.1093/bioinformatics/btq429","title":"Datamonkey 2010: a suite of phylogenetic analysis tools for evolutionary biology","year":2010,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1235,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"AIDS Vancouver","funders":"National Institute of Allergy and Infectious Diseases; Canadian Institutes of Health Research","keywords":"Suite; Phylogenetic tree; Biology; Identification (biology); Documentation; Selection (genetic algorithm); Computer science; Evolutionary biology; Computational biology; Data science; Machine learning; Genetics; Gene; Ecology; Geography; Programming language","score_opus":0.018902856069135335,"score_gpt":0.25981028669467615,"score_spread":0.24090743062554082,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2140028978","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0037785748,0.0015306001,0.25058848,0.0016010766,0.0010812014,0.001072793,0.34665474,0.38362148,0.010070988],"genre_scores_gemma":[0.022026643,0.0021581862,0.42544436,0.0020096232,0.00031982412,0.004721412,0.36275682,0.17329876,0.0072643827],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9955656,0.0013903613,0.0007311962,0.00082866685,0.0010700356,0.00041411872],"domain_scores_gemma":[0.986852,0.007336787,0.0010288489,0.00229237,0.0013555575,0.0011344212],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.011411946,0.0038863905,0.004115955,0.006185757,0.0021972896,0.005240904,0.009641999,0.002112593,0.08203473],"category_scores_gemma":[0.030266192,0.004187778,0.0040288065,0.0077736326,0.001591745,0.009692608,0.008088112,0.008636242,0.052166138],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0017919925,0.0001960557,0.0035883063,0.005497076,0.0007872539,0.0007824228,0.0011300549,0.0031391208,0.007329491,0.018078724,0.8748193,0.08286019],"study_design_scores_gemma":[0.0013476803,0.00014227834,0.00589846,0.0014829765,0.00037300427,0.0010936582,0.0003532002,0.021460202,0.011980231,0.06144589,0.8938604,0.0005620357],"about_ca_topic_score_codex":0.004073719,"about_ca_topic_score_gemma":0.0046585114,"teacher_disagreement_score":0.08203473,"about_ca_system_score_codex":0.0015840093,"about_ca_system_score_gemma":0.005468432,"threshold_uncertainty_score":0.27443355},"labels":[],"label_agreement":null},{"id":"W2140048165","doi":"10.1093/bioinformatics/bth239","title":"Estimating and comparing the rates of gene discovery and expressed sequence tag (EST) frequencies in EST surveys","year":2004,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":35,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canadian Institute for Advanced Research; Dalhousie University","funders":"Genome Atlantic; Natural Sciences and Engineering Research Council of Canada; Canadian Institute for Advanced Research; Alfred P. Sloan Foundation","keywords":"Redundancy (engineering); Sample (material); Sequence (biology); Sample size determination; Computer science; False discovery rate; Gene; Statistics; Biology; Genetics; Mathematics","score_opus":0.03646441197339777,"score_gpt":0.281118808471943,"score_spread":0.2446543964985452,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2140048165","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.31478024,0.0008546767,0.6811216,0.00022413638,0.000054056083,0.00016367702,0.0008703294,0.0006621754,0.0012690685],"genre_scores_gemma":[0.7046326,0.00079683634,0.28830445,0.00020377008,0.000102029095,0.0005055965,0.003905191,0.00044960028,0.0010999026],"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9835814,0.00691458,0.0011406059,0.003950963,0.0037609444,0.0006514186],"domain_scores_gemma":[0.7676194,0.1942192,0.01908358,0.0097982725,0.008150605,0.0011290048],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.03534972,0.001097685,0.0013326807,0.005785954,0.0006534455,0.0029291334,0.002323945,0.002108306,0.00067541376],"category_scores_gemma":[0.14558934,0.00096669997,0.0015626885,0.004842318,0.0021963292,0.004599827,0.0025944433,0.0017863017,0.0005343599],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008818449,0.0002484949,0.59487575,0.0007833528,0.00096763,0.00039665808,0.0011120011,0.1948139,0.028174737,0.023814043,0.0012087511,0.1527229],"study_design_scores_gemma":[0.00006289668,0.0005961453,0.17575356,0.00014474992,0.0002734448,0.0013814986,0.00075580284,0.70741206,0.056976803,0.051485084,0.0048540533,0.00030395057],"about_ca_topic_score_codex":0.001673812,"about_ca_topic_score_gemma":0.001373609,"teacher_disagreement_score":0.03534972,"about_ca_system_score_codex":0.0017183684,"about_ca_system_score_gemma":0.0009927725,"threshold_uncertainty_score":0.18694937},"labels":[],"label_agreement":null},{"id":"W2140154634","doi":"10.1093/bioinformatics/btr046","title":"SHRiMP2: Sensitive yet Practical Short Read Mapping","year":2011,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":365,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of Toronto","funders":"Canadian Institutes of Health Research","keywords":"Executable; Computer science; Shrimp; Space (punctuation); Sensitivity (control systems); Source code; Code (set theory); Data mining; Programming language; Biology; Operating system; Engineering; Set (abstract data type)","score_opus":0.0499284744864106,"score_gpt":0.2654698643525903,"score_spread":0.21554138986617968,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2140154634","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.02522522,0.00060054637,0.7541982,0.0007038187,0.0010526979,0.0006012954,0.035540864,0.17224011,0.009837268],"genre_scores_gemma":[0.0849712,0.0005171804,0.7402457,0.0010878368,0.00026823638,0.0016822093,0.10206743,0.04038066,0.028779533],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9981798,0.0002855566,0.00015414323,0.00053369,0.0007267412,0.000120253455],"domain_scores_gemma":[0.9975883,0.00075355486,0.00014500458,0.0006741054,0.00062418624,0.00021483412],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0032818613,0.002035207,0.0010035761,0.0008575113,0.00092454057,0.0018689949,0.0028675813,0.000774864,0.02306305],"category_scores_gemma":[0.0065115974,0.0014746835,0.0013409811,0.0010197923,0.00047348387,0.0015725589,0.0035742805,0.003048306,0.016511442],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0027258613,0.0002013592,0.006769337,0.0019238062,0.00044611632,0.0007369116,0.0006977162,0.010335311,0.28546125,0.007964789,0.33402887,0.34870866],"study_design_scores_gemma":[0.0006062617,0.00047177903,0.0060968753,0.00025174292,0.00013928082,0.0012379267,0.00016148665,0.09837721,0.3542326,0.013447865,0.52456856,0.0004084595],"about_ca_topic_score_codex":0.0013890928,"about_ca_topic_score_gemma":0.0019538354,"teacher_disagreement_score":0.02306305,"about_ca_system_score_codex":0.00045340782,"about_ca_system_score_gemma":0.0012324243,"threshold_uncertainty_score":0.07715362},"labels":[],"label_agreement":null},{"id":"W2140163842","doi":"10.1093/bioinformatics/btu178","title":"rTANDEM, an R/Bioconductor package for MS/MS protein identification","year":2014,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Advanced Proteomics Techniques and Applications","field":"Chemistry","cited_by":12,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université Laval","funders":"","keywords":"Bioconductor; R package; Computer science; Pipeline (software); Identification (biology); JavaScript; Interface (matter); Software; Data mining; Operating system; Programming language; Biology","score_opus":0.02169888587174565,"score_gpt":0.28549764496068214,"score_spread":0.26379875908893646,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2140163842","genre_codex":"dataset","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0024267123,0.0040973523,0.2626985,0.001526894,0.0012695463,0.00052002247,0.3977947,0.31295046,0.016715862],"genre_scores_gemma":[0.016482677,0.0026131568,0.352602,0.0022159733,0.0005205409,0.002887558,0.41010648,0.18837713,0.024194544],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9960956,0.0010609718,0.00039585095,0.0013056775,0.0008612418,0.00028057885],"domain_scores_gemma":[0.99470633,0.0020787108,0.0008211877,0.0010213902,0.0010449479,0.00032747243],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005012436,0.0036732943,0.0039219423,0.0049433606,0.0014495351,0.0047094896,0.0067066676,0.0019004198,0.13525231],"category_scores_gemma":[0.018303504,0.0018898814,0.0030315192,0.004605341,0.0009811779,0.0037390864,0.0037337241,0.0037545771,0.17830765],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0004928773,0.000031058753,0.0011210472,0.0037136856,0.0006464512,0.00022359364,0.00015150788,0.0015663168,0.005787839,0.0061409464,0.94807976,0.032044776],"study_design_scores_gemma":[0.0003082914,0.00009406024,0.0028974626,0.000551883,0.0005045028,0.0007477965,0.00007395967,0.013661262,0.010989319,0.02820346,0.94178027,0.000187664],"about_ca_topic_score_codex":0.0026811464,"about_ca_topic_score_gemma":0.003231916,"teacher_disagreement_score":0.13525231,"about_ca_system_score_codex":0.0011813692,"about_ca_system_score_gemma":0.0034206,"threshold_uncertainty_score":0.45246404},"labels":[],"label_agreement":null},{"id":"W2140505514","doi":"10.1093/bioinformatics/btt217","title":"A weighted sampling algorithm for the design of RNA sequences with targeted secondary structure and nucleotide distribution","year":2013,"lang":"en","type":"article","venue":"Bioinformatics","topic":"RNA and protein synthesis mechanisms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":60,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"McGill University","funders":"","keywords":"Sampling (signal processing); Algorithm; Nucleic acid secondary structure; Nucleotide; Computational biology; RNA; Distribution (mathematics); Computer science; Mathematics; Biology; Combinatorics; Genetics; Gene","score_opus":0.011839337860578933,"score_gpt":0.2168797599005093,"score_spread":0.2050404220399304,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2140505514","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0059285765,0.00008090102,0.992732,0.000023994504,0.000012920668,0.000059314236,0.000027359054,0.00043073104,0.0007042595],"genre_scores_gemma":[0.074342415,0.00012999355,0.92272305,0.000096458294,0.000021284668,0.00038307434,0.00025896128,0.00032995403,0.0017148189],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9992786,0.0002322107,0.000038299953,0.00015425934,0.0002431951,0.0000533857],"domain_scores_gemma":[0.99925107,0.0004268238,0.00006618164,0.00006644635,0.00014313553,0.000046219502],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012861302,0.00097275525,0.00074046326,0.00069143344,0.00041771575,0.0005522483,0.00095493003,0.00073222275,0.003013025],"category_scores_gemma":[0.0025978298,0.00053055276,0.00081336935,0.000728697,0.0005583777,0.0005996796,0.0008280898,0.00088378746,0.0009700814],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00029984818,0.00018297584,0.0011926732,0.0003782106,0.00009824121,0.0001600997,0.00017419891,0.54104537,0.059809547,0.026347445,0.0036685164,0.36664286],"study_design_scores_gemma":[0.00006478842,0.00014505477,0.000089980706,0.000013735389,0.000016593309,0.00006826752,0.000010716799,0.9804646,0.007969569,0.007750669,0.0033946512,0.00001135626],"about_ca_topic_score_codex":0.0010798146,"about_ca_topic_score_gemma":0.0019721817,"teacher_disagreement_score":0.003013025,"about_ca_system_score_codex":0.0006254063,"about_ca_system_score_gemma":0.0011210947,"threshold_uncertainty_score":0.010079563},"labels":[],"label_agreement":null},{"id":"W2140593394","doi":"10.1093/bioinformatics/btq265","title":"FuncBase : a resource for quantitative gene function annotation","year":2010,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":16,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"National Institute of Neurological Disorders and Stroke; National Human Genome Research Institute; National Heart, Lung, and Blood Institute","keywords":"Annotation; Function (biology); Gene ontology; Computer science; Resource (disambiguation); Gene Annotation; Web resource; Linkage (software); Gene; Ontology; Computational biology; Focus (optics); Information retrieval; Biology; World Wide Web; Genome; Artificial intelligence; Genetics; Gene expression","score_opus":0.012546597740124498,"score_gpt":0.2442021333390203,"score_spread":0.2316555355988958,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2140593394","genre_codex":"dataset","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0023722192,0.0018553043,0.1312685,0.00041527173,0.00043878384,0.00031318236,0.6718081,0.18266779,0.008860832],"genre_scores_gemma":[0.006223932,0.0009153269,0.09397561,0.00028010004,0.00009855737,0.00085299136,0.87219304,0.021789249,0.0036711537],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9982449,0.00021208795,0.00024875792,0.0005297612,0.0006068161,0.00015781554],"domain_scores_gemma":[0.99738616,0.0007311768,0.0003740581,0.00054560875,0.00070120144,0.0002617033],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0026572163,0.0055316277,0.0034751198,0.009084136,0.0018741122,0.0037079488,0.007923505,0.0022231191,0.08980221],"category_scores_gemma":[0.0068099955,0.002042617,0.0024405373,0.007653236,0.0006306803,0.0034306848,0.003575918,0.0033106112,0.10495503],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001041761,0.00018251881,0.001937872,0.0054223267,0.00036977034,0.0004954465,0.0002413521,0.0032386126,0.027419748,0.007920476,0.85557914,0.09615103],"study_design_scores_gemma":[0.00040900905,0.00016003443,0.005570791,0.00086896395,0.00035397743,0.0013318842,0.0001237016,0.014238157,0.03681893,0.034118693,0.90563685,0.0003689411],"about_ca_topic_score_codex":0.0035508957,"about_ca_topic_score_gemma":0.0021581438,"teacher_disagreement_score":0.08980221,"about_ca_system_score_codex":0.0014703843,"about_ca_system_score_gemma":0.0028372488,"threshold_uncertainty_score":0.30041832},"labels":[],"label_agreement":null},{"id":"W2141010659","doi":"10.1093/bioinformatics/btm570","title":"Automatic synchronization and distribution of biological databases and software over low-bandwidth networks among developing countries","year":2007,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetics, Bioinformatics, and Biomedical Research","field":"Biochemistry, Genetics and Molecular Biology","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"International Development Research Centre","keywords":"BitTorrent; Download; File Transfer Protocol; Computer science; Distributed database; Software; Bandwidth (computing); Developing country; Database; Synchronization (alternating current); Computer network; The Internet; World Wide Web; Operating system","score_opus":0.014920001751989988,"score_gpt":0.2693107648705665,"score_spread":0.2543907631185765,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2141010659","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.84580195,0.00039938808,0.13949245,0.00032016062,0.00006536822,0.000389477,0.0017375452,0.005747104,0.0060465466],"genre_scores_gemma":[0.9413769,0.00028743807,0.05474473,0.000034043656,0.000016859518,0.00014230056,0.0022224365,0.00016008884,0.0010152408],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99907947,0.00032683686,0.000071776885,0.00018760341,0.0002510145,0.00008326073],"domain_scores_gemma":[0.9946413,0.0020750354,0.00057531684,0.0010812936,0.0014315383,0.0001955014],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002036835,0.0003275156,0.00026660622,0.0010056343,0.0007344703,0.0007109627,0.0010404459,0.00045255394,0.0018722115],"category_scores_gemma":[0.008614513,0.00016800434,0.00016791704,0.0014815024,0.0004220331,0.0019715184,0.0012138827,0.00037980452,0.0010619406],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.002217621,0.00072210946,0.17033447,0.0007712053,0.00013642902,0.0010262452,0.0018949568,0.13661045,0.10970314,0.0063933395,0.01372627,0.5564638],"study_design_scores_gemma":[0.000254952,0.0010222341,0.07624399,0.00010166331,0.00014547662,0.0007695536,0.0014943745,0.71129715,0.17733356,0.0051534073,0.026079003,0.000104694176],"about_ca_topic_score_codex":0.004196533,"about_ca_topic_score_gemma":0.0016367121,"teacher_disagreement_score":0.004196533,"about_ca_system_score_codex":0.0005013384,"about_ca_system_score_gemma":0.0008114409,"threshold_uncertainty_score":0.01077199},"labels":[],"label_agreement":null},{"id":"W2141010693","doi":"10.1093/bioinformatics/btm217","title":"Bayesian association of haplotypes and non-genetic factors to regulatory and phenotypic variation in human populations","year":2007,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"Wellcome Trust; Microsoft Research","keywords":"Biology; Single-nucleotide polymorphism; Haplotype; Genetics; Genetic variation; Genetic association; Phenotype; Copy-number variation; Tag SNP; Quantitative trait locus; Genotype; Gene; Computational biology; Genome","score_opus":0.011397283085800163,"score_gpt":0.2575955666381467,"score_spread":0.24619828355234655,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2141010693","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.58408093,0.0014991863,0.41082212,0.0012888783,0.000037611244,0.0000696971,0.000714722,0.00028229828,0.0012045405],"genre_scores_gemma":[0.96654874,0.00065166823,0.031052876,0.00022618576,0.000063683794,0.00009384025,0.0005906542,0.000037006866,0.00073538115],"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9973054,0.001786351,0.00009163638,0.00058843306,0.00014609373,0.00008198556],"domain_scores_gemma":[0.9761066,0.020932171,0.0014335929,0.0009371684,0.00030434417,0.00028611027],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007019529,0.00034774505,0.0007996834,0.0009659279,0.000389657,0.0009364456,0.00091459625,0.000839928,0.0018535719],"category_scores_gemma":[0.026137298,0.00051765743,0.0006503822,0.000993069,0.0012471724,0.00086200394,0.00095711555,0.0010378812,0.00023829506],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010150294,0.00016239138,0.5735598,0.00034114817,0.0013158714,0.00065537536,0.00067686377,0.28087872,0.004925639,0.035096742,0.001715688,0.0996568],"study_design_scores_gemma":[0.00012985057,0.00013979203,0.14130609,0.000070810456,0.0003110784,0.0006408345,0.00010616647,0.72117895,0.0011240263,0.13337447,0.0015448272,0.00007311886],"about_ca_topic_score_codex":0.004959157,"about_ca_topic_score_gemma":0.0047046733,"teacher_disagreement_score":0.007019529,"about_ca_system_score_codex":0.00056842336,"about_ca_system_score_gemma":0.00066996,"threshold_uncertainty_score":0.037123263},"labels":[],"label_agreement":null},{"id":"W2141014201","doi":"10.1093/bioinformatics/btt182","title":"Predicting the functional consequences of cancer-associated amino acid substitutions","year":2013,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Cancer Genomics and Diagnostics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":282,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Institute of Nutrition, Metabolism and Diabetes; Biotechnology and Biological Sciences Research Council; Directorate for Biological Sciences; Medical Research Council; University of Bristol","keywords":"Missense mutation; Computational biology; Computer science; Mutation; Exome; Neutral mutation; Genome; Cancer; Exome sequencing; Genetics; Biology; Gene","score_opus":0.014641703727665044,"score_gpt":0.23337261783227423,"score_spread":0.21873091410460918,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2141014201","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8857361,0.0014471699,0.09778075,0.00039116762,0.000049471288,0.00008493039,0.005544161,0.006977688,0.001988541],"genre_scores_gemma":[0.9165873,0.0005592594,0.07382801,0.00010799479,0.000024513603,0.00004799265,0.007850441,0.00020517198,0.0007893439],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99972147,0.00006287095,0.000026750395,0.00009526515,0.00006722381,0.000026416412],"domain_scores_gemma":[0.99876213,0.00079959683,0.00012657266,0.00007764928,0.00017909418,0.000055014658],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00094016554,0.00094579166,0.000652675,0.00091048755,0.00034070795,0.0006354598,0.0007841614,0.000900898,0.0016480553],"category_scores_gemma":[0.0027690306,0.00016953857,0.00047989772,0.0009145165,0.00025049632,0.00043329844,0.00033405048,0.00043966476,0.00074638706],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0018656338,0.0004334877,0.14447297,0.0013439072,0.00037152835,0.0009763408,0.000113633534,0.5426825,0.10894266,0.0021218518,0.007049032,0.18962637],"study_design_scores_gemma":[0.00004896367,0.00018004182,0.02872026,0.000031856132,0.00012913797,0.00072443887,0.000031822212,0.9032842,0.06189811,0.0033065234,0.0016163684,0.00002817776],"about_ca_topic_score_codex":0.0033138334,"about_ca_topic_score_gemma":0.004462557,"teacher_disagreement_score":0.0033138334,"about_ca_system_score_codex":0.0005409425,"about_ca_system_score_gemma":0.0007420594,"threshold_uncertainty_score":0.0065891147},"labels":[],"label_agreement":null},{"id":"W2141064379","doi":"10.1093/bioinformatics/bts262","title":"Two effective methods for correcting experimental high-throughput screening data","year":2012,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Computational Drug Discovery Methods","field":"Computer Science","cited_by":19,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; Université du Québec à Montréal","funders":"","keywords":"Computer science; Preprocessor; Process (computing); Data mining; Throughput; Systematic error; Error detection and correction; Data pre-processing; Statistics; Algorithm; Artificial intelligence; Mathematics","score_opus":0.09587939461284134,"score_gpt":0.4428652021612223,"score_spread":0.346985807548381,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2141064379","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0021623904,0.00037835835,0.99466467,0.0002595026,0.0002728526,0.00013228964,0.00020739704,0.0015358108,0.00038671136],"genre_scores_gemma":[0.04278406,0.0006671553,0.95008516,0.00045200382,0.0003768666,0.0007269567,0.0013907232,0.00075268873,0.0027643752],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.97907937,0.0049917614,0.0013171252,0.003085514,0.010925158,0.00060105446],"domain_scores_gemma":[0.9414789,0.023665095,0.0065259025,0.0143428575,0.013092753,0.0008944657],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.013310618,0.0031668309,0.0025246176,0.0050004525,0.0013395911,0.0026788176,0.0050085397,0.0030575667,0.0043779113],"category_scores_gemma":[0.058442347,0.0011073912,0.002531876,0.0041369824,0.0025561482,0.0038831003,0.004138688,0.004765072,0.0032454561],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007927514,0.00043081932,0.010375174,0.0016619866,0.00067372713,0.00058398437,0.0004071405,0.052258763,0.043730244,0.035765134,0.014851062,0.83846915],"study_design_scores_gemma":[0.00032212966,0.0006208229,0.012839085,0.00036868494,0.00048197285,0.0024532874,0.00019670023,0.6795952,0.15779638,0.087893166,0.056838006,0.0005945959],"about_ca_topic_score_codex":0.0016748641,"about_ca_topic_score_gemma":0.002307685,"teacher_disagreement_score":0.013310618,"about_ca_system_score_codex":0.0014510308,"about_ca_system_score_gemma":0.0032225105,"threshold_uncertainty_score":0.0703941},"labels":[],"label_agreement":null},{"id":"W2141222038","doi":"10.1093/bioinformatics/btn305","title":"FindPeaks 3.1: a tool for identifying areas of enrichment from massively parallel short-read sequencing technology","year":2008,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":270,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"BC Cancer Agency","funders":"Genome British Columbia; Michael Smith Health Research BC; Genome Canada","keywords":"Computer science; Massive parallel sequencing; Software; Java; DNA sequencing; Genomics; Massively parallel; Source code; File format; Genome; Database; Operating system; Biology; DNA; Genetics; Gene","score_opus":0.024781069853669895,"score_gpt":0.25104059722478206,"score_spread":0.22625952737111216,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2141222038","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0045089354,0.0004578414,0.42952615,0.00025340455,0.00018362633,0.0003358823,0.048400395,0.5128748,0.0034590515],"genre_scores_gemma":[0.02108546,0.00045086074,0.79666483,0.00054680463,0.000091769114,0.002410681,0.099377885,0.07315614,0.0062155533],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9984187,0.00019536009,0.00018496269,0.00042587225,0.0006186733,0.00015647507],"domain_scores_gemma":[0.99746346,0.0013869599,0.0003269427,0.00028823275,0.0003763845,0.00015799004],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0028552006,0.0034140993,0.0023957063,0.0056382366,0.0015213231,0.0025082617,0.0042819986,0.0015980167,0.04883084],"category_scores_gemma":[0.0075533222,0.0029626866,0.0026684985,0.0034495627,0.00065479433,0.0022196157,0.0029414196,0.0027895242,0.030788615],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0022477582,0.0002688165,0.005513729,0.004714708,0.0007934744,0.001508331,0.0007263915,0.0084019,0.06714151,0.0078762965,0.608126,0.29268113],"study_design_scores_gemma":[0.0012679079,0.00039968282,0.010445864,0.00087660487,0.00052522984,0.0025848898,0.000306959,0.1797103,0.21154165,0.049400404,0.54218763,0.00075292226],"about_ca_topic_score_codex":0.0019169024,"about_ca_topic_score_gemma":0.002481442,"teacher_disagreement_score":0.04883084,"about_ca_system_score_codex":0.00091314124,"about_ca_system_score_gemma":0.0023336157,"threshold_uncertainty_score":0.16335541},"labels":[],"label_agreement":null},{"id":"W2141316905","doi":"10.1093/bioinformatics/bts726","title":"G<scp>e</scp>S<scp>to</scp>D<scp>ifferent</scp>: a Cytoscape plugin for the generation and the identification of gene regulatory networks describing a stochastic cell differentiation process","year":2013,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene Regulatory Network Analysis","field":"Biochemistry, Genetics and Molecular Biology","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"National Institute of General Medical Sciences","keywords":"Identification (biology); Computer science; Gene regulatory network; Plug-in; Computational biology; Systems biology; Biology; Theoretical computer science; Gene; Genetics; Gene expression","score_opus":0.013052161573079488,"score_gpt":0.21098444103755504,"score_spread":0.19793227946447556,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2141316905","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0074141263,0.000627417,0.14861895,0.0010788668,0.0009292478,0.00053884287,0.19290093,0.6022173,0.045674365],"genre_scores_gemma":[0.0776974,0.0015378207,0.14002231,0.0031008061,0.00047577193,0.0032088982,0.4269497,0.24500172,0.10200558],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9996816,0.000029000917,0.000019852267,0.00008348501,0.0001433311,0.000042669057],"domain_scores_gemma":[0.9990477,0.00031709898,0.000053568565,0.00021821327,0.00018908845,0.00017431144],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005190038,0.0016594597,0.0010800775,0.002052827,0.00094112585,0.0015117984,0.0026169491,0.0018893111,0.2574092],"category_scores_gemma":[0.0024255903,0.0010767899,0.001387746,0.0015110917,0.00055155816,0.001920226,0.0020050223,0.0021048689,0.12496445],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00033792856,0.00013636566,0.0012901246,0.0015377362,0.00012255347,0.0008117857,0.0001463604,0.0023643232,0.015878947,0.005695001,0.8840812,0.08759766],"study_design_scores_gemma":[0.00060473295,0.00006729878,0.0045277635,0.0003426926,0.00008984615,0.0010673173,0.000060877413,0.02836965,0.07323516,0.019870704,0.871565,0.00019901025],"about_ca_topic_score_codex":0.0048101246,"about_ca_topic_score_gemma":0.0077860565,"teacher_disagreement_score":0.2574092,"about_ca_system_score_codex":0.00089648133,"about_ca_system_score_gemma":0.001163188,"threshold_uncertainty_score":0.8611195},"labels":[],"label_agreement":null},{"id":"W2141363231","doi":"10.1093/bioinformatics/btq416","title":"OWL2Perl: creating Perl modules from OWL class definitions","year":2010,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Biomedical Text Mining and Ontologies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"St. Paul's Hospital","funders":"Natural Sciences and Engineering Research Council of Canada; Genome Canada; Canarie","keywords":"Perl; Computer science; RDF; Class (philosophy); Programming language; World Wide Web; Software; Resource (disambiguation); License; Semantic Web; Software engineering; Artificial intelligence; Operating system","score_opus":0.024532266281811578,"score_gpt":0.25449899228227674,"score_spread":0.22996672600046517,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2141363231","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0024334297,0.00012890642,0.8960788,0.00045946,0.00020709414,0.00054511824,0.0087794745,0.084096,0.0072717196],"genre_scores_gemma":[0.040303774,0.00045671477,0.86772597,0.0009061791,0.00014035638,0.001363511,0.04687598,0.03197961,0.010247943],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9964833,0.0007774074,0.00045847445,0.00054959615,0.0014810426,0.00024999934],"domain_scores_gemma":[0.9934534,0.002669481,0.00067313394,0.0016564472,0.0012715074,0.00027615431],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007455349,0.0018020999,0.0008197199,0.0028038346,0.000989591,0.0031686875,0.004368017,0.0013343387,0.021556273],"category_scores_gemma":[0.0105353175,0.0018147686,0.0022373123,0.0018778811,0.0011526314,0.0064134495,0.0034764446,0.0030502034,0.012664567],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005823979,0.0005662879,0.0032612616,0.0030368424,0.00041783892,0.001390149,0.0015976684,0.012751428,0.02393344,0.13202535,0.33270976,0.48772752],"study_design_scores_gemma":[0.00018510401,0.00008820302,0.0010753631,0.0005656466,0.00013257559,0.0010350225,0.00024300344,0.12750332,0.056428958,0.068613335,0.74392366,0.0002057707],"about_ca_topic_score_codex":0.0029678985,"about_ca_topic_score_gemma":0.003619972,"teacher_disagreement_score":0.021556273,"about_ca_system_score_codex":0.001130185,"about_ca_system_score_gemma":0.0024371315,"threshold_uncertainty_score":0.07211298},"labels":[],"label_agreement":null},{"id":"W2141415922","doi":"10.1093/bioinformatics/btr017","title":"Computational refinement of post-translational modifications predicted from tandem mass spectrometry","year":2011,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Advanced Proteomics Techniques and Applications","field":"Chemistry","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of Toronto","funders":"Ontario Ministry of Research and Innovation; Natural Sciences and Engineering Research Council of Canada; Canadian Institutes of Health Research","keywords":"Computer science; Cluster analysis; Database search engine; Noise (video); Tandem mass spectrometry; Posttranslational modification; Proteome; Data mining; Algorithm; Artificial intelligence; Pattern recognition (psychology); Bioinformatics; Mass spectrometry; Search engine; Chemistry; Biology","score_opus":0.02368803982480069,"score_gpt":0.2529056610913337,"score_spread":0.22921762126653303,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2141415922","genre_codex":"empirical","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.51276636,0.0011753136,0.4688292,0.0009316401,0.00015120042,0.00033674436,0.0019140092,0.012036843,0.0018586455],"genre_scores_gemma":[0.5597392,0.00020252028,0.43373182,0.0004552359,0.0000792972,0.00023010247,0.00439511,0.0003954302,0.0007714457],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99872977,0.0003991679,0.00010068469,0.00039553785,0.0002623229,0.00011248847],"domain_scores_gemma":[0.99221206,0.0058172448,0.0007207478,0.00038424894,0.000632761,0.00023285748],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004116535,0.0016255071,0.0019692734,0.0023435606,0.00085528835,0.0011175388,0.002866281,0.0016954286,0.0017842266],"category_scores_gemma":[0.012546262,0.00074484025,0.0017388874,0.0015345913,0.00089653727,0.0013617714,0.0011583671,0.0015087015,0.0005151652],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0012013782,0.00026559917,0.015625427,0.00035969133,0.00024284258,0.0002749042,0.00007069435,0.89960676,0.005405882,0.0015032315,0.002884505,0.07255914],"study_design_scores_gemma":[0.0000206048,0.000032804015,0.0002443025,0.00000353729,0.000012451332,0.000022575497,0.000006387624,0.9981223,0.00081274926,0.0006011008,0.00011728125,0.0000039984707],"about_ca_topic_score_codex":0.008269764,"about_ca_topic_score_gemma":0.009191288,"teacher_disagreement_score":0.008269764,"about_ca_system_score_codex":0.0010039619,"about_ca_system_score_gemma":0.002690459,"threshold_uncertainty_score":0.021770597},"labels":[],"label_agreement":null},{"id":"W2141462396","doi":"10.1093/bioinformatics/btm629","title":"A hybrid model for robust detection of transcription factor binding sites","year":2008,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":15,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"University of Toronto; Ontario Genomics Institute; Genome Canada","keywords":"Transcription factor; Computer science; Factor (programming language); DNA binding site; Binding site; Computational biology; Biology; Genetics; Promoter; Gene; Programming language; Gene expression","score_opus":0.02511979486837297,"score_gpt":0.21658238976924557,"score_spread":0.19146259490087258,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2141462396","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.052255,0.0004799796,0.94259083,0.0004978248,0.00006928983,0.000075864344,0.0005460155,0.00089935167,0.002585885],"genre_scores_gemma":[0.8161456,0.00040656028,0.16729988,0.0005057161,0.00011481288,0.00055784744,0.0016416891,0.00037128312,0.012956625],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99932265,0.00019091084,0.000026551223,0.00024308155,0.00012841739,0.000088426714],"domain_scores_gemma":[0.9974795,0.0018170542,0.0002041406,0.000100633864,0.00029909358,0.00009965379],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0017297042,0.0011596931,0.0014557276,0.0008752192,0.00041872665,0.0014018227,0.002441731,0.0022115798,0.0036219538],"category_scores_gemma":[0.005269806,0.00060834724,0.0013275936,0.00060239463,0.0011694005,0.0009997969,0.0012403392,0.0017347819,0.0010582851],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00015085754,0.00004318762,0.0013326709,0.00005866966,0.000041404954,0.00007497232,0.00003865186,0.97624433,0.0024418456,0.006719275,0.00084465503,0.012009486],"study_design_scores_gemma":[0.0000056661374,0.00001037166,0.00008180528,0.0000026662678,0.0000042332676,0.000009917172,0.0000022724512,0.997914,0.00016617318,0.0016692455,0.00012888864,0.0000048818474],"about_ca_topic_score_codex":0.008069862,"about_ca_topic_score_gemma":0.005305419,"teacher_disagreement_score":0.008069862,"about_ca_system_score_codex":0.0013654473,"about_ca_system_score_gemma":0.0010932939,"threshold_uncertainty_score":0.016045809},"labels":[],"label_agreement":null},{"id":"W2141629057","doi":"10.1093/bioinformatics/btl030","title":"Efficient estimation of graphlet frequency distributions in protein–protein interaction networks","year":2006,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":137,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Ontario Institute for Cancer Research; University of Toronto","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Heuristics; Scalability; Heuristic; Computer science; Biological network; Software; Algorithm; Theoretical computer science; Artificial intelligence; Mathematics","score_opus":0.005303035860545529,"score_gpt":0.21728683723952533,"score_spread":0.2119838013789798,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2141629057","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.09489736,0.00029181284,0.90234655,0.00022897686,0.000015009116,0.000056083234,0.00034346973,0.0011498708,0.0006708202],"genre_scores_gemma":[0.51271564,0.00035894342,0.48345214,0.000079081954,0.00005160026,0.00017141513,0.0021250667,0.00019331533,0.00085284637],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99941444,0.00021305123,0.000027426342,0.00010329678,0.00019298618,0.00004886822],"domain_scores_gemma":[0.99227315,0.0060921717,0.00062565965,0.00044260937,0.00040691416,0.00015945612],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016086149,0.00055031467,0.0007979289,0.002244009,0.00040270516,0.00092533196,0.0012637224,0.0010808556,0.0012306102],"category_scores_gemma":[0.014144306,0.00055237964,0.0005122674,0.001565604,0.00064597453,0.001529157,0.00082329,0.00097008876,0.000753876],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002851895,0.000107349995,0.009635455,0.00022929962,0.000073925876,0.00018264768,0.0001965882,0.83930856,0.010286762,0.0106363045,0.002922788,0.1261351],"study_design_scores_gemma":[0.000007702194,0.0000083436535,0.0007788947,0.0000038113935,0.0000030040478,0.000039962815,0.000011176549,0.99236804,0.0011493354,0.0054239975,0.00020141467,0.000004322278],"about_ca_topic_score_codex":0.0029057222,"about_ca_topic_score_gemma":0.003143906,"teacher_disagreement_score":0.0029057222,"about_ca_system_score_codex":0.0010335358,"about_ca_system_score_gemma":0.0006919176,"threshold_uncertainty_score":0.008507252},"labels":[],"label_agreement":null},{"id":"W2142027180","doi":"10.1093/bioinformatics/bts716","title":"SCARPA: scaffolding reads with practical algorithms","year":2012,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genome Rearrangement Algorithms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":69,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"SickKids Foundation; Hospital for Sick Children; University of Toronto","funders":"Natural Sciences and Engineering Research Council of Canada; Canadian Institutes of Health Research; University of Toronto","keywords":"Scaffold; Contig; Computer science; Algorithm; Heuristic; Process (computing); Genome; Biology; Artificial intelligence; Programming language; Genetics","score_opus":0.021609353128708805,"score_gpt":0.2760012413538531,"score_spread":0.2543918882251443,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2142027180","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0032584714,0.00023467247,0.9559729,0.00022112565,0.0000702211,0.00014678904,0.0013170387,0.036399145,0.0023796454],"genre_scores_gemma":[0.03128113,0.00018018822,0.9575377,0.00019085415,0.00006545763,0.00043537217,0.0043244576,0.003972357,0.0020123874],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9977508,0.0006718731,0.00016141617,0.0006731067,0.0005985653,0.0001443134],"domain_scores_gemma":[0.99205244,0.004875791,0.00040459173,0.0016029307,0.00086919573,0.00019503683],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003320474,0.0017750384,0.0012024025,0.0013698302,0.0012275769,0.0019857641,0.0035193793,0.0019684245,0.023336137],"category_scores_gemma":[0.02029159,0.0013743261,0.001300073,0.0024406165,0.0012225914,0.0026216654,0.002742869,0.0028699192,0.013103426],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010452551,0.0002101913,0.0019942082,0.0017191207,0.00013386515,0.00045836504,0.0005362143,0.13787994,0.024741763,0.044768885,0.102857746,0.6836544],"study_design_scores_gemma":[0.0004089683,0.00022401387,0.00053988036,0.00015675435,0.000045748995,0.00049291115,0.00013969411,0.8002643,0.030306982,0.09174236,0.07559622,0.00008210001],"about_ca_topic_score_codex":0.0022341965,"about_ca_topic_score_gemma":0.002927397,"teacher_disagreement_score":0.023336137,"about_ca_system_score_codex":0.0008162389,"about_ca_system_score_gemma":0.002009928,"threshold_uncertainty_score":0.07806718},"labels":[],"label_agreement":null},{"id":"W2142280055","doi":"10.1093/bioinformatics/btt710","title":"ChromoHub V2: cancer genomics","year":2013,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Protein Degradation and Inhibitors","field":"Biochemistry, Genetics and Molecular Biology","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Structural Genomics Consortium; University of Toronto","funders":"Ministero dello Sviluppo Economico; Ontario Ministry of Economic Development and Innovation; Ontario Genomics Institute; Canadian Institutes of Health Research; Ontario Genomics; Genome Canada; Wellcome Trust; Eli Lilly Canada; GlaxoSmithKline; Pfizer; Eli Lilly and Company","keywords":"Chromatin; Genomics; Epigenetics; Histone; Computational biology; Biology; Genome; Cancer; Chromatin remodeling; Genetics; Gene","score_opus":0.008853294606747695,"score_gpt":0.229220323251506,"score_spread":0.2203670286447583,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2142280055","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0071418,0.0007288155,0.10138848,0.0013657025,0.00044712494,0.00045984608,0.17755456,0.6880853,0.022828303],"genre_scores_gemma":[0.1049388,0.001115094,0.17868778,0.003375559,0.0003959489,0.002280721,0.49224696,0.18662865,0.030330421],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9993369,0.00009749757,0.00003424681,0.00017452677,0.0002538761,0.00010301597],"domain_scores_gemma":[0.99886096,0.00047383527,0.00007449574,0.00019343523,0.00020010879,0.00019702762],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016342128,0.0014208679,0.0011257194,0.001935425,0.00062615867,0.0019590668,0.002732826,0.00092030666,0.090784706],"category_scores_gemma":[0.0037120674,0.0008821549,0.0010909291,0.0019893206,0.0004588766,0.0014384408,0.0022900682,0.0017298359,0.043565005],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001834553,0.000100777805,0.0030832037,0.0010277473,0.00018150837,0.0002900241,0.00017627343,0.0016865401,0.006698904,0.0080137355,0.90789306,0.06901358],"study_design_scores_gemma":[0.0015281924,0.00024662106,0.01091975,0.00024415867,0.00021691562,0.00086938863,0.000114502676,0.058262914,0.04956036,0.032703467,0.8450487,0.00028490947],"about_ca_topic_score_codex":0.0043860953,"about_ca_topic_score_gemma":0.0041309814,"teacher_disagreement_score":0.090784706,"about_ca_system_score_codex":0.0012441386,"about_ca_system_score_gemma":0.0016005154,"threshold_uncertainty_score":0.3037051},"labels":[],"label_agreement":null},{"id":"W2142295792","doi":"10.1093/bioinformatics/btu522","title":"Individual-level analysis of differential expression of genes and pathways for personalized medicine","year":2014,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":138,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"","keywords":"Normalization (sociology); Gene; Personalized medicine; Population; Computational biology; Biology; Computer science; Bioinformatics; Genetics; Medicine","score_opus":0.04343514193106566,"score_gpt":0.27577847301456077,"score_spread":0.2323433310834951,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2142295792","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.28913808,0.0020949692,0.6898505,0.00083228434,0.00008105588,0.00016261941,0.010784687,0.0032750163,0.0037808057],"genre_scores_gemma":[0.8359257,0.00036196402,0.15748793,0.00018337097,0.0000690976,0.00016529903,0.0044506653,0.00018902766,0.0011670054],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9991861,0.00028093986,0.00003887145,0.00026325948,0.00017186617,0.00005895641],"domain_scores_gemma":[0.99754936,0.0014585146,0.000292945,0.00036398644,0.00024462142,0.00009059362],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0019295609,0.0004226664,0.00067292596,0.0011567071,0.00028363772,0.00072551344,0.000519345,0.000387596,0.004359314],"category_scores_gemma":[0.0041888105,0.00015398688,0.00048766972,0.0018618124,0.0005204126,0.00052698544,0.0005696659,0.0007626437,0.0009977695],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0021264977,0.00043740915,0.14197886,0.0018632174,0.0007700864,0.0005193946,0.00034223418,0.08754752,0.21257597,0.016539771,0.013940339,0.52135867],"study_design_scores_gemma":[0.0001366563,0.000790837,0.22086729,0.00011841875,0.00048598053,0.0016680948,0.00027367525,0.5322359,0.14512274,0.07086844,0.027247347,0.00018458905],"about_ca_topic_score_codex":0.0006121907,"about_ca_topic_score_gemma":0.0009679063,"teacher_disagreement_score":0.004359314,"about_ca_system_score_codex":0.00043867112,"about_ca_system_score_gemma":0.0006488276,"threshold_uncertainty_score":0.014583349},"labels":[],"label_agreement":null},{"id":"W2142319534","doi":"10.1093/bioinformatics/btg072","title":"Using multiple interdependencyto separate functional from phylogenetic correlations in protein alignments","year":2003,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Protein Structure and Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":126,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University Health Network; Ontario Institute for Cancer Research","funders":"","keywords":"Phylogenetic tree; Multiple sequence alignment; Sequence (biology); Computational biology; Substitution (logic); Cluster analysis; Biology; Sequence alignment; Phylogenetics; Protein superfamily; Protein sequencing; Evolutionary biology; Computer science; Genetics; Peptide sequence; Artificial intelligence; Gene","score_opus":0.019866193769249967,"score_gpt":0.2498401684537612,"score_spread":0.22997397468451122,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2142319534","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.4511411,0.0009308051,0.53982913,0.00040046562,0.000051240535,0.00021040297,0.0018255834,0.002332809,0.0032785232],"genre_scores_gemma":[0.84377223,0.00027607998,0.15272446,0.0000771535,0.00006564313,0.00016956757,0.0017407173,0.00035753308,0.000816528],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9945379,0.0017206471,0.0005510783,0.0012897754,0.001606153,0.00029449057],"domain_scores_gemma":[0.97193754,0.017750442,0.0042635445,0.003351074,0.0021521328,0.0005452077],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0040673725,0.00094960287,0.0012288784,0.0041293968,0.000857026,0.00151648,0.001305243,0.0009142913,0.0026160246],"category_scores_gemma":[0.025172532,0.0006591019,0.00075675873,0.00353424,0.0009820393,0.002657823,0.0020287288,0.0015015268,0.0008772233],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0018103512,0.00042231043,0.33297673,0.0016926259,0.0009572634,0.0015201909,0.0008530722,0.08511404,0.115432,0.020376617,0.0060458696,0.43279892],"study_design_scores_gemma":[0.00007265139,0.0002693711,0.13652559,0.00009532439,0.0002668237,0.0015603246,0.00016971982,0.7592267,0.060206145,0.03600492,0.0054437616,0.00015873804],"about_ca_topic_score_codex":0.0008207639,"about_ca_topic_score_gemma":0.0021765262,"teacher_disagreement_score":0.0041293968,"about_ca_system_score_codex":0.0008244064,"about_ca_system_score_gemma":0.0008280677,"threshold_uncertainty_score":0.021510601},"labels":[],"label_agreement":null},{"id":"W2142963728","doi":"10.1093/bioinformatics/btt602","title":"Exploring high dimensional data with Butterfly: a novel classification algorithm based on discrete dynamical systems","year":2013,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Neural Networks and Applications","field":"Computer Science","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Queen's University; University Health Network","funders":"Government of Ontario; Eli Lilly Canada; University Health Network; Natural Sciences and Engineering Research Council of Canada; Ottawa Hospital Research Institute; Ontario Institute for Cancer Research; Eli Lilly and Company","keywords":"Computer science; Cluster analysis; Representation (politics); Set (abstract data type); Univariate; Dynamical systems theory; Visualization; Algorithm; Artificial intelligence; External Data Representation; Pattern recognition (psychology); Data mining; Machine learning; Multivariate statistics","score_opus":0.09020328266180674,"score_gpt":0.2489412473830473,"score_spread":0.15873796472124055,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2142963728","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0067024855,0.000088917346,0.98890823,0.00022693934,0.000039222316,0.0000751431,0.00014675506,0.0030130139,0.0007993199],"genre_scores_gemma":[0.076206714,0.00009552614,0.91996723,0.00017273055,0.00004259001,0.00017706196,0.0005132246,0.00037018178,0.0024547728],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9994653,0.0000692611,0.000035854722,0.00016865082,0.00021428781,0.000046559653],"domain_scores_gemma":[0.99898,0.0004943057,0.000088950634,0.00013920372,0.00022984961,0.000067677625],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008631286,0.00091585424,0.0008214657,0.0015185738,0.00075050205,0.0017245756,0.0019741238,0.0010854795,0.0070700287],"category_scores_gemma":[0.0035078146,0.0003817624,0.0008298228,0.0010854882,0.00093934895,0.002256695,0.0019620499,0.0012318745,0.001701405],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003177168,0.00010739547,0.0038614045,0.00019045609,0.00010575063,0.0002943631,0.00045775005,0.14703412,0.015520646,0.034140516,0.018606033,0.7793639],"study_design_scores_gemma":[0.000031682426,0.000036513266,0.00032411478,0.000017571308,0.000009126837,0.00012419002,0.000040256662,0.9759975,0.00394263,0.014703274,0.0047542094,0.000018820332],"about_ca_topic_score_codex":0.006188132,"about_ca_topic_score_gemma":0.0059010605,"teacher_disagreement_score":0.0070700287,"about_ca_system_score_codex":0.00094344077,"about_ca_system_score_gemma":0.00096759835,"threshold_uncertainty_score":0.0236516},"labels":[],"label_agreement":null},{"id":"W2143093088","doi":"10.1093/bioinformatics/btp378","title":"ISOLATE: a computational strategy for identifying the primary origin of cancers using high-throughput sequencing","year":2009,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Cancer Diagnosis and Treatment","field":"Medicine","cited_by":40,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Computational biology; Biology; Profiling (computer programming); Sample size determination; Gene expression profiling; Computer science; Gene; Machine learning; Artificial intelligence; Gene expression; Genetics; Statistics; Mathematics","score_opus":0.09569313300894752,"score_gpt":0.35252440953347447,"score_spread":0.256831276524527,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2143093088","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.030974822,0.00013570824,0.96053964,0.00046186693,0.000050129514,0.00017493899,0.0011464123,0.0055224667,0.0009939934],"genre_scores_gemma":[0.26943496,0.00018871279,0.72026384,0.0005911005,0.00016973099,0.00093654793,0.005131607,0.00091406,0.0023694674],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9995276,0.00017861256,0.000023463504,0.00011736622,0.00011253861,0.000040548726],"domain_scores_gemma":[0.9970824,0.0022172094,0.00013508467,0.00021602698,0.00022469358,0.00012466143],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0019290633,0.001116254,0.0012336945,0.0016180922,0.00068789703,0.0011639739,0.0021279177,0.001291302,0.0037686364],"category_scores_gemma":[0.0070000123,0.00071294286,0.0015995953,0.0009910252,0.00064557174,0.00091780216,0.001240467,0.0013074434,0.0010379351],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007805157,0.0002761877,0.009493216,0.00030888384,0.00044397547,0.00040086493,0.00014171583,0.79173845,0.005177949,0.008105471,0.014168219,0.16896464],"study_design_scores_gemma":[0.000028709224,0.000023675233,0.00027636855,0.000004136899,0.000017968841,0.000028112676,0.000008536099,0.99511683,0.0005683701,0.0034016115,0.0005194199,0.000006272618],"about_ca_topic_score_codex":0.004867741,"about_ca_topic_score_gemma":0.009519019,"teacher_disagreement_score":0.004867741,"about_ca_system_score_codex":0.00066371,"about_ca_system_score_gemma":0.0021105926,"threshold_uncertainty_score":0.012607276},"labels":[],"label_agreement":null},{"id":"W2143124024","doi":"10.1093/bioinformatics/btn416","title":"ZOOM! Zillions of oligos mapped","year":2008,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":211,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo; Western University","funders":"National Human Genome Research Institute","keywords":"Zoom; Computer science; Reference genome; Human genome; Software; Sequence assembly; Computational biology; Genome; DNA sequencing; Biology; Genetics; Gene; Operating system; Transcriptome","score_opus":0.016696739985166367,"score_gpt":0.2223268851516874,"score_spread":0.205630145166521,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2143124024","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.056316137,0.0023795455,0.5736812,0.003621862,0.003185958,0.0007715317,0.026551934,0.26664856,0.066843346],"genre_scores_gemma":[0.1653832,0.0015487621,0.7344662,0.0022802376,0.00036753557,0.0010583772,0.016238399,0.031080421,0.04757692],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99964035,0.000048907383,0.000025578647,0.00008744823,0.00014399295,0.00005366428],"domain_scores_gemma":[0.99848515,0.0005791261,0.00008359856,0.0003901424,0.00029432282,0.0001676462],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00077145785,0.0013253436,0.00069865567,0.0018835927,0.0009343659,0.0011810282,0.0013157437,0.00093151926,0.08947511],"category_scores_gemma":[0.0040381406,0.0007330959,0.0007781969,0.0013920048,0.0005272278,0.0016846604,0.0021923804,0.0015053262,0.017588686],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0017278629,0.00017912501,0.0037375963,0.0013503244,0.00017381155,0.0013811664,0.0007456563,0.0074637565,0.19805206,0.01056185,0.287815,0.48681182],"study_design_scores_gemma":[0.0004594099,0.00037962978,0.012033878,0.00047345823,0.00015538205,0.0015374421,0.00048664238,0.09928761,0.34441736,0.027733663,0.5126563,0.00037919558],"about_ca_topic_score_codex":0.0027995317,"about_ca_topic_score_gemma":0.0041961204,"teacher_disagreement_score":0.08947511,"about_ca_system_score_codex":0.0006710933,"about_ca_system_score_gemma":0.0006790108,"threshold_uncertainty_score":0.29932404},"labels":[],"label_agreement":null},{"id":"W2143227538","doi":"10.1093/bioinformatics/btq637","title":"Differential effects of chromatin regulators and transcription factors on gene regulation: a nucleosomal perspective","year":2010,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"Canadian Institutes of Health Research","keywords":"Nucleosome; Chromatin; Biology; Gene; Transcription factor; Promoter; Transcriptional regulation; Genetics; Regulation of gene expression; Saccharomyces cerevisiae; Computational biology; Transcription (linguistics); Histone; Gene expression","score_opus":0.0034549751453332596,"score_gpt":0.1960566982394446,"score_spread":0.19260172309411133,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2143227538","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.67273265,0.06163088,0.22361565,0.002625512,0.0002692311,0.00007589298,0.0081382515,0.0013256671,0.029586285],"genre_scores_gemma":[0.9251448,0.013013012,0.052770194,0.00071382435,0.00009917499,0.000045565295,0.003949272,0.00017478378,0.004089348],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9998516,0.000027397416,0.000007963421,0.00006701746,0.000029234405,0.000016766946],"domain_scores_gemma":[0.9997811,0.000084736195,0.00003619982,0.000031252363,0.000031997333,0.000034769862],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00032513472,0.0002599013,0.00039168634,0.00044912423,0.00019303804,0.00056020275,0.00033805257,0.00022750211,0.002341862],"category_scores_gemma":[0.00043384216,0.00012434945,0.00021735679,0.0009758929,0.00036160694,0.00054077053,0.00039038385,0.0003211517,0.00044661755],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009314742,0.000051135048,0.027425239,0.0016125313,0.000116758085,0.00039959355,0.00037855818,0.003704841,0.84444493,0.026605748,0.0017500797,0.09257909],"study_design_scores_gemma":[0.000087821725,0.0004989048,0.34091052,0.00030430325,0.0004848615,0.0021911787,0.00074801676,0.027259788,0.471873,0.05695057,0.098571084,0.00011994685],"about_ca_topic_score_codex":0.0021233582,"about_ca_topic_score_gemma":0.0025269745,"teacher_disagreement_score":0.002341862,"about_ca_system_score_codex":0.00046186356,"about_ca_system_score_gemma":0.00018910282,"threshold_uncertainty_score":0.007834315},"labels":[],"label_agreement":null},{"id":"W2143438047","doi":"10.1093/bioinformatics/btr224","title":"Mapping ancestral genomes with massive gene loss: A matrix sandwich problem","year":2011,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genome Rearrangement Algorithms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"Natural Sciences and Engineering Research Council of Canada; Agence Nationale de la Recherche","keywords":"Genome; Extant taxon; Biology; Gene; Computational biology; Heuristic; Gene duplication; Genetics; Evolutionary biology; Computer science; Artificial intelligence","score_opus":0.018437889191453578,"score_gpt":0.22572372873115842,"score_spread":0.20728583953970484,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2143438047","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.119687624,0.0001310543,0.876386,0.0005084287,0.000026011623,0.00009353907,0.00025312626,0.00074244133,0.0021717965],"genre_scores_gemma":[0.31667912,0.00013782208,0.67950463,0.00015816018,0.00002829173,0.00014679087,0.0009922958,0.00021262126,0.0021403173],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9994203,0.0001944017,0.000030407136,0.00018721496,0.00010099991,0.0000666863],"domain_scores_gemma":[0.99730814,0.001858246,0.00017586624,0.00031414008,0.00017065613,0.00017301524],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001210551,0.0008628334,0.0009822302,0.00084692676,0.00075373094,0.0013374852,0.0017826374,0.0015171241,0.005971674],"category_scores_gemma":[0.006245815,0.00079213065,0.0010796691,0.0010995122,0.0009642238,0.002282309,0.0019725207,0.0014643304,0.0007791422],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005348752,0.00024006549,0.004483825,0.00035089918,0.00008920479,0.0007100649,0.00039319127,0.7196151,0.0163091,0.07143383,0.004995082,0.18084483],"study_design_scores_gemma":[0.00009060389,0.00011857456,0.00068565615,0.000021607155,0.000028820903,0.00035772135,0.000119963704,0.91545767,0.006934821,0.074468896,0.0016907474,0.00002485809],"about_ca_topic_score_codex":0.002170656,"about_ca_topic_score_gemma":0.0018583243,"teacher_disagreement_score":0.005971674,"about_ca_system_score_codex":0.000645696,"about_ca_system_score_gemma":0.0009871385,"threshold_uncertainty_score":0.019977212},"labels":[],"label_agreement":null},{"id":"W2143693563","doi":"10.1093/bioinformatics/btu373","title":"Mirsynergy: detecting synergistic miRNA regulatory modules by overlapping neighbourhood expansion","year":2014,"lang":"en","type":"article","venue":"Bioinformatics","topic":"MicroRNA in disease regulation","field":"Biochemistry, Genetics and Molecular Biology","cited_by":84,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Bioconductor; microRNA; Computational biology; Cluster analysis; Computer science; Biology; Gene regulatory network; Gene; Gene expression; Machine learning; Genetics","score_opus":0.006147777979144508,"score_gpt":0.21085898556060106,"score_spread":0.20471120758145656,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2143693563","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.17220901,0.00066510437,0.8127729,0.00014130455,0.00004266403,0.00025050304,0.0011351671,0.009841996,0.0029413528],"genre_scores_gemma":[0.42420056,0.00018315318,0.5692021,0.00010429496,0.000024871195,0.00036880228,0.0025565466,0.00080565887,0.0025539564],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9991617,0.0002042582,0.000043428572,0.00029773946,0.00022240082,0.00007045376],"domain_scores_gemma":[0.9993167,0.00029466796,0.00012106581,0.00011305126,0.00010111183,0.000053444994],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011433864,0.00072845607,0.00083025394,0.0015370307,0.0006751263,0.00074826216,0.001232635,0.0006328618,0.0029290037],"category_scores_gemma":[0.0025801407,0.00049236254,0.0010240956,0.0009224996,0.0005137674,0.0006523364,0.0016671505,0.0004103706,0.0009938319],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0017461325,0.0003098765,0.032626588,0.0011321486,0.00094071194,0.000659322,0.00085805764,0.23696116,0.18834746,0.01788069,0.013535317,0.50500256],"study_design_scores_gemma":[0.00009884766,0.00021296657,0.0069433344,0.000029957935,0.00011528426,0.00047719013,0.00011376331,0.93270534,0.043515164,0.00903549,0.0066899536,0.00006272297],"about_ca_topic_score_codex":0.0016839491,"about_ca_topic_score_gemma":0.0042786105,"teacher_disagreement_score":0.0029290037,"about_ca_system_score_codex":0.0005575072,"about_ca_system_score_gemma":0.00088028243,"threshold_uncertainty_score":0.009798467},"labels":[],"label_agreement":null},{"id":"W2143801201","doi":"10.1093/bioinformatics/btn463","title":"Improving subcellular localization prediction using text classification and the gene ontology","year":2008,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Machine Learning in Bioinformatics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":44,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Natural Sciences and Engineering Research Council of Canada; University of Alberta","keywords":"Gene ontology; Computer science; Hierarchy; Mechanism (biology); Protein function prediction; Ontology; Function (biology); Artificial intelligence; Data mining; Sequence (biology); Protein sequencing; Information retrieval; Natural language processing; Computational biology; Machine learning; Gene; Protein function; Biology; Peptide sequence; Genetics; Gene expression","score_opus":0.015766448245103544,"score_gpt":0.22927907549808127,"score_spread":0.21351262725297773,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2143801201","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.4635204,0.004625105,0.47680056,0.0038450228,0.00062147167,0.00077193225,0.021445615,0.021086948,0.0072830105],"genre_scores_gemma":[0.50310487,0.0017754008,0.4554005,0.0004888122,0.0006628499,0.00042960315,0.034569513,0.0003276881,0.0032407294],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9987503,0.0002881712,0.00021888416,0.0002438811,0.00042038405,0.00007834631],"domain_scores_gemma":[0.9877203,0.007587499,0.0012316555,0.0005773404,0.0026355635,0.000247623],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002007649,0.0014509347,0.0009944419,0.008045859,0.000565708,0.0012538509,0.0011187162,0.0011267108,0.002855836],"category_scores_gemma":[0.012881912,0.00019404711,0.00093241874,0.0065144515,0.00036174833,0.0027032567,0.00079311593,0.0010047388,0.0024146573],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009998189,0.000853808,0.04836559,0.001583748,0.00023914006,0.0010030824,0.00023017335,0.022925023,0.051164072,0.0018363162,0.032658402,0.83814085],"study_design_scores_gemma":[0.00034439628,0.0004985646,0.034787185,0.00025172878,0.00048971066,0.0010576862,0.00046029157,0.8528111,0.07986943,0.012826492,0.016476745,0.00012663283],"about_ca_topic_score_codex":0.0039688866,"about_ca_topic_score_gemma":0.0034047219,"teacher_disagreement_score":0.008045859,"about_ca_system_score_codex":0.0005435321,"about_ca_system_score_gemma":0.001117004,"threshold_uncertainty_score":0.010617614},"labels":[],"label_agreement":null},{"id":"W2143859503","doi":"10.1093/bioinformatics/btr100","title":"Human variation database: an open-source database template for genomic discovery","year":2011,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Rare Diseases","field":"Biochemistry, Genetics and Molecular Biology","cited_by":12,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"BC Cancer Agency","funders":"Genome British Columbia; Michael Smith Health Research BC; Genome Canada","keywords":"Computer science; Database; Database schema; Java; Source code; Application programming interface; Pooling; Software; Interface (matter); Database application; Graphical user interface; Modular design; Open source; Database testing; Information retrieval; Database design; Operating system","score_opus":0.03992812043407963,"score_gpt":0.28016559752281417,"score_spread":0.24023747708873455,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2143859503","genre_codex":"dataset","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.001946827,0.0017601039,0.19099097,0.0016235972,0.0007242787,0.0006436362,0.69839686,0.090280585,0.013633257],"genre_scores_gemma":[0.009674876,0.0012840979,0.18231404,0.0014254256,0.00024522285,0.0016705594,0.7731602,0.0226263,0.0075992798],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99687093,0.00058571016,0.0006943081,0.0008591662,0.0008009371,0.00018885017],"domain_scores_gemma":[0.9931433,0.0025151025,0.00082207395,0.0016342049,0.0011334952,0.0007518034],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.006489173,0.0015986062,0.0018840223,0.004018043,0.0010292537,0.0039958768,0.005337215,0.0027165082,0.06093091],"category_scores_gemma":[0.018891871,0.0015241359,0.0013865603,0.006619706,0.00062910863,0.0030303434,0.0054786666,0.0026334946,0.06944311],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0012271582,0.000093355186,0.0033491931,0.003601529,0.00029053376,0.00067187665,0.00050816126,0.0020664039,0.009131729,0.020094594,0.8654835,0.09348201],"study_design_scores_gemma":[0.000364072,0.00008081439,0.0032564925,0.00047541372,0.00012994776,0.00097157416,0.00012458365,0.0029217803,0.0060841264,0.022752343,0.96271074,0.0001279657],"about_ca_topic_score_codex":0.003217576,"about_ca_topic_score_gemma":0.0037805212,"teacher_disagreement_score":0.99466276,"about_ca_system_score_codex":0.0009539706,"about_ca_system_score_gemma":0.003723063,"threshold_uncertainty_score":0.20383418},"labels":[],"label_agreement":null},{"id":"W2144015117","doi":"10.1093/bioinformatics/btu277","title":"Deep learning of the tissue-regulated splicing code","year":2014,"lang":"en","type":"article","venue":"Bioinformatics","topic":"RNA Research and Splicing","field":"Biochemistry, Genetics and Molecular Biology","cited_by":487,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto; Canadian Institute for Advanced Research","funders":"Natural Sciences and Engineering Research Council of Canada; Canadian Institutes of Health Research","keywords":"Computer science; Context (archaeology); Artificial intelligence; RNA splicing; Deep learning; Source code; Machine learning; Alternative splicing; Hyperparameter; Computational biology; Graphics processing unit; Biology; Gene; RNA; Genetics; Exon","score_opus":0.008313087259342699,"score_gpt":0.25031227657918176,"score_spread":0.24199918931983905,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2144015117","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.23951918,0.00042586715,0.75469565,0.0008370077,0.000048567166,0.000021973428,0.0008227018,0.0009909426,0.0026380995],"genre_scores_gemma":[0.92119795,0.00028701153,0.073603585,0.0001958406,0.000029228639,0.000049816175,0.0010462053,0.00007034257,0.0035201118],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99990106,0.000021343842,0.0000042961806,0.00003260711,0.000022388365,0.000018297247],"domain_scores_gemma":[0.99952066,0.00022921992,0.000055541834,0.00005260737,0.0001095068,0.000032459393],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004916419,0.0003675195,0.00034420597,0.00031036243,0.0001682395,0.0004166865,0.00075822335,0.0005652695,0.0013431208],"category_scores_gemma":[0.0015983689,0.00026016834,0.00035150745,0.0004325356,0.0005058513,0.00072441314,0.0004358663,0.0011548728,0.00039451034],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00010065322,0.000051849947,0.0044626896,0.00006597708,0.00004239759,0.000100649384,0.000050036193,0.89457816,0.014850147,0.01714457,0.0019490478,0.06660387],"study_design_scores_gemma":[0.0000034710874,0.0000068812074,0.0003800974,0.00000351262,0.0000031099291,0.000013442679,0.0000032133307,0.9885462,0.0014909143,0.009365415,0.0001806613,0.000003051852],"about_ca_topic_score_codex":0.004104778,"about_ca_topic_score_gemma":0.006095548,"teacher_disagreement_score":0.004104778,"about_ca_system_score_codex":0.0007366023,"about_ca_system_score_gemma":0.0006814935,"threshold_uncertainty_score":0.008161783},"labels":[],"label_agreement":null},{"id":"W2144222696","doi":"10.1093/bioinformatics/btu346","title":"CLImAT: accurate detection of copy number alteration and loss of heterozygosity in impure and aneuploid tumor samples using whole-genome sequencing data","year":2014,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Cancer Genomics and Diagnostics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":55,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"BC Cancer Agency; National Natural Science Foundation of China","keywords":"Loss of heterozygosity; Biology; Genome; Aneuploidy; Genetics; Computational biology; Whole genome sequencing; DNA sequencing; Ploidy; Normalization (sociology); Copy-number variation; Allele; Gene; Chromosome","score_opus":0.02941690551158587,"score_gpt":0.27309311595221203,"score_spread":0.24367621044062615,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2144222696","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.19379613,0.0017319078,0.4977548,0.0011400137,0.00026129026,0.00064265844,0.08960049,0.20745754,0.0076152226],"genre_scores_gemma":[0.3496698,0.0006269907,0.54800683,0.0006575497,0.00012128183,0.00095535757,0.086557165,0.009511706,0.0038931833],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9990283,0.00015081235,0.000056984172,0.00042674027,0.0002746244,0.00006261311],"domain_scores_gemma":[0.99730265,0.0012875374,0.00048080453,0.00033756136,0.00043200952,0.00015941377],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0022858547,0.0013987261,0.0013112369,0.0027218342,0.0011192061,0.0017575803,0.0017314637,0.0009819092,0.006639864],"category_scores_gemma":[0.007469038,0.0008611562,0.0014277705,0.0021613243,0.0006471944,0.0011132514,0.0015272944,0.0011342722,0.0030782204],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0033159396,0.00046864033,0.13495632,0.004035296,0.0022300605,0.0026463084,0.0013106717,0.16980821,0.13366205,0.009948618,0.2665931,0.27102476],"study_design_scores_gemma":[0.00031866925,0.00033048276,0.054064665,0.00015178947,0.00027681186,0.0019063685,0.00023670764,0.8253235,0.065118775,0.010746829,0.041261762,0.0002637302],"about_ca_topic_score_codex":0.005417542,"about_ca_topic_score_gemma":0.013416997,"teacher_disagreement_score":0.006639864,"about_ca_system_score_codex":0.0007293883,"about_ca_system_score_gemma":0.0016107828,"threshold_uncertainty_score":0.022212565},"labels":[],"label_agreement":null},{"id":"W2144514889","doi":"10.1093/bioinformatics/btl125","title":"Dragon Promoter Mapper (DPM): a Bayesian framework for modelling promoter structures","year":2006,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Guelph","funders":"","keywords":"Promoter; Enhancer; Motif (music); Computational biology; Sequence motif; Gene; Bayesian probability; Computer science; Biology; Genetics; Artificial intelligence; Transcription factor; Gene expression; Physics","score_opus":0.007655762186004872,"score_gpt":0.22247093817564653,"score_spread":0.21481517598964167,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2144514889","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0018767603,0.0002422473,0.99288493,0.00013126667,0.00002653748,0.000030267473,0.001730665,0.00231223,0.0007650178],"genre_scores_gemma":[0.09156537,0.0008502464,0.8933579,0.00012604371,0.000072139075,0.0005086942,0.005987202,0.0014429528,0.0060894694],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9992034,0.00030097077,0.000042522828,0.00020085875,0.00019864953,0.00005363407],"domain_scores_gemma":[0.99872714,0.00085716246,0.00010666409,0.000108093634,0.00012481704,0.00007612274],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0018355541,0.0011073264,0.0011274248,0.0014404977,0.0007556727,0.0015772599,0.003286522,0.0019490775,0.013806178],"category_scores_gemma":[0.0049559535,0.0012182131,0.001376338,0.0013475682,0.00072532985,0.0013476606,0.001035185,0.0019478776,0.0046562147],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00038519927,0.00006906835,0.003272057,0.0007840708,0.00014199232,0.00040532832,0.00030464042,0.6481186,0.010015431,0.09623183,0.026300758,0.21397094],"study_design_scores_gemma":[0.000033184537,0.000019680436,0.00028740676,0.000040246643,0.000023474899,0.000113873975,0.00001960702,0.923877,0.003248815,0.052451532,0.019849189,0.000035996065],"about_ca_topic_score_codex":0.0062673385,"about_ca_topic_score_gemma":0.008926296,"teacher_disagreement_score":0.013806178,"about_ca_system_score_codex":0.001093907,"about_ca_system_score_gemma":0.0016165983,"threshold_uncertainty_score":0.04618621},"labels":[],"label_agreement":null},{"id":"W2144629166","doi":"10.1093/bioinformatics/bts078","title":"WaVPeak: picking NMR peaks through wavelet-based smoothing and volume-based filtering","year":2012,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Protein Structure and Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":64,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Hong Kong University of Science and Technology; York University; King Abdullah University of Science and Technology; Renmin University of China; National Natural Science Foundation of China; Fundamental Research Funds for the Central Universities; University of Toronto","keywords":"Smoothing; Wavelet; Filter (signal processing); Computer science; False positive paradox; Benchmark (surveying); Maxima and minima; Noise (video); Spectral line; Volume (thermodynamics); Artificial intelligence; Nuclear magnetic resonance; Pattern recognition (psychology); Algorithm; Biological system; Mathematics; Physics; Computer vision","score_opus":0.01111331327904974,"score_gpt":0.23334582484257474,"score_spread":0.222232511563525,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2144629166","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.11294005,0.008168924,0.65821886,0.0007963661,0.0012063807,0.0019757336,0.06671339,0.14477144,0.0052087517],"genre_scores_gemma":[0.08934059,0.0012127848,0.75957143,0.000327505,0.00016257171,0.00081271084,0.1417114,0.0031222878,0.0037388392],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9971353,0.00031676862,0.00027233665,0.00088066334,0.0011270082,0.00026801345],"domain_scores_gemma":[0.99657476,0.0011304118,0.0003037371,0.0009581803,0.0008606914,0.00017216703],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002692764,0.004720548,0.0027410514,0.006245359,0.0012110197,0.002753339,0.0065589002,0.0032487465,0.0053559802],"category_scores_gemma":[0.00783195,0.0010162818,0.0032809444,0.004378462,0.0009896368,0.002188365,0.00326601,0.0026502914,0.0057042595],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0019508596,0.0008850836,0.0067894533,0.00285674,0.00071323756,0.0006431151,0.00020499794,0.1001467,0.04282711,0.0027765608,0.16113956,0.67906654],"study_design_scores_gemma":[0.00055322365,0.0007443537,0.010487803,0.0002427458,0.00021129579,0.0016773326,0.00020561133,0.8339377,0.08389689,0.007556917,0.06021756,0.00026852527],"about_ca_topic_score_codex":0.014324543,"about_ca_topic_score_gemma":0.018284265,"teacher_disagreement_score":0.014324543,"about_ca_system_score_codex":0.0011120436,"about_ca_system_score_gemma":0.0020553833,"threshold_uncertainty_score":0.028482318},"labels":[],"label_agreement":null},{"id":"W2145023212","doi":"10.1093/bioinformatics/btr657","title":"Prediction and analysis of nucleotide-binding residues using sequence and sequence-derived structural descriptors","year":2011,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Machine Learning in Bioinformatics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":137,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Killam Trusts","keywords":"Sequence (biology); Computational biology; Sequence alignment; Peptide sequence; Sequence analysis; Nucleotide; Multiple sequence alignment; Protein sequencing; Binding site; Sequence logo; Sequence motif; Consensus sequence; Conserved sequence; Biology; Biochemistry; Gene","score_opus":0.05324823264434206,"score_gpt":0.2742169801333416,"score_spread":0.22096874748899956,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2145023212","genre_codex":"empirical","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.77254283,0.0017132353,0.19591454,0.00036002145,0.00006279009,0.00021261696,0.020635717,0.0052239816,0.0033341763],"genre_scores_gemma":[0.8655009,0.0007592326,0.0852695,0.00010360051,0.000045283716,0.00015043253,0.046374694,0.00030587314,0.001490436],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9997222,0.000053990756,0.000018104636,0.00008743862,0.00008820404,0.000030098621],"domain_scores_gemma":[0.99922323,0.00034407864,0.00014152203,0.000056431505,0.00015196671,0.00008272071],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00069431105,0.0008379219,0.0007504874,0.0016990176,0.00029911858,0.0005180706,0.000573416,0.0005378157,0.002573912],"category_scores_gemma":[0.0018199405,0.00021411608,0.00054852315,0.0015091103,0.00022836468,0.00050354144,0.00039523985,0.00064873946,0.0017033175],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.002004312,0.0011494147,0.1426285,0.0012999576,0.00041253312,0.0011662105,0.00014958867,0.4579119,0.14230533,0.0030123182,0.01946664,0.22849333],"study_design_scores_gemma":[0.000060534327,0.0002565118,0.017838651,0.00004445552,0.00011381003,0.00037418006,0.000047349782,0.9467272,0.027939165,0.0023776556,0.0041860924,0.000034389766],"about_ca_topic_score_codex":0.0025692398,"about_ca_topic_score_gemma":0.004739906,"teacher_disagreement_score":0.002573912,"about_ca_system_score_codex":0.0005088843,"about_ca_system_score_gemma":0.00082530983,"threshold_uncertainty_score":0.008610606},"labels":[],"label_agreement":null},{"id":"W2145258109","doi":"10.1093/bioinformatics/btg134","title":"Genescript: DNA sequence annotation pipeline","year":2003,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":12,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Hospital for Sick Children","funders":"","keywords":"Annotation; Pipeline (software); Computer science; Sequence (biology); Download; DNA sequencing; Computational biology; DNA; Sequence analysis; Information retrieval; Data mining; Artificial intelligence; World Wide Web; Genetics; Programming language; Biology","score_opus":0.022625402548178693,"score_gpt":0.24450245365557152,"score_spread":0.22187705110739284,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2145258109","genre_codex":"dataset","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0018221692,0.0006387183,0.14451337,0.00062172295,0.000678207,0.0011126667,0.60650045,0.23456131,0.0095515],"genre_scores_gemma":[0.0075477674,0.0005941746,0.22211744,0.00083522225,0.00019037763,0.0038761257,0.70564735,0.046445668,0.012745885],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9986891,0.00026082576,0.00021452854,0.00041219126,0.00025986688,0.00016348541],"domain_scores_gemma":[0.9983577,0.00053365174,0.00022898531,0.00023694227,0.0004313688,0.00021129345],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003048148,0.0037730895,0.0026924529,0.0040341904,0.0015165356,0.0026365332,0.004964534,0.0018269783,0.12333555],"category_scores_gemma":[0.0052404897,0.0031850836,0.002695735,0.0032998163,0.0006771832,0.0021350353,0.0019088663,0.0042985715,0.17059891],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0019508622,0.00016506074,0.0009826705,0.0043956693,0.00027316104,0.0003513046,0.00032416207,0.001288966,0.047819786,0.0044737225,0.8780653,0.05990937],"study_design_scores_gemma":[0.0010420086,0.00032790322,0.0031867088,0.00076947105,0.00037432832,0.0006620006,0.00021528323,0.012198778,0.06861785,0.018191487,0.894019,0.00039512827],"about_ca_topic_score_codex":0.002698072,"about_ca_topic_score_gemma":0.0031286667,"teacher_disagreement_score":0.12333555,"about_ca_system_score_codex":0.001246667,"about_ca_system_score_gemma":0.0037805787,"threshold_uncertainty_score":0.4125985},"labels":[],"label_agreement":null},{"id":"W2145433078","doi":"10.1093/bioinformatics/btm169","title":"Polyploids, genome halving and phylogeny","year":2007,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Chromosomal and Genetic Variations","field":"Agricultural and Biological Sciences","cited_by":18,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Ottawa","funders":"Natural Sciences and Engineering Research Council of Canada; Canadian Institute for Advanced Research","keywords":"Genome; Polyploid; Ploidy; Phylogenetics; Biology; Evolutionary biology; Genomics; Genome evolution; Computational biology; Genetics; Gene","score_opus":0.014016557060256464,"score_gpt":0.20459001863550538,"score_spread":0.19057346157524893,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2145433078","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.104237124,0.004759753,0.88211876,0.0028647173,0.000067638895,0.0000407117,0.00019379506,0.00071486813,0.005002616],"genre_scores_gemma":[0.54151964,0.0019839543,0.4533552,0.0003705463,0.00009942435,0.000060297596,0.0004322489,0.00020684133,0.001971806],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99956256,0.00014298558,0.000024303547,0.00015385868,0.00008359489,0.000032618056],"domain_scores_gemma":[0.9973361,0.0017242176,0.000387389,0.00019755501,0.00018274425,0.00017200304],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014307064,0.00039321106,0.00070560427,0.00067266467,0.00079154543,0.0013777438,0.0011745123,0.0008635322,0.0027784782],"category_scores_gemma":[0.005848103,0.00030718086,0.00037379275,0.0012550832,0.0017852469,0.0025597871,0.0011833812,0.0014777981,0.00030068686],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00022132916,0.00008991609,0.013094979,0.00057447626,0.00009203996,0.00019351453,0.0003586827,0.33057353,0.008533091,0.34781486,0.004728062,0.29372555],"study_design_scores_gemma":[0.000036378093,0.000100673584,0.0043747863,0.00006339583,0.000035488592,0.00042688477,0.00018814491,0.37816617,0.0058427434,0.6001042,0.010621291,0.00003983208],"about_ca_topic_score_codex":0.0024482114,"about_ca_topic_score_gemma":0.0021816355,"teacher_disagreement_score":0.0027784782,"about_ca_system_score_codex":0.0011445071,"about_ca_system_score_gemma":0.00079425034,"threshold_uncertainty_score":0.009294927},"labels":[],"label_agreement":null},{"id":"W2145481313","doi":"10.1093/bioinformatics/btp044","title":"Radié: visualizing taxon properties and parsimonious mappings using a radial phylogenetic tree","year":2009,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Dalhousie University","funders":"Canada Research Chairs; Genome Atlantic; Auckland University of Technology, New Zealand","keywords":"Phylogenetic tree; Character (mathematics); Tree (set theory); Character evolution; Phylogenetics; Computer science; Java; Biology; Evolutionary biology; Combinatorics; Mathematics; Genetics; Clade","score_opus":0.03208165806107003,"score_gpt":0.23892651207457635,"score_spread":0.20684485401350633,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2145481313","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.03841253,0.00090142863,0.81778693,0.00080860197,0.00011620534,0.00025862124,0.02144543,0.11231438,0.007955888],"genre_scores_gemma":[0.12594625,0.0006817629,0.84639037,0.0001412925,0.0000290287,0.00048685144,0.0126082245,0.011105873,0.0026103095],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9995134,0.00014925745,0.00004751118,0.00012810745,0.000116141324,0.000045477736],"domain_scores_gemma":[0.99820745,0.0011561238,0.00014644387,0.00017501315,0.000218094,0.0000967802],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015693934,0.0015603356,0.000981637,0.0052977824,0.0007616433,0.0027381137,0.0019149102,0.0012388062,0.028978221],"category_scores_gemma":[0.0061542722,0.00087982713,0.0012777676,0.0034378394,0.0004859554,0.002912537,0.002440419,0.0015497361,0.004942509],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0017040243,0.00037691812,0.0112013,0.004419045,0.0005423777,0.0014945595,0.005624638,0.086790584,0.054702993,0.062308427,0.1790219,0.5918132],"study_design_scores_gemma":[0.00040790657,0.00017708896,0.010326047,0.00060039863,0.00015319651,0.0013420214,0.0011790781,0.71877754,0.021947093,0.0703124,0.17446898,0.00030831306],"about_ca_topic_score_codex":0.0038683966,"about_ca_topic_score_gemma":0.005318788,"teacher_disagreement_score":0.028978221,"about_ca_system_score_codex":0.00061795645,"about_ca_system_score_gemma":0.0008123275,"threshold_uncertainty_score":0.09694183},"labels":[],"label_agreement":null},{"id":"W2145494207","doi":"10.1093/bioinformatics/btn260","title":"Optimal vaccination schedules using simulated annealing","year":2008,"lang":"en","type":"article","venue":"Bioinformatics","topic":"vaccines and immunoinformatics approaches","field":"Biochemistry, Genetics and Molecular Biology","cited_by":36,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Ministero dell'Università e della Ricerca; Alberta Agricultural Research Institute","keywords":"Simulated annealing; Computer science; Software; Genetic algorithm; Mathematical optimization; Distributed computing; Algorithm; Machine learning; Mathematics","score_opus":0.030868948883166217,"score_gpt":0.25857468700424174,"score_spread":0.22770573812107553,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2145494207","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.07524203,0.0012823364,0.8933871,0.0009920237,0.00025875744,0.00041528465,0.00033639948,0.00094274915,0.027143301],"genre_scores_gemma":[0.6675961,0.0007847894,0.32015774,0.00037419537,0.000101825935,0.0008019136,0.000686946,0.00034831645,0.009148233],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9995524,0.00018146841,0.000021515354,0.000079468125,0.00009788412,0.000067246554],"domain_scores_gemma":[0.9988815,0.00073533016,0.00011781265,0.000071576884,0.0001316665,0.00006211631],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007761446,0.00066692696,0.0009427525,0.00084870006,0.00045370718,0.0006913827,0.00072922255,0.00095017673,0.008463776],"category_scores_gemma":[0.004071595,0.0005593989,0.0008513672,0.00046951434,0.00061000016,0.0007578275,0.00069954613,0.000857763,0.0012306718],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00011812761,0.00005959097,0.0004145157,0.00011909463,0.000038727547,0.00003559426,0.000049268234,0.9568156,0.0017879076,0.011218827,0.0024919838,0.02685079],"study_design_scores_gemma":[0.000052322874,0.00008663544,0.00023431511,0.000028095741,0.000025645852,0.000019234472,0.000016320972,0.98049015,0.0010631512,0.014924447,0.003049127,0.000010603297],"about_ca_topic_score_codex":0.0022976843,"about_ca_topic_score_gemma":0.0023843322,"teacher_disagreement_score":0.008463776,"about_ca_system_score_codex":0.0012361535,"about_ca_system_score_gemma":0.0012501362,"threshold_uncertainty_score":0.028314114},"labels":[],"label_agreement":null},{"id":"W2145562754","doi":"10.1093/bioinformatics/btu462","title":"ASTRAL: genome-scale coalescent-based species tree estimation","year":2014,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1422,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Howard Hughes Medical Institute; University of Alberta; National Science Foundation","keywords":"Coalescent theory; Tree (set theory); Genome; Population; Computational biology; Concatenation (mathematics); Scale (ratio); Computer science; Biology; Data mining; Theoretical computer science; Gene; Geography; Mathematics; Genetics; Phylogenetic tree; Cartography; Combinatorics","score_opus":0.010755306190773325,"score_gpt":0.21322452821859994,"score_spread":0.2024692220278266,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2145562754","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.03422132,0.0024123518,0.759843,0.001326481,0.0005967966,0.00042442352,0.06839023,0.12763071,0.005154725],"genre_scores_gemma":[0.11269709,0.0010312566,0.7370113,0.0009168837,0.00030960928,0.0012211873,0.121384926,0.022793042,0.0026346848],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9972934,0.0009869606,0.0001616391,0.00083548814,0.00060462387,0.00011776643],"domain_scores_gemma":[0.99052215,0.005895558,0.0007694204,0.0014706075,0.00090942916,0.00043287515],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0055013504,0.0023386355,0.0024691303,0.00399511,0.0017145233,0.0032169397,0.0055641886,0.0021999874,0.013577123],"category_scores_gemma":[0.0315985,0.0017293338,0.0028991108,0.0044294354,0.0013085079,0.0033728757,0.0033647474,0.0043351394,0.012027522],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0020167744,0.00037743725,0.032523602,0.0055145593,0.0036224923,0.0012662058,0.0018098606,0.23938799,0.02787967,0.041912667,0.41105044,0.2326383],"study_design_scores_gemma":[0.0009903616,0.00019718084,0.007878468,0.00035420462,0.0006020186,0.0010492804,0.0002501792,0.8245521,0.011175505,0.06756886,0.08505779,0.00032405742],"about_ca_topic_score_codex":0.0053829183,"about_ca_topic_score_gemma":0.009631455,"teacher_disagreement_score":0.013577123,"about_ca_system_score_codex":0.00095631846,"about_ca_system_score_gemma":0.002306886,"threshold_uncertainty_score":0.04542005},"labels":[],"label_agreement":null},{"id":"W2145636602","doi":"10.1093/bioinformatics/btq092","title":"High quality SNP calling using Illumina data at shallow coverage","year":2010,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":55,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"BC Cancer Agency","funders":"","keywords":"Single-nucleotide polymorphism; SNP; Genome; Reference genome; Software; Computer science; Computational biology; Data mining; Genetics; Biology; Gene; Genotype","score_opus":0.059569126268171346,"score_gpt":0.30447150022535385,"score_spread":0.2449023739571825,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2145636602","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.07045021,0.00032187602,0.9125956,0.00016631576,0.000066348235,0.00016055546,0.0043986957,0.008641871,0.0031985035],"genre_scores_gemma":[0.07089928,0.00014435533,0.9173196,0.00017552327,0.000034301003,0.00023615948,0.007922213,0.0006494891,0.002619153],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9980108,0.00043999535,0.00011441397,0.0005996287,0.00073559117,0.00009954204],"domain_scores_gemma":[0.99808526,0.0007350492,0.00016623562,0.00049676944,0.00039228259,0.00012440883],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001538961,0.0007609062,0.0009349679,0.00089887413,0.0005909666,0.0011032762,0.0010084785,0.0007793215,0.0039666933],"category_scores_gemma":[0.004147177,0.0005169875,0.0006187429,0.0014104424,0.00034404357,0.00065431674,0.0010704936,0.0012377459,0.003951241],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00070741365,0.00011394018,0.019057108,0.00066417287,0.00030140454,0.0003821211,0.00051129865,0.014873867,0.689311,0.0051125167,0.009657196,0.2593079],"study_design_scores_gemma":[0.00015255698,0.00039629353,0.034029424,0.00009697453,0.00020699049,0.0016497481,0.00015232465,0.22663017,0.6880868,0.010414734,0.037911627,0.00027237515],"about_ca_topic_score_codex":0.0014681462,"about_ca_topic_score_gemma":0.0037246128,"teacher_disagreement_score":0.0039666933,"about_ca_system_score_codex":0.00032450608,"about_ca_system_score_gemma":0.00074612605,"threshold_uncertainty_score":0.013269901},"labels":[],"label_agreement":null},{"id":"W2145991146","doi":"10.1093/bioinformatics/btt352","title":"Harnessing virtual machines to simplify next-generation DNA sequencing analysis","year":2013,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Scientific Computing and Data Management","field":"Decision Sciences","cited_by":32,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal; Institute for Research in Immunology and Cancer","funders":"Institut National de la Recherche Agronomique","keywords":"Computer science; Software; Benchmarking; Pace; Data science; Field (mathematics); Software engineering; Operating system","score_opus":0.18150046374526624,"score_gpt":0.35659954027095925,"score_spread":0.175099076525693,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2145991146","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.06304096,0.0034833818,0.90203035,0.0028217116,0.0010265398,0.00038246723,0.0003121431,0.013558121,0.013344307],"genre_scores_gemma":[0.47660172,0.0019024465,0.5127926,0.0014063022,0.0003585662,0.0004821291,0.0008248174,0.0019122845,0.0037191426],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99603015,0.0019969349,0.00019402964,0.00038968064,0.0009904532,0.00039864762],"domain_scores_gemma":[0.9888182,0.005678134,0.0006971879,0.0028348707,0.0012747049,0.0006967991],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004876357,0.0007635429,0.0005369761,0.000926105,0.00096064433,0.0039914157,0.0025056968,0.0012048085,0.0033967318],"category_scores_gemma":[0.016129224,0.0006817468,0.0006480434,0.0008540713,0.0018023701,0.0043524047,0.004344525,0.0017611622,0.0014961795],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014676198,0.0002879922,0.019699773,0.0015112764,0.00035532427,0.001811384,0.0024506645,0.19014482,0.093018785,0.17060754,0.05224594,0.46639892],"study_design_scores_gemma":[0.00018580012,0.00048601403,0.003841999,0.00053888763,0.00015342061,0.0014922009,0.00070912496,0.44942978,0.10193298,0.16368982,0.27733728,0.0002026694],"about_ca_topic_score_codex":0.0011576451,"about_ca_topic_score_gemma":0.00093798514,"teacher_disagreement_score":0.004876357,"about_ca_system_score_codex":0.0010157586,"about_ca_system_score_gemma":0.0014891955,"threshold_uncertainty_score":0.025788963},"labels":[],"label_agreement":null},{"id":"W2146076890","doi":"10.1093/bioinformatics/btp204","title":"Prediction of sub-cavity binding preferences using an adaptive physicochemical structure representation","year":2009,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Computational Drug Discovery Methods","field":"Computer Science","cited_by":23,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"Bill and Melinda Gates Foundation","keywords":"Structural similarity; Computer science; Structural bioinformatics; Leverage (statistics); Similarity (geometry); Representation (politics); Template; Set (abstract data type); Drug discovery; Binding site; Protein structure; Computational biology; Data mining; Biological system; Artificial intelligence; Bioinformatics; Chemistry; Biology; Genetics","score_opus":0.08881122272780319,"score_gpt":0.32411579657060197,"score_spread":0.2353045738427988,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2146076890","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.15129133,0.00025229712,0.8432694,0.00013776097,0.000013448128,0.0001544045,0.00027123292,0.0032524914,0.0013576853],"genre_scores_gemma":[0.60470146,0.00021029239,0.39230558,0.00014828959,0.000020158777,0.00021989853,0.00091376697,0.0003145068,0.0011660082],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99960977,0.000087967484,0.000025116178,0.00012727683,0.000108578846,0.000041275875],"domain_scores_gemma":[0.9992817,0.0003133164,0.00011556701,0.00010816397,0.00013003925,0.00005115223],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006452239,0.00075160316,0.0009879364,0.0010966281,0.00037968613,0.0008232363,0.0013600962,0.0010922743,0.001619513],"category_scores_gemma":[0.00207178,0.00034277036,0.00080079754,0.00069238036,0.0004953043,0.000930036,0.0007599403,0.0010008495,0.00064573943],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002964207,0.0002890975,0.0064895907,0.00013203164,0.000079164565,0.00017620603,0.000068674955,0.77577895,0.065013066,0.00335159,0.0016893549,0.14663574],"study_design_scores_gemma":[0.000010674465,0.000025791433,0.0003088728,0.0000025746642,0.000006417087,0.000043912052,0.000004300675,0.99445045,0.004009286,0.0009129828,0.00021870573,0.0000061213477],"about_ca_topic_score_codex":0.0023286375,"about_ca_topic_score_gemma":0.0024972253,"teacher_disagreement_score":0.0023286375,"about_ca_system_score_codex":0.0006099458,"about_ca_system_score_gemma":0.0010061892,"threshold_uncertainty_score":0.005417824},"labels":[],"label_agreement":null},{"id":"W2146158248","doi":"10.1093/bioinformatics/bth934","title":"Genomic features in the breakpoint regions between syntenic blocks","year":2004,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":32,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Ottawa","funders":"Natural Sciences and Engineering Research Council of Canada; Canada Research Chairs","keywords":"Breakpoint; Synteny; Genome; Biology; Genetics; Computational biology; Event (particle physics); Chromosome; Chromosomal inversion; Gene; Karyotype; Physics","score_opus":0.012190793580367123,"score_gpt":0.22931640811605053,"score_spread":0.2171256145356834,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2146158248","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99369794,0.0003689061,0.004740158,0.000020724614,0.0000016807162,0.000015180263,0.0005211497,0.00006669615,0.00056761864],"genre_scores_gemma":[0.9950818,0.000070774135,0.0030772542,0.000016909828,0.000007170021,0.000017983073,0.0015103804,0.000022312333,0.0001954692],"study_design_codex":"bench_or_experimental","study_design_gemma":"observational","domain_scores_codex":[0.99976283,0.000028472641,0.000023260909,0.00010832385,0.000047717047,0.000029426034],"domain_scores_gemma":[0.99791294,0.00094592274,0.00074044906,0.00017324221,0.000101274825,0.00012621275],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00027214474,0.00015779726,0.00032525283,0.0019032692,0.00036891183,0.00047269286,0.0003449515,0.00041519108,0.0026435857],"category_scores_gemma":[0.002050806,0.0002230626,0.00021340862,0.0018218047,0.00041850712,0.00046689992,0.0005993559,0.00033552793,0.00056522636],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0019418397,0.00011612522,0.25052476,0.0007832345,0.0002975225,0.002980847,0.0019807548,0.0050585796,0.6939289,0.0036918267,0.0003948798,0.03830075],"study_design_scores_gemma":[0.0000651973,0.0002848747,0.944011,0.00005358393,0.0002432864,0.004348979,0.00033357978,0.004928939,0.036822308,0.005570769,0.0033046484,0.0000328126],"about_ca_topic_score_codex":0.0006265031,"about_ca_topic_score_gemma":0.0009664301,"teacher_disagreement_score":0.0026435857,"about_ca_system_score_codex":0.00015946715,"about_ca_system_score_gemma":0.00012444908,"threshold_uncertainty_score":0.00884366},"labels":[],"label_agreement":null},{"id":"W2146223044","doi":"10.1093/bioinformatics/btt050","title":"Oxfold: kinetic folding of RNA using stochastic context-free grammars and evolutionary information","year":2013,"lang":"en","type":"article","venue":"Bioinformatics","topic":"RNA and protein synthesis mechanisms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":15,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"Biotechnology and Biological Sciences Research Council; Engineering and Physical Sciences Research Council; Directorate for Biological Sciences","keywords":"Folding (DSP implementation); Context (archaeology); Rule-based machine translation; Computer science; RNA; Nucleic acid secondary structure; Protein folding; Kinetics; Computational biology; Theoretical computer science; Artificial intelligence; Biology; Physics; Genetics; Gene; Engineering; Biochemistry","score_opus":0.011620894942128486,"score_gpt":0.2146457541753658,"score_spread":0.20302485923323732,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2146223044","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.06022894,0.00018687695,0.9289823,0.00022600275,0.000069803646,0.00008526665,0.000738667,0.0066773877,0.0028047562],"genre_scores_gemma":[0.44498873,0.00019440455,0.54741067,0.00015069525,0.0000456744,0.00028819067,0.0017353862,0.002003742,0.003182511],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99956244,0.0001797221,0.000026441117,0.00011174524,0.00008969987,0.000029908097],"domain_scores_gemma":[0.99829525,0.0012384402,0.000091180395,0.00018057627,0.00013313629,0.00006141871],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0017055284,0.00072772545,0.00084029103,0.0006459564,0.00038154624,0.0009252544,0.0014648199,0.00094310415,0.0044133784],"category_scores_gemma":[0.004217398,0.0005476609,0.00086888834,0.0005483943,0.0007138726,0.0011273565,0.00092704396,0.0009237278,0.00069208874],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001521922,0.000052114592,0.0016863347,0.0001354116,0.00004606199,0.00013725548,0.000088402965,0.9243469,0.0027525823,0.026028985,0.0028291568,0.04174465],"study_design_scores_gemma":[0.000030991025,0.000014552766,0.000102564234,0.000009473179,0.000005715327,0.000041671494,0.000006932047,0.9775463,0.0020701124,0.019249938,0.00091505307,0.000006707247],"about_ca_topic_score_codex":0.0026175492,"about_ca_topic_score_gemma":0.0025433518,"teacher_disagreement_score":0.0044133784,"about_ca_system_score_codex":0.0009863526,"about_ca_system_score_gemma":0.0014055389,"threshold_uncertainty_score":0.01476419},"labels":[],"label_agreement":null},{"id":"W2146279700","doi":"10.1093/bioinformatics/bti675","title":"METIS: multiple extraction techniques for informative sentences","year":2005,"lang":"en","type":"article","venue":"Bioinformatics","topic":"RNA and protein synthesis mechanisms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":25,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Engineering and Physical Sciences Research Council; Ministry of Economy, Trade and Industry","keywords":"Metis; Annotation; Computer science; Sentence; Information retrieval; Support vector machine; Component (thermodynamics); Natural language processing; Artificial intelligence; World Wide Web","score_opus":0.016057162470740523,"score_gpt":0.27316759891147785,"score_spread":0.25711043644073733,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2146279700","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0043150424,0.0006010558,0.899995,0.0007122611,0.0003482953,0.00092916394,0.024402667,0.064967126,0.0037293027],"genre_scores_gemma":[0.024269555,0.00032550032,0.925922,0.00017524185,0.00027304824,0.0009496954,0.039881725,0.0047239754,0.0034792256],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.995993,0.0012190818,0.00071364903,0.0008651425,0.001036425,0.00017266486],"domain_scores_gemma":[0.98675156,0.0073261615,0.0011078427,0.0016677399,0.0028605703,0.00028619933],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005039754,0.0031561852,0.0011713849,0.010029011,0.0017523041,0.002777184,0.002631878,0.0014330202,0.040246624],"category_scores_gemma":[0.020003216,0.0015829892,0.0020302837,0.005359787,0.00078659185,0.0039086347,0.0040724096,0.0025535515,0.022228755],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00068819063,0.00015989188,0.0014046541,0.0030703016,0.0003301732,0.0011971586,0.0013537295,0.0023113356,0.03502827,0.015334971,0.21264444,0.72647685],"study_design_scores_gemma":[0.00057750964,0.00057986786,0.006025814,0.0010035035,0.0008711059,0.0051240204,0.0021466215,0.23103277,0.13949889,0.09219618,0.52051425,0.00042949547],"about_ca_topic_score_codex":0.0010299394,"about_ca_topic_score_gemma":0.0021591699,"teacher_disagreement_score":0.040246624,"about_ca_system_score_codex":0.00074461516,"about_ca_system_score_gemma":0.0020618604,"threshold_uncertainty_score":0.13463843},"labels":[],"label_agreement":null},{"id":"W2146384526","doi":"10.1093/bioinformatics/btl611","title":"Dynamic exploration and editing of KEGG pathway diagrams","year":2006,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":88,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Leibniz-Institut für Pflanzengenetik und Kulturpflanzenforschung; Institute of Genetics","keywords":"KEGG; Visualization; Computer science; Encyclopedia; Markup language; Resource (disambiguation); Data visualization; Graph drawing; World Wide Web; Information retrieval; Presentation (obstetrics); Human–computer interaction; Information visualization; Data mining; XML; Biology","score_opus":0.006553937449244133,"score_gpt":0.2026285416479405,"score_spread":0.19607460419869638,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2146384526","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.011967904,0.00041185325,0.9145643,0.0007175974,0.00017153443,0.00022973114,0.00791006,0.05845816,0.005568868],"genre_scores_gemma":[0.07771508,0.00080644543,0.8965566,0.00015756769,0.00009464702,0.0005483588,0.009911826,0.010397674,0.0038117585],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9985115,0.00046803924,0.0001905744,0.0003475608,0.00040822173,0.0000740601],"domain_scores_gemma":[0.98947746,0.007345167,0.0004211499,0.0013373034,0.0011433961,0.00027550326],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0030588896,0.0015661765,0.000804979,0.004105462,0.00076860015,0.0031811919,0.0018616439,0.0006298021,0.012473636],"category_scores_gemma":[0.015439275,0.00066322787,0.0011523976,0.0029257066,0.0005326018,0.0024906015,0.002673564,0.0014628565,0.0034147096],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014989072,0.0002007435,0.004634221,0.0037020121,0.00031461503,0.0018504888,0.006362642,0.023658343,0.07952071,0.04562406,0.10737084,0.7252624],"study_design_scores_gemma":[0.00036492784,0.0001638843,0.0051729623,0.00081761123,0.0003358537,0.002438169,0.0011806141,0.15318203,0.1470719,0.11055296,0.5783251,0.00039400876],"about_ca_topic_score_codex":0.0011450432,"about_ca_topic_score_gemma":0.0013045401,"teacher_disagreement_score":0.012473636,"about_ca_system_score_codex":0.00050122896,"about_ca_system_score_gemma":0.0012418336,"threshold_uncertainty_score":0.041728497},"labels":[],"label_agreement":null},{"id":"W2146705459","doi":"10.1093/bioinformatics/btr635","title":"Inferring sequence regions under functional divergence in duplicate genes","year":2011,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University","funders":"","keywords":"Functional divergence; Divergence (linguistics); Biology; Gene duplication; Phylogenetic tree; Computational biology; Gene; Genetics; Evolutionary biology; Gene family; Genome","score_opus":0.05988863102699207,"score_gpt":0.24481273053407207,"score_spread":0.18492409950708,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2146705459","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9245264,0.00028166934,0.07408319,0.00004440488,0.0000052493738,0.00001698238,0.00035706328,0.0002417024,0.0004432923],"genre_scores_gemma":[0.96711135,0.000075291406,0.0318731,0.00003277336,0.000007540253,0.000019130484,0.0007209301,0.000039607203,0.00012020727],"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99951994,0.00008588873,0.00003559047,0.00019622479,0.000107669766,0.000054637174],"domain_scores_gemma":[0.99737144,0.0013865543,0.0006121637,0.0002023744,0.00022634381,0.00020105897],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011183241,0.00036910054,0.00073466636,0.0016422481,0.00061046635,0.00054597954,0.000897197,0.00083857286,0.00060699694],"category_scores_gemma":[0.004216589,0.0003857,0.00080882816,0.0012868636,0.000577598,0.00065723725,0.0007778454,0.00048689218,0.00020262787],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014589529,0.0002216977,0.4951274,0.00074908946,0.0003759859,0.003525174,0.0010998716,0.20547463,0.19857675,0.007045502,0.0008651746,0.08547981],"study_design_scores_gemma":[0.000058652844,0.00029447972,0.11461346,0.00006421937,0.00023786574,0.003945911,0.000520556,0.7902494,0.07271361,0.0147995,0.0024214638,0.00008084079],"about_ca_topic_score_codex":0.001979226,"about_ca_topic_score_gemma":0.0014575169,"teacher_disagreement_score":0.001979226,"about_ca_system_score_codex":0.000722532,"about_ca_system_score_gemma":0.00067266915,"threshold_uncertainty_score":0.005914271},"labels":[],"label_agreement":null},{"id":"W2146891463","doi":"10.1093/bioinformatics/btm604","title":"Mutual information without the influence of phylogeny or entropy dramatically improves residue contact prediction","year":2007,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Protein Structure and Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":537,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"Canada Research Chairs","keywords":"Mutual information; Entropy (arrow of time); Computer science; Phylogenetic tree; Metric (unit); Information theory; Multiple sequence alignment; Computational biology; Protein family; Identification (biology); Artificial intelligence; Statistical physics; Theoretical computer science; Sequence alignment; Biology; Mathematics; Genetics; Physics; Statistics; Gene; Peptide sequence","score_opus":0.0039831791568075295,"score_gpt":0.2248982414525908,"score_spread":0.22091506229578325,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2146891463","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.4630944,0.00089530647,0.5219526,0.0004615702,0.000049411487,0.00006728773,0.0011270021,0.007648244,0.004704049],"genre_scores_gemma":[0.84708774,0.00021291108,0.14967832,0.000077891484,0.000044040302,0.000042774893,0.0015143586,0.0004007203,0.0009412807],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99887544,0.00033223745,0.0000634911,0.00025593775,0.00036948218,0.00010346363],"domain_scores_gemma":[0.9971137,0.0017081839,0.00032426429,0.00037802424,0.0002762147,0.00019953409],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0018328488,0.0010262956,0.0011121072,0.0018598073,0.000592741,0.0009052335,0.00089250377,0.00076147425,0.0019523769],"category_scores_gemma":[0.007454407,0.00043056716,0.0006542653,0.0010977848,0.00056525547,0.002206326,0.001593683,0.001015703,0.00091408833],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0015667052,0.00053518236,0.05036226,0.0006590985,0.0005537424,0.00047090033,0.0003204693,0.28720075,0.13648103,0.009148666,0.009087847,0.5036133],"study_design_scores_gemma":[0.000031515956,0.00015160802,0.008667748,0.000022655278,0.00006652721,0.00018709956,0.000023746175,0.9495293,0.034836087,0.005053828,0.001390835,0.00003916083],"about_ca_topic_score_codex":0.0016416416,"about_ca_topic_score_gemma":0.0034017814,"teacher_disagreement_score":0.0019523769,"about_ca_system_score_codex":0.00055951544,"about_ca_system_score_gemma":0.00087139435,"threshold_uncertainty_score":0.009693086},"labels":[],"label_agreement":null},{"id":"W2147492358","doi":"10.1093/bioinformatics/18.12.1696","title":"DNACompress: fast and effective DNA sequence compression","year":2002,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Algorithms and Data Compression","field":"Computer Science","cited_by":210,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Bioinformatics Solutions (Canada); Western University","funders":"National Science Foundation","keywords":"Compression (physics); Sequence (biology); DNA; Computer science; DNA sequencing; Data compression; Computational biology; Algorithm; Genetics; Biology; Materials science; Composite material","score_opus":0.021486470274000594,"score_gpt":0.23858404324737703,"score_spread":0.21709757297337645,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2147492358","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.03246675,0.0020878066,0.8701677,0.00077987515,0.00047597173,0.0003362939,0.005522962,0.077695474,0.01046714],"genre_scores_gemma":[0.118822075,0.0009165324,0.84883064,0.0003600698,0.00024098397,0.00072024367,0.012414291,0.0037377423,0.01395739],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9994324,0.00007205144,0.000045570516,0.000103910235,0.00030528073,0.00004085179],"domain_scores_gemma":[0.99912006,0.00035457566,0.000087373126,0.00016591104,0.00021379702,0.000058201775],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007324831,0.0011686556,0.00055818196,0.0020201239,0.00055097125,0.0008478376,0.0012513102,0.00078450405,0.014409398],"category_scores_gemma":[0.0027052835,0.00054530863,0.00040753194,0.0018175566,0.000492989,0.0010320103,0.0012072697,0.0011707062,0.0073053464],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007889232,0.0001256345,0.0008934613,0.00056272355,0.00007034222,0.0004213182,0.00016181634,0.012756411,0.08729523,0.014869772,0.10966554,0.7723889],"study_design_scores_gemma":[0.00050999183,0.00031925965,0.001916713,0.00015087707,0.00007012073,0.0013295701,0.00009356661,0.3799454,0.47183505,0.023258809,0.12046427,0.000106424195],"about_ca_topic_score_codex":0.0009566056,"about_ca_topic_score_gemma":0.001350841,"teacher_disagreement_score":0.014409398,"about_ca_system_score_codex":0.00045018297,"about_ca_system_score_gemma":0.00070343126,"threshold_uncertainty_score":0.048204243},"labels":[],"label_agreement":null},{"id":"W2147626660","doi":"10.1093/bioinformatics/btr591","title":"Optimized application of penalized regression methods to diverse genomic data","year":2011,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":99,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University Health Network; Ontario Institute for Cancer Research","funders":"Division of Biological Infrastructure; National Science Foundation","keywords":"Elastic net regularization; Lasso (programming language); Feature selection; Computer science; Univariate; Regression; Data mining; R package; Predictive modelling; Regression analysis; Machine learning; Artificial intelligence; Statistics; Mathematics; Multivariate statistics","score_opus":0.08353315231449081,"score_gpt":0.3586795711320192,"score_spread":0.2751464188175284,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2147626660","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.029733839,0.00015233301,0.9678335,0.00027810154,0.000031199357,0.00006817115,0.00031485513,0.0008971882,0.0006908325],"genre_scores_gemma":[0.29421186,0.00017910829,0.7017221,0.00027279306,0.00004096669,0.00049464655,0.0013927654,0.00052162696,0.0011641474],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99725324,0.001763252,0.00011380931,0.00040647236,0.00035836114,0.00010501266],"domain_scores_gemma":[0.98871464,0.008532397,0.00057964685,0.0007568452,0.0012523148,0.00016419067],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0059875883,0.00078970264,0.0006406039,0.0006443608,0.0003858375,0.0008337926,0.0016635845,0.00090346957,0.0020169502],"category_scores_gemma":[0.019623147,0.0003686898,0.00067942607,0.00079098676,0.00077306293,0.0006151317,0.0009985808,0.0016100109,0.00066974165],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00013611624,0.000067881636,0.0025223868,0.00014134345,0.000070660906,0.00012009432,0.00006790586,0.9473239,0.00475012,0.0056300825,0.0020303922,0.037139177],"study_design_scores_gemma":[0.000014673312,0.000021419311,0.0003380259,0.000009343359,0.000004679217,0.000021583703,0.0000067076317,0.99487597,0.0016248818,0.002559188,0.0005160198,0.0000075401686],"about_ca_topic_score_codex":0.0031273183,"about_ca_topic_score_gemma":0.003158405,"teacher_disagreement_score":0.0059875883,"about_ca_system_score_codex":0.0006573933,"about_ca_system_score_gemma":0.00143908,"threshold_uncertainty_score":0.031665742},"labels":[],"label_agreement":null},{"id":"W2147702777","doi":"10.1093/bioinformatics/btr229","title":"Sequence-based prediction of protein crystallization, purification and production propensity","year":2011,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Enzyme Structure and Function","field":"Materials Science","cited_by":77,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Natural Sciences and Engineering Research Council of Canada; Killam Trusts","keywords":"Crystallization; Protein crystallization; Computer science; Process (computing); Production (economics); Set (abstract data type); In silico; Selection (genetic algorithm); Sequence (biology); Quality (philosophy); Diffraction; Algorithm; Data mining; Biological system; Artificial intelligence; Chemistry; Biology; Physics; Genetics","score_opus":0.0539457064322674,"score_gpt":0.21324081686334262,"score_spread":0.15929511043107522,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2147702777","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.5867427,0.0020625556,0.40348226,0.00046695315,0.000034506174,0.00015382122,0.002437853,0.0029579776,0.0016614229],"genre_scores_gemma":[0.9224925,0.00060251774,0.07263508,0.0000926382,0.000037493523,0.00010389135,0.003211637,0.00009280754,0.0007314226],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9991571,0.00018572877,0.00007886107,0.0002775367,0.00023173075,0.000069052],"domain_scores_gemma":[0.99491227,0.0029666298,0.0008355272,0.00030908495,0.0007856495,0.00019081564],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002578736,0.00089476653,0.000775877,0.0011337104,0.00027248514,0.0008824441,0.0006442162,0.00075705047,0.0009047618],"category_scores_gemma":[0.0070607597,0.00030710062,0.0005149004,0.0007918737,0.00050989934,0.0010116787,0.00056923146,0.00087311224,0.00080417906],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00065694796,0.00033658498,0.24953447,0.0007263597,0.00022023538,0.00037053743,0.0001235431,0.5815008,0.032252956,0.0015689285,0.0035385734,0.12916996],"study_design_scores_gemma":[0.000017553863,0.00016425826,0.024187738,0.000031389834,0.000040069375,0.00023286437,0.00001923441,0.9504293,0.0224768,0.0015159373,0.00086130376,0.00002360257],"about_ca_topic_score_codex":0.0027247637,"about_ca_topic_score_gemma":0.0023699123,"teacher_disagreement_score":0.0027247637,"about_ca_system_score_codex":0.0006408002,"about_ca_system_score_gemma":0.001107865,"threshold_uncertainty_score":0.013637841},"labels":[],"label_agreement":null},{"id":"W2147803109","doi":"10.1093/bioinformatics/btv274","title":"SpeeDB: fast structural protein searches","year":2015,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Protein Structure and Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Carleton University","funders":"Natural Sciences and Engineering Research Council of Canada; National Research Foundation","keywords":"Protein Data Bank (RCSB PDB); Computer science; Protein Data Bank; Protein superfamily; Interface (matter); Computational biology; Function (biology); Protein structure; Identification (biology); Stability (learning theory); Data mining; Information retrieval; Chemistry; Biology; Machine learning; Stereochemistry; Biochemistry; Genetics","score_opus":0.019252973532231186,"score_gpt":0.2571173499731011,"score_spread":0.2378643764408699,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2147803109","genre_codex":"software","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.018806735,0.008657285,0.25821048,0.002188583,0.0012366356,0.0014892253,0.09853584,0.582022,0.028853206],"genre_scores_gemma":[0.08144568,0.0045737987,0.57561594,0.0011985332,0.00035783005,0.0035412004,0.2583914,0.063021004,0.011854616],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99698204,0.000933293,0.00020273149,0.0005357156,0.0010829614,0.0002632455],"domain_scores_gemma":[0.99600023,0.0022921604,0.00026191902,0.00050760625,0.00068838825,0.0002495919],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004484468,0.00542726,0.003273695,0.00470646,0.0016970419,0.0031884916,0.0065457863,0.004034518,0.067827925],"category_scores_gemma":[0.014566805,0.0028175868,0.0024136316,0.0053423615,0.00088075147,0.0048320643,0.0036916516,0.0032478985,0.056068372],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0018957425,0.00029046266,0.0017286333,0.0032259752,0.00062449346,0.000583367,0.00022599864,0.0122064585,0.009297173,0.012780426,0.84307516,0.11406612],"study_design_scores_gemma":[0.0048635434,0.00061494904,0.0037824018,0.0011098188,0.00039787998,0.0021447544,0.00032207477,0.4155662,0.020157026,0.066264085,0.48437032,0.00040692682],"about_ca_topic_score_codex":0.0030168032,"about_ca_topic_score_gemma":0.0044703106,"teacher_disagreement_score":0.067827925,"about_ca_system_score_codex":0.0013200138,"about_ca_system_score_gemma":0.0025797568,"threshold_uncertainty_score":0.22690701},"labels":[],"label_agreement":null},{"id":"W2147849606","doi":"10.1093/bioinformatics/btq171","title":"PepC: proteomics software for identifying differentially expressed proteins based on spectral counting","year":2010,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Advanced Proteomics Techniques and Applications","field":"Chemistry","cited_by":41,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Human Genome Research Institute; Canadian Institutes of Health Research; National Institutes of Health","keywords":"Pipeline (software); Computer science; Software; Proteomics; Java; MIT License; False discovery rate; Computational biology; Biology; Programming language; Biochemistry; Gene","score_opus":0.01931066443580819,"score_gpt":0.2745229701509697,"score_spread":0.2552123057151615,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2147849606","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0042151767,0.00030775648,0.710965,0.00015139632,0.00012133162,0.00035457368,0.01656581,0.26517868,0.0021403767],"genre_scores_gemma":[0.022392599,0.00044746272,0.89791214,0.0004507697,0.00007530463,0.0021340747,0.034235813,0.0379392,0.0044125626],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9982734,0.00021504809,0.00025805325,0.000447947,0.0006786885,0.00012695498],"domain_scores_gemma":[0.9972717,0.0011996231,0.00039676987,0.000373793,0.0006188832,0.00013918107],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004167208,0.0031893281,0.0020082186,0.004835973,0.0011064543,0.0017458659,0.0032581121,0.0013700397,0.023032866],"category_scores_gemma":[0.0067506717,0.0017639626,0.001655152,0.0033674254,0.0007865663,0.0025762243,0.0021758077,0.0029126548,0.017285347],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0023192686,0.000413603,0.0074550845,0.0043243775,0.00070549274,0.00090999994,0.0005487282,0.0072922306,0.25757158,0.010621408,0.35345083,0.35438734],"study_design_scores_gemma":[0.0007441404,0.00037188502,0.01832979,0.00038452953,0.0003190392,0.0033624433,0.00014088856,0.26238886,0.3931995,0.026022391,0.29419225,0.0005442703],"about_ca_topic_score_codex":0.0012951412,"about_ca_topic_score_gemma":0.0011746085,"teacher_disagreement_score":0.023032866,"about_ca_system_score_codex":0.0008745935,"about_ca_system_score_gemma":0.0017017421,"threshold_uncertainty_score":0.07705265},"labels":[],"label_agreement":null},{"id":"W2148132327","doi":"10.1093/bioinformatics/18.12.1692","title":"RED: the analysis, management and dissemination ofexpressed sequence tags","year":2002,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Molecular Biology Techniques and Applications","field":"Biochemistry, Genetics and Molecular Biology","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"Medical Research Council; National Institutes of Health","keywords":"Sequence (biology); Sequence analysis; World Wide Web; Web site; Expressed sequence tag; DNA sequencing; Library science; Computer science; Database; Biology; Information retrieval; The Internet; Genetics; DNA; Complementary DNA; Gene","score_opus":0.012029396377031123,"score_gpt":0.2626744045564294,"score_spread":0.2506450081793983,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2148132327","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.010897736,0.0011480922,0.35900363,0.0013646082,0.0005573567,0.0010317543,0.21789438,0.3902965,0.017806012],"genre_scores_gemma":[0.032865085,0.0016874047,0.28074402,0.0010880344,0.00023379007,0.0014275655,0.601686,0.040881146,0.039386973],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99795157,0.00031879975,0.00024893944,0.00044655325,0.00084493385,0.00018921072],"domain_scores_gemma":[0.99666387,0.0008521173,0.00042801534,0.0009855139,0.0006709959,0.00039943957],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.006019212,0.0019379403,0.0013765588,0.003025533,0.0009306299,0.0037305115,0.0038425014,0.0011111541,0.040786427],"category_scores_gemma":[0.0060592247,0.0011292864,0.00085024245,0.002773792,0.0004710147,0.002143804,0.0026208402,0.0024835144,0.06885179],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0021054475,0.00027643837,0.0024199395,0.0015314455,0.00019606006,0.00094895065,0.00046856553,0.0017009604,0.08547497,0.0061635524,0.65188754,0.24682607],"study_design_scores_gemma":[0.00046460264,0.00023613051,0.0072732205,0.00023668849,0.00016490744,0.0010956072,0.00014942684,0.01732094,0.16301481,0.006362216,0.80340064,0.0002809269],"about_ca_topic_score_codex":0.0016582198,"about_ca_topic_score_gemma":0.0012699395,"teacher_disagreement_score":0.040786427,"about_ca_system_score_codex":0.000823589,"about_ca_system_score_gemma":0.0014888927,"threshold_uncertainty_score":0.13644415},"labels":[],"label_agreement":null},{"id":"W2148133654","doi":"10.1093/bioinformatics/btm221","title":"Modeling recurrent DNA copy number alterations in array CGH data","year":2007,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomic variations and chromosomal abnormalities","field":"Biochemistry, Genetics and Molecular Biology","cited_by":71,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"Genome British Columbia; Michael Smith Health Research BC","keywords":"Comparative genomic hybridization; Inference; Computer science; Computational biology; Copy-number variation; Biology; Data set; Set (abstract data type); Artificial intelligence; Genetics; Genome; Gene","score_opus":0.029263217990079266,"score_gpt":0.2809631552105036,"score_spread":0.25169993722042433,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2148133654","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.19743288,0.00020999063,0.79830354,0.00067103625,0.000019827688,0.0000884341,0.0015180339,0.0013615922,0.00039458423],"genre_scores_gemma":[0.81746674,0.00025331747,0.17670976,0.00019268805,0.000059193175,0.0002748719,0.0028879498,0.00013793571,0.0020175967],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99836916,0.0005022025,0.00009734326,0.00067117,0.00023264367,0.0001274855],"domain_scores_gemma":[0.9880784,0.008965379,0.0010135642,0.0013352662,0.00042912876,0.0001782215],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004008551,0.0008076171,0.0009953732,0.0012689589,0.00048346943,0.0010722689,0.0022151298,0.0016982349,0.00096188724],"category_scores_gemma":[0.01392954,0.0007638151,0.0010975777,0.0015533445,0.0015491246,0.002109247,0.0010125147,0.001967137,0.00044475563],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00022816157,0.00010097273,0.019785836,0.00009017655,0.0000699489,0.00018561137,0.00024682525,0.9343272,0.005684707,0.006664156,0.00054592313,0.032070573],"study_design_scores_gemma":[0.00000998297,0.000023632969,0.00163196,0.000004559644,0.000013344583,0.00005468086,0.000010805505,0.98561674,0.0019032194,0.010534153,0.00018446802,0.0000123062855],"about_ca_topic_score_codex":0.0071381815,"about_ca_topic_score_gemma":0.010076298,"teacher_disagreement_score":0.0071381815,"about_ca_system_score_codex":0.0013072224,"about_ca_system_score_gemma":0.00097397424,"threshold_uncertainty_score":0.021199465},"labels":[],"label_agreement":null},{"id":"W2148271643","doi":"10.1093/bioinformatics/btv060","title":"Computational identification of MoRFs in protein sequences","year":2015,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Machine Learning in Bioinformatics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":99,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"Natural Sciences and Engineering Research Council of Canada; Canadian Institutes of Health Research; Genome Canada","keywords":"Computer science; Support vector machine; Identification (biology); Sequence (biology); Artificial intelligence; Computational biology; Data mining; Set (abstract data type); Machine learning; Pattern recognition (psychology); Biology; Genetics","score_opus":0.01588503249904103,"score_gpt":0.2752276049529452,"score_spread":0.25934257245390413,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2148271643","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7194423,0.0022543017,0.25676125,0.00083334156,0.000120497454,0.00022198707,0.003997831,0.012544261,0.00382435],"genre_scores_gemma":[0.71350205,0.0005899885,0.2706887,0.00023199864,0.00010429908,0.00029127244,0.012221291,0.00039292822,0.0019774376],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9995797,0.00008469967,0.000025234673,0.00013213488,0.00013030462,0.000047816487],"domain_scores_gemma":[0.99782336,0.001386467,0.00022324071,0.00014755985,0.00027579354,0.00014357449],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010267275,0.0008362245,0.0010584746,0.0018159454,0.00048812904,0.00063945464,0.0012095459,0.00091482635,0.0020921673],"category_scores_gemma":[0.003998599,0.0003181357,0.00059755915,0.0013621775,0.0003943637,0.0010886665,0.0007414183,0.00085381296,0.00072319567],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008525687,0.0003622669,0.029543687,0.0006587738,0.00021992567,0.0004074811,0.00007990881,0.7600415,0.008196912,0.0033466069,0.012709175,0.18358131],"study_design_scores_gemma":[0.000010804519,0.000030342853,0.0007392006,0.0000050393146,0.0000064320143,0.00003127104,0.000009251903,0.9970336,0.0008836264,0.0009864133,0.0002608873,0.0000030637304],"about_ca_topic_score_codex":0.0042425157,"about_ca_topic_score_gemma":0.0059372378,"teacher_disagreement_score":0.0042425157,"about_ca_system_score_codex":0.0006756626,"about_ca_system_score_gemma":0.0013996175,"threshold_uncertainty_score":0.008435607},"labels":[],"label_agreement":null},{"id":"W2148315211","doi":"10.1093/bioinformatics/btl662","title":"Computing exact<i>P</i>-values for DNA motifs","year":2007,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":50,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"National Human Genome Research Institute","keywords":"Heuristic; Computation; Computer science; Dynamic programming; Ranking (information retrieval); Value (mathematics); Algorithm; Software; Exact solutions in general relativity; Position (finance); Theoretical computer science; Mathematics; Artificial intelligence; Machine learning","score_opus":0.008630630522480112,"score_gpt":0.244304805498116,"score_spread":0.2356741749756359,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2148315211","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.05085009,0.00094954594,0.9400318,0.0006887195,0.0000725896,0.000088412795,0.0014376848,0.0029921348,0.00288885],"genre_scores_gemma":[0.4123034,0.00064787804,0.58169514,0.00030195285,0.00013621611,0.00030792775,0.002553217,0.00056043325,0.0014939172],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99614054,0.0013230013,0.00028130368,0.0012036476,0.00082742696,0.0002240978],"domain_scores_gemma":[0.9576208,0.03520249,0.0021382172,0.002987218,0.0014770115,0.00057430175],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004775111,0.0012301174,0.0016687361,0.0022205864,0.00086015655,0.0023262603,0.0023203702,0.0019346927,0.005852446],"category_scores_gemma":[0.055154722,0.00071265385,0.0007461979,0.002932978,0.002014571,0.0034476635,0.0015171517,0.0021302954,0.0023496142],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013411369,0.00044441817,0.032598104,0.00206299,0.0003379688,0.00092105696,0.00041186006,0.4732353,0.017613815,0.09047753,0.016134024,0.36442176],"study_design_scores_gemma":[0.000117152566,0.00018042565,0.0027383184,0.000112351205,0.0000500397,0.0012008933,0.000109333654,0.615937,0.012094958,0.36351034,0.0038880731,0.0000610956],"about_ca_topic_score_codex":0.0008886301,"about_ca_topic_score_gemma":0.0010420558,"teacher_disagreement_score":0.005852446,"about_ca_system_score_codex":0.0011380005,"about_ca_system_score_gemma":0.00178729,"threshold_uncertainty_score":0.025253475},"labels":[],"label_agreement":null},{"id":"W2148425737","doi":"10.1093/bioinformatics/btu558","title":"BioBloom tools: fast, accurate and memory-efficient host species sequence screening using bloom filters","year":2014,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":134,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canada's Michael Smith Genome Sciences Centre","funders":"Genome British Columbia; Genome Canada","keywords":"Bloom filter; False positive paradox; Computer science; Set (abstract data type); Sequence (biology); Filter (signal processing); Host (biology); Software; False positives and false negatives; Data set; Data structure; Data mining; Algorithm; Artificial intelligence; Programming language; Biology","score_opus":0.04576000212109506,"score_gpt":0.2551211840184869,"score_spread":0.20936118189739183,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2148425737","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.023901084,0.0012283518,0.64667505,0.0006451826,0.0002506405,0.00059578323,0.035497185,0.2855703,0.005636415],"genre_scores_gemma":[0.06936495,0.00059463165,0.85803294,0.0005197127,0.000093453724,0.001506423,0.047338597,0.016769283,0.005780076],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99790937,0.00045685106,0.00021933796,0.00048001227,0.0007746591,0.00015965258],"domain_scores_gemma":[0.99606776,0.0022821187,0.00044304068,0.00045386382,0.00049854175,0.00025459693],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004967114,0.0017875955,0.0016705642,0.004585905,0.0016136849,0.0027342963,0.0029776786,0.0014905339,0.019997349],"category_scores_gemma":[0.014124917,0.0016728682,0.0011609105,0.0029896635,0.0007670175,0.003995651,0.0029068233,0.001698318,0.017748341],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0029499112,0.00040295997,0.0149031235,0.003303602,0.000499701,0.0007779607,0.0011981226,0.007655056,0.22070502,0.016378146,0.3658251,0.36540127],"study_design_scores_gemma":[0.00089434447,0.00043542747,0.012163836,0.00070070487,0.00022564444,0.0015811182,0.00042270625,0.24891512,0.3259519,0.04803422,0.36019135,0.0004836101],"about_ca_topic_score_codex":0.002277156,"about_ca_topic_score_gemma":0.0030214547,"teacher_disagreement_score":0.019997349,"about_ca_system_score_codex":0.0013364721,"about_ca_system_score_gemma":0.0022047972,"threshold_uncertainty_score":0.06689781},"labels":[],"label_agreement":null},{"id":"W2148683465","doi":"10.1093/bioinformatics/btq332","title":"Savant: genome browser for high-throughput sequencing data","year":2010,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":117,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of Toronto","funders":"Natural Sciences and Engineering Research Council of Canada; Canadian Institutes of Health Research","keywords":"Throughput; Computer science; Genome; DNA sequencing; Genome browser; Computational biology; Biology; Genomics; Genetics; DNA; Gene; Operating system","score_opus":0.03208935018264604,"score_gpt":0.26003887064980213,"score_spread":0.2279495204671561,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2148683465","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.005509686,0.0018448959,0.18466671,0.0013132348,0.00062492705,0.00064449135,0.35798517,0.43627715,0.011133822],"genre_scores_gemma":[0.02487545,0.0015529444,0.32883114,0.0009138044,0.00015196195,0.0023521204,0.5796827,0.05505816,0.006581805],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9981805,0.000622624,0.00020455114,0.0003002927,0.0005313805,0.00016061422],"domain_scores_gemma":[0.9949816,0.0026466195,0.00042422567,0.0005989023,0.00075518945,0.0005934909],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004901134,0.0028802785,0.0019118594,0.0035554306,0.0011365494,0.0027361445,0.0043925177,0.0021809943,0.05725955],"category_scores_gemma":[0.010328654,0.0020468817,0.001589443,0.0040273257,0.0005443852,0.002917647,0.0026384827,0.0032345515,0.029713137],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014280094,0.00010268027,0.0017074597,0.002751568,0.00025823,0.00075325987,0.0006387348,0.0028689927,0.019802384,0.007854022,0.9265075,0.035327263],"study_design_scores_gemma":[0.001192828,0.00020230334,0.0068273214,0.001031691,0.0002149521,0.0014034803,0.00026863095,0.043610297,0.019094009,0.03880033,0.8869898,0.00036433846],"about_ca_topic_score_codex":0.0056184162,"about_ca_topic_score_gemma":0.010372703,"teacher_disagreement_score":0.05725955,"about_ca_system_score_codex":0.00088362035,"about_ca_system_score_gemma":0.0033935742,"threshold_uncertainty_score":0.19155228},"labels":[],"label_agreement":null},{"id":"W2148694532","doi":"10.1093/bioinformatics/bts007","title":"CHROMATRA: a Galaxy tool for visualizing genome-wide chromatin signatures","year":2012,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":14,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Child and Family Research Institute; Simon Fraser University; University of British Columbia","funders":"Canadian Institutes of Health Research; Michael Smith Health Research BC","keywords":"Workflow; Visualization; Computer science; Genome; Computational biology; Software; Chromatin; Genome browser; Gene; Data mining; Biology; Genomics; Genetics; Database; Programming language","score_opus":0.00881905620344901,"score_gpt":0.24328570023083532,"score_spread":0.23446664402738632,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2148694532","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0075336774,0.0013325303,0.3229776,0.00055099826,0.0002083046,0.00015275269,0.18012878,0.4776269,0.009488533],"genre_scores_gemma":[0.072635695,0.0016020915,0.44858003,0.0009865221,0.00014147208,0.0012626959,0.36334133,0.090469025,0.02098115],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9993777,0.00009939729,0.000040499366,0.00013514564,0.00026014552,0.00008697942],"domain_scores_gemma":[0.99905354,0.0003137256,0.00009109124,0.00023461628,0.00017790258,0.00012903877],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00096098904,0.0016262718,0.0017358143,0.0032193244,0.00068635616,0.0019938403,0.0029203338,0.0014350552,0.09203953],"category_scores_gemma":[0.0023063396,0.0012703169,0.0012609444,0.0022889203,0.00030384236,0.0013670998,0.0022383938,0.0018994099,0.042363506],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010330401,0.00008217782,0.0035379585,0.003184313,0.00046722323,0.00056875515,0.0005816473,0.0031737047,0.13177079,0.0051265545,0.68524,0.16523384],"study_design_scores_gemma":[0.00048980856,0.00010960585,0.013365927,0.00029350852,0.00022172391,0.0016561729,0.00023141771,0.040101834,0.12922493,0.0138228545,0.80002004,0.00046220378],"about_ca_topic_score_codex":0.0043925866,"about_ca_topic_score_gemma":0.0071522873,"teacher_disagreement_score":0.09203953,"about_ca_system_score_codex":0.00057162944,"about_ca_system_score_gemma":0.0008303718,"threshold_uncertainty_score":0.30790287},"labels":[],"label_agreement":null},{"id":"W2148762636","doi":"10.1093/bioinformatics/bth436","title":"Modeling interactome: scale-free or geometric?","year":2004,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":712,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Princess Margaret Cancer Centre; University Health Network; University of Toronto","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Computer science; Interactome; Geometric networks; Random graph; Network model; Scale (ratio); Complex network; Scale-free network; Measure (data warehouse); Artificial intelligence; Data mining; Theoretical computer science; Machine learning; Graph","score_opus":0.01418155438029658,"score_gpt":0.24152108462962651,"score_spread":0.22733953024932993,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2148762636","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.18534626,0.0056579155,0.7888477,0.0075438195,0.00021911241,0.0002798148,0.0031106668,0.0019067185,0.0070879753],"genre_scores_gemma":[0.8705749,0.0055040834,0.11628444,0.0011195514,0.00034924058,0.00058223616,0.0031365268,0.0004684779,0.0019806009],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99818015,0.00071762933,0.000068293506,0.00061968534,0.0003413601,0.0000729044],"domain_scores_gemma":[0.9908014,0.00589544,0.0014303299,0.0010997402,0.0004480252,0.0003250491],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0033437093,0.0012947929,0.0010320515,0.0017704571,0.00055529096,0.0017520615,0.0021204806,0.0018008488,0.0019942685],"category_scores_gemma":[0.020895109,0.0005204027,0.0009270709,0.0021182974,0.0019218908,0.0046618027,0.0014174702,0.0011882806,0.0005044465],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00012630787,0.00007855846,0.018039754,0.00089837983,0.0003334129,0.00037351923,0.00029034124,0.8114678,0.0035960136,0.13149828,0.0072121383,0.02608554],"study_design_scores_gemma":[0.00002954701,0.000053996046,0.0033938712,0.00005881025,0.00007685247,0.00038440348,0.000070131704,0.7264813,0.0012485938,0.26338,0.0047844965,0.000037920378],"about_ca_topic_score_codex":0.0022959232,"about_ca_topic_score_gemma":0.0017373983,"teacher_disagreement_score":0.0033437093,"about_ca_system_score_codex":0.0015483997,"about_ca_system_score_gemma":0.00064021203,"threshold_uncertainty_score":0.017683446},"labels":[],"label_agreement":null},{"id":"W2148889460","doi":"10.1093/bioinformatics/bti1128","title":"Recovering haplotype structure through recombination and gene conversion","year":2005,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"","keywords":"Haplotype; Recombination; Gene conversion; Genetics; Gene; Computational biology; Computer science; Biology; Allele","score_opus":0.010641531232120648,"score_gpt":0.23894569477016747,"score_spread":0.22830416353804683,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2148889460","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.12589182,0.00045641308,0.8708289,0.00034218366,0.000018671915,0.00006458613,0.00034898755,0.0006546575,0.0013937288],"genre_scores_gemma":[0.5410882,0.0005894358,0.45408756,0.00008790864,0.000046657657,0.000084550906,0.0016521151,0.00014557905,0.002218021],"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9993562,0.00023322305,0.000037762573,0.00016397685,0.0001463604,0.000062397914],"domain_scores_gemma":[0.9960718,0.0028049056,0.00039756365,0.00032331006,0.00021959361,0.00018288754],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0017677719,0.00046622212,0.0007704738,0.001541683,0.00053652027,0.0009497296,0.0011596505,0.000765761,0.002160523],"category_scores_gemma":[0.0075377426,0.0004020249,0.0007204215,0.001970418,0.0009973538,0.0013828033,0.0012212268,0.0012195783,0.0005769639],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00043940652,0.0002125828,0.027675986,0.00028144606,0.00011689464,0.00041078965,0.00025190026,0.6760049,0.005744928,0.02304694,0.0021143067,0.26369995],"study_design_scores_gemma":[0.00008456082,0.00009214187,0.0043398095,0.000040224444,0.000053217733,0.00062270515,0.00005696414,0.9265677,0.005987495,0.05998928,0.00213393,0.00003202278],"about_ca_topic_score_codex":0.003110111,"about_ca_topic_score_gemma":0.0024566664,"teacher_disagreement_score":0.003110111,"about_ca_system_score_codex":0.00051919674,"about_ca_system_score_gemma":0.0012253154,"threshold_uncertainty_score":0.0093489885},"labels":[],"label_agreement":null},{"id":"W2148967786","doi":"10.1093/bioinformatics/btp431","title":"Path: a tool to facilitate pathway-based genetic association analysis","year":2009,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":20,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"Canadian Institutes of Health Research","keywords":"dbSNP; Single-nucleotide polymorphism; SNP; KEGG; Genome browser; OMIM : Online Mendelian Inheritance in Man; Genetic association; Tag SNP; Genome-wide association study; Computational biology; Computer science; Biology; Genetics; Genome; Genomics; Gene; Genotype; Gene ontology","score_opus":0.017030154304975766,"score_gpt":0.23987850454393989,"score_spread":0.22284835023896413,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2148967786","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.004796984,0.0006171801,0.5890896,0.0011709991,0.0003472209,0.0007470454,0.08580778,0.31222725,0.0051959353],"genre_scores_gemma":[0.043581426,0.00145178,0.81352645,0.0012110686,0.00016092145,0.0048221126,0.094803914,0.03328436,0.007158025],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9988298,0.00039359677,0.00014070464,0.00028169775,0.00027044376,0.00008373116],"domain_scores_gemma":[0.9923765,0.005959536,0.00039922076,0.0005054017,0.00046566312,0.00029374746],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0042203404,0.0019040195,0.0013288517,0.0038122677,0.00084926025,0.002166069,0.002758344,0.0010775811,0.07287227],"category_scores_gemma":[0.013183162,0.0014212081,0.0020879533,0.003067561,0.0006637615,0.002060809,0.004309416,0.0020526024,0.017611582],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0018779399,0.00027647938,0.01357423,0.005716479,0.0013295091,0.0021901992,0.00095400633,0.016022857,0.008264681,0.030351976,0.6823678,0.23707384],"study_design_scores_gemma":[0.0022401696,0.00028534787,0.0109942295,0.0011198941,0.0006915399,0.002822943,0.0003705485,0.15585908,0.013220525,0.1476719,0.6643487,0.000375097],"about_ca_topic_score_codex":0.0028412517,"about_ca_topic_score_gemma":0.0043031946,"teacher_disagreement_score":0.07287227,"about_ca_system_score_codex":0.00075641484,"about_ca_system_score_gemma":0.0030848328,"threshold_uncertainty_score":0.24378204},"labels":[],"label_agreement":null},{"id":"W2149153852","doi":"10.1093/bioinformatics/btr245","title":"Optimally discriminative subnetwork markers predict response to chemotherapy","year":2011,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":99,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University; University of British Columbia","funders":"Canadian Institutes of Health Research","keywords":"Subnetwork; Generalizability theory; Discriminative model; Computer science; Artificial intelligence; Computational biology; Machine learning; Biology; Mathematics; Statistics","score_opus":0.013374317781432606,"score_gpt":0.22781538550732428,"score_spread":0.21444106772589167,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2149153852","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6534908,0.00096874475,0.3359241,0.0009129152,0.000056716664,0.00020252913,0.0028704973,0.0024086442,0.003165057],"genre_scores_gemma":[0.9192028,0.00024229314,0.075801164,0.00007488217,0.000065254244,0.00012815099,0.0030836728,0.000084476196,0.0013171773],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9996505,0.000093297065,0.000028665407,0.000101344114,0.00008526587,0.000040885698],"domain_scores_gemma":[0.99815494,0.0010249758,0.0003724438,0.00011882581,0.00023116807,0.00009757431],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00063123833,0.0008269725,0.0006674089,0.0025303722,0.00025633126,0.0008561829,0.0005029337,0.0005474104,0.0015626737],"category_scores_gemma":[0.004685895,0.0001991666,0.0003721554,0.0012520631,0.00035301765,0.00054333627,0.0005409595,0.00044927088,0.00046325577],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00092450035,0.0003482363,0.1711915,0.00027948088,0.00023466774,0.00021205419,0.00009571197,0.38865188,0.021588575,0.0023501234,0.005593713,0.4085297],"study_design_scores_gemma":[0.000053730342,0.00010657649,0.028963167,0.000026773083,0.00008490344,0.00015009112,0.000030330848,0.95279074,0.009787459,0.0066565285,0.0013298743,0.000019822013],"about_ca_topic_score_codex":0.0024889377,"about_ca_topic_score_gemma":0.0037797864,"teacher_disagreement_score":0.0025303722,"about_ca_system_score_codex":0.00081561937,"about_ca_system_score_gemma":0.00074322283,"threshold_uncertainty_score":0.005917728},"labels":[],"label_agreement":null},{"id":"W2149202898","doi":"10.1093/bioinformatics/btq041","title":"Identifying biologically relevant differences between metagenomic communities","year":2010,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Microbial Community Ecology and Physiology","field":"Environmental Science","cited_by":974,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Dalhousie University","funders":"","keywords":"Metagenomics; Software; Biology; Ecology; Pairwise comparison; Computer science; Computational biology; Data science; Artificial intelligence","score_opus":0.044821397992979205,"score_gpt":0.25582985422559856,"score_spread":0.21100845623261935,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2149202898","genre_codex":"empirical","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6664079,0.0037843627,0.30034888,0.0018853591,0.00018291666,0.00035311934,0.019098312,0.003878611,0.004060526],"genre_scores_gemma":[0.75317186,0.00094623666,0.23510844,0.00030378203,0.0000954472,0.0002560527,0.0092302,0.0003575744,0.0005304589],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9982646,0.00035054228,0.00015136972,0.0006499344,0.00049438956,0.00008927254],"domain_scores_gemma":[0.99318624,0.0036167298,0.0013979019,0.0006450276,0.00076845643,0.00038560093],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0027855902,0.00084729405,0.00060332083,0.0023046427,0.0007788931,0.0016823579,0.0009114254,0.0006448735,0.002307886],"category_scores_gemma":[0.009167531,0.00032037953,0.00076030794,0.0023084492,0.0010111491,0.0013994543,0.0016342774,0.0008992335,0.0007109637],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00085957634,0.00025063226,0.17000833,0.0030517252,0.0005579901,0.0004896257,0.00095118576,0.008311972,0.66747445,0.0041493224,0.0030241744,0.14087099],"study_design_scores_gemma":[0.0001291785,0.00056047505,0.4595034,0.00032616296,0.00071902614,0.0019341543,0.0010672647,0.06594516,0.39959115,0.04908232,0.020894818,0.0002468018],"about_ca_topic_score_codex":0.0007908639,"about_ca_topic_score_gemma":0.00097455486,"teacher_disagreement_score":0.0027855902,"about_ca_system_score_codex":0.0005080354,"about_ca_system_score_gemma":0.00095829455,"threshold_uncertainty_score":0.014731824},"labels":[],"label_agreement":null},{"id":"W2149260201","doi":"10.1093/bioinformatics/bts511","title":"Optimal timepoint sampling in high-throughput gene expression experiments","year":2012,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Lakehead University","funders":"Natural Sciences and Engineering Research Council of Canada; U.S. Department of Energy","keywords":"Sampling (signal processing); Computer science; Throughput; Selection (genetic algorithm); Data mining; False discovery rate; Expression (computer science); Machine learning; Computational biology; Gene; Biology; Genetics","score_opus":0.032554689900855506,"score_gpt":0.2989092124500508,"score_spread":0.2663545225491953,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2149260201","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.027389824,0.00026491145,0.97106296,0.0001814506,0.000017917208,0.000112867194,0.0001598993,0.0005714782,0.00023868638],"genre_scores_gemma":[0.4583853,0.00058708695,0.53778505,0.00021543981,0.00007694221,0.00082652067,0.00096531777,0.00016665769,0.0009917397],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99626726,0.0018214625,0.0001879231,0.00078977115,0.0007836553,0.00015008319],"domain_scores_gemma":[0.98404884,0.012332476,0.001128184,0.0012753678,0.00090743817,0.00030772862],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.009819525,0.0011574494,0.0018381132,0.0008514735,0.0006117448,0.0011420546,0.002285087,0.0015865858,0.0009591632],"category_scores_gemma":[0.023874119,0.0007783192,0.0013105808,0.0016745086,0.0017972029,0.0020826794,0.0011265508,0.0021731849,0.0004623208],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010465853,0.00024646646,0.005874855,0.00036076765,0.00013696904,0.0001732693,0.0001648204,0.89242464,0.02284117,0.012797308,0.0010146924,0.06291853],"study_design_scores_gemma":[0.000054656753,0.00011180483,0.0010289581,0.000010164187,0.000028930066,0.00004659119,0.000015586786,0.9710977,0.011420206,0.015576483,0.00058764796,0.000021282518],"about_ca_topic_score_codex":0.0026384185,"about_ca_topic_score_gemma":0.0023347104,"teacher_disagreement_score":0.009819525,"about_ca_system_score_codex":0.0016941226,"about_ca_system_score_gemma":0.0016426196,"threshold_uncertainty_score":0.051931202},"labels":[],"label_agreement":null},{"id":"W2150153992","doi":"10.1093/bioinformatics/btm560","title":"Petri net-based method for the analysis of the dynamics of signal propagation in signaling pathways","year":2007,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene Regulatory Network Analysis","field":"Biochemistry, Genetics and Molecular Biology","cited_by":45,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Polytechnique Montréal","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Petri net; Computer science; SIGNAL (programming language); Representation (politics); Signal-flow graph; Dynamics (music); Network dynamics; Process (computing); Signal transduction; Systems biology; Theoretical computer science; Computational biology; Distributed computing; Biology; Cell biology; Mathematics; Programming language","score_opus":0.011995817193112481,"score_gpt":0.2528520046799133,"score_spread":0.24085618748680082,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2150153992","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0010972006,0.000048759328,0.99805033,0.000016246964,0.000016600636,0.000021498701,0.000048392656,0.00029411103,0.00040682033],"genre_scores_gemma":[0.09036157,0.0002730763,0.9058548,0.00004633797,0.000042433883,0.00032609786,0.0003888675,0.00020033903,0.0025064864],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99960214,0.00013267285,0.000027451668,0.00006460739,0.00014372416,0.00002951697],"domain_scores_gemma":[0.9987154,0.00094684365,0.000063314896,0.000059118895,0.00016801317,0.000047360063],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009132943,0.0010452066,0.0006636913,0.00179572,0.0005241695,0.0007850635,0.0012668709,0.0006614497,0.0044815335],"category_scores_gemma":[0.0023356634,0.0003759925,0.001169786,0.0010561066,0.0007748211,0.0010497221,0.00059267605,0.0011643858,0.0008819464],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00012863684,0.00008180009,0.0008091879,0.00024790765,0.00008341376,0.000271367,0.00011709703,0.78829056,0.012771754,0.11150962,0.0016890789,0.083999574],"study_design_scores_gemma":[0.000006614463,0.000006808605,0.0000402189,0.0000064160945,0.00000653767,0.000020476342,0.0000032908317,0.9872649,0.0007869656,0.010816553,0.0010357902,0.000005487453],"about_ca_topic_score_codex":0.00487258,"about_ca_topic_score_gemma":0.0038398395,"teacher_disagreement_score":0.00487258,"about_ca_system_score_codex":0.0009880827,"about_ca_system_score_gemma":0.0012840721,"threshold_uncertainty_score":0.0149921775},"labels":[],"label_agreement":null},{"id":"W2150515766","doi":"10.1093/bioinformatics/btq305","title":"Efficient learning of microbial genotype–phenotype association rules","year":2010,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":29,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Dalhousie University","funders":"","keywords":"Biology; Phenotype; Association (psychology); Computational biology; Phylogenetic tree; Genotype; Genome-wide association study; Association rule learning; Genetic association; Genotype-phenotype distinction; Gene; Genetics; Data mining; Computer science; Single-nucleotide polymorphism; Psychology","score_opus":0.006223568153281153,"score_gpt":0.22609678617860202,"score_spread":0.21987321802532087,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2150515766","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.08766766,0.0004006738,0.9070342,0.00034438082,0.00002964138,0.000111469875,0.000981649,0.0020418526,0.0013884868],"genre_scores_gemma":[0.49292666,0.00023741007,0.49997246,0.00021295037,0.00008103597,0.00023187078,0.0048995544,0.00015008765,0.0012880662],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9974577,0.00085822726,0.00020494076,0.00069816085,0.0006118279,0.00016923077],"domain_scores_gemma":[0.98524636,0.011240453,0.0008979383,0.0010673762,0.0013432632,0.00020468912],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0038599756,0.0008718482,0.001612114,0.0024125837,0.00059796934,0.0016225245,0.002396936,0.0012749424,0.0016876784],"category_scores_gemma":[0.015658975,0.0004980682,0.0011277889,0.0024225565,0.0005504146,0.001562423,0.0014859162,0.0018065907,0.0012040174],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00037619213,0.00042198988,0.039551355,0.0002925794,0.0002844618,0.00043439076,0.00021611157,0.24472,0.005447198,0.005008801,0.004425908,0.69882107],"study_design_scores_gemma":[0.000018284338,0.000032220065,0.0016963697,0.000012382154,0.000031970878,0.00011835495,0.00002701781,0.986331,0.0018607369,0.009180196,0.0006834848,0.000007910631],"about_ca_topic_score_codex":0.0022079456,"about_ca_topic_score_gemma":0.00290244,"teacher_disagreement_score":0.0038599756,"about_ca_system_score_codex":0.00050056196,"about_ca_system_score_gemma":0.0012116394,"threshold_uncertainty_score":0.020413756},"labels":[],"label_agreement":null},{"id":"W2150547367","doi":"10.1093/bioinformatics/btu786","title":"Tissue-aware data integration approach for the inference of pathway interactions in metazoan organisms","year":2014,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":103,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Human Genome Research Institute; National Institute of General Medical Sciences; Canadian Institute for Advanced Research","keywords":"Inference; Scalability; Computational biology; Compendium; Computer science; Genome; Biology; Software; Biological data; Source code; Data type; Bioinformatics; Gene; Database; Genetics; Artificial intelligence; Programming language","score_opus":0.03736471980374437,"score_gpt":0.2869714272434773,"score_spread":0.24960670743973293,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2150547367","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.036213193,0.0017047,0.912909,0.0004454977,0.000117319556,0.00017555676,0.024851408,0.021914326,0.001668996],"genre_scores_gemma":[0.14925759,0.00081313855,0.7988067,0.00030104612,0.00007436584,0.00038066032,0.048214074,0.0010411757,0.001111311],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99876046,0.00023393337,0.00009819031,0.00049108855,0.00033566114,0.00008065545],"domain_scores_gemma":[0.998139,0.00088363135,0.00013812818,0.00034077748,0.00033453348,0.00016383664],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0023199033,0.0014733977,0.0012435149,0.005471988,0.0010887702,0.0020753862,0.0028255698,0.0014895437,0.0040681744],"category_scores_gemma":[0.0060139336,0.0009437878,0.004225957,0.0043356973,0.00062302983,0.0016432672,0.003172722,0.0024656514,0.0018289874],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0012638903,0.00044771767,0.06982442,0.0031724873,0.0030274712,0.002763709,0.00091473607,0.42246038,0.06997891,0.027557028,0.029318573,0.36927065],"study_design_scores_gemma":[0.000072165465,0.000097730794,0.007944075,0.000117427924,0.00041497615,0.0005711043,0.00024251798,0.909814,0.015930958,0.04089122,0.02382355,0.000080321886],"about_ca_topic_score_codex":0.010874242,"about_ca_topic_score_gemma":0.022369854,"teacher_disagreement_score":0.010874242,"about_ca_system_score_codex":0.0011423865,"about_ca_system_score_gemma":0.0027192584,"threshold_uncertainty_score":0.021621883},"labels":[],"label_agreement":null},{"id":"W2150717117","doi":"10.1093/bioinformatics/btr444","title":"Bayesian prediction of tissue-regulated splicing using RNA sequence and cellular context","year":2011,"lang":"en","type":"article","venue":"Bioinformatics","topic":"RNA Research and Splicing","field":"Biochemistry, Genetics and Molecular Biology","cited_by":116,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of Toronto","funders":"Canadian Institutes of Health Research","keywords":"RNA splicing; Source code; Context (archaeology); Alternative splicing; Feature (linguistics); Computational biology; Exon; Computer science; Inference; Gibbs sampling; Bayesian probability; Biology; Gene; RNA; Artificial intelligence; Genetics","score_opus":0.03686261043592832,"score_gpt":0.2611324581602233,"score_spread":0.22426984772429498,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2150717117","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.19121216,0.00081841607,0.80320036,0.00023521521,0.000024070305,0.000060568396,0.0018193757,0.0013584013,0.0012713176],"genre_scores_gemma":[0.7767373,0.00044121515,0.21602218,0.00012585464,0.000064709086,0.00011748586,0.004981449,0.00020079967,0.0013090331],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9994388,0.00020775136,0.000033838292,0.00017365396,0.000108759035,0.000037151127],"domain_scores_gemma":[0.9971372,0.0021093895,0.00025184307,0.0001440414,0.00026598512,0.00009146928],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002017773,0.0006997935,0.0007432357,0.0013499793,0.00028898614,0.0006825666,0.0008027855,0.00078349997,0.0014068473],"category_scores_gemma":[0.005440364,0.0004541174,0.00072169583,0.00087711256,0.00046778732,0.0006181393,0.00052848115,0.00092908746,0.0006816107],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009028896,0.00015825899,0.07639172,0.00035337976,0.00027128722,0.00027068687,0.0001587895,0.6962564,0.04625419,0.0068574105,0.0031458514,0.16897914],"study_design_scores_gemma":[0.000028146465,0.000030843876,0.006896109,0.00001655897,0.000033966957,0.00009122508,0.000010976212,0.97730154,0.0055800476,0.009405502,0.000589712,0.000015428272],"about_ca_topic_score_codex":0.0063343607,"about_ca_topic_score_gemma":0.011121531,"teacher_disagreement_score":0.0063343607,"about_ca_system_score_codex":0.0007048391,"about_ca_system_score_gemma":0.00090265815,"threshold_uncertainty_score":0.012594998},"labels":[],"label_agreement":null},{"id":"W2150856587","doi":"10.1093/bioinformatics/bth057","title":"Cluster Analyzer for Transcription Sites (CATS): a C++-based program for identifying clustered transcription factor binding sites","year":2004,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"D-Wave Systems (Canada)","funders":"Howard Hughes Medical Institute","keywords":"Transcription factor; DNA binding site; Consensus sequence; Biology; Computational biology; Transcription (linguistics); Genome; Binding site; CATS; DNA; Genetics; DNA sequencing; Gene; Promoter; Computer science; Base sequence; Gene expression","score_opus":0.03079255120416373,"score_gpt":0.2834839935097028,"score_spread":0.2526914423055391,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2150856587","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.006029956,0.00033175098,0.58834434,0.00023052497,0.00018552071,0.00091974257,0.032327004,0.36763597,0.003995128],"genre_scores_gemma":[0.021643164,0.0003076931,0.89489645,0.0002630565,0.000069294685,0.002671524,0.04335851,0.03159762,0.0051927944],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9989737,0.00017425149,0.00012370571,0.0003202794,0.000309724,0.00009829382],"domain_scores_gemma":[0.99822205,0.0008954777,0.0002006014,0.0001467133,0.0003879388,0.00014715554],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0028072752,0.003193524,0.0016657813,0.002690125,0.0011765485,0.0016895971,0.0038493727,0.00096532446,0.04301214],"category_scores_gemma":[0.0044073607,0.0014900182,0.0015186484,0.0021513668,0.0007266378,0.0016773008,0.0014520199,0.0031208447,0.024001395],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0038547628,0.00039146072,0.0062490143,0.0041093966,0.00064972485,0.000911689,0.001060218,0.01707024,0.07808197,0.0126294885,0.574352,0.30064002],"study_design_scores_gemma":[0.0020111904,0.00067598023,0.011124448,0.00061572244,0.0004202724,0.0016783696,0.00035183592,0.39674488,0.13366629,0.04368518,0.4084252,0.000600643],"about_ca_topic_score_codex":0.003397576,"about_ca_topic_score_gemma":0.0037409249,"teacher_disagreement_score":0.04301214,"about_ca_system_score_codex":0.0010015706,"about_ca_system_score_gemma":0.0020106903,"threshold_uncertainty_score":0.14388996},"labels":[],"label_agreement":null},{"id":"W2150869900","doi":"10.1093/bioinformatics/btt313","title":"SPANNER: taxonomic assignment of sequences using pyramid matching of similarity profiles","year":2013,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":14,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Dalhousie University","funders":"","keywords":"Taxonomic rank; Computer science; Rank (graph theory); Similarity (geometry); Set (abstract data type); Biology; Artificial intelligence; Mathematics; Ecology; Combinatorics","score_opus":0.02345590031270034,"score_gpt":0.23882277232233448,"score_spread":0.21536687200963414,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2150869900","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.03569048,0.00059111457,0.7519349,0.00026190866,0.00024297068,0.0006433722,0.018466268,0.18652382,0.00564522],"genre_scores_gemma":[0.11028772,0.00031311545,0.8417719,0.0002458982,0.00008238679,0.0005100836,0.03518044,0.009137442,0.0024710558],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9978643,0.00022637792,0.00022991786,0.0009240347,0.00063824846,0.00011721886],"domain_scores_gemma":[0.99729925,0.00074203673,0.0005132868,0.0007572244,0.0004898295,0.0001984155],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003293892,0.001926909,0.001126252,0.005479667,0.0011203571,0.0018683278,0.002498315,0.0014930641,0.015899444],"category_scores_gemma":[0.012031008,0.0009578992,0.001390897,0.0042312196,0.00081346114,0.0044660047,0.0033874598,0.0013420839,0.009509721],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.002947028,0.0004631025,0.020838352,0.002776336,0.00069388375,0.0009154664,0.0011824057,0.016154189,0.12422863,0.013717955,0.10088455,0.71519804],"study_design_scores_gemma":[0.00042247068,0.0011847432,0.019215405,0.00046855694,0.0002926545,0.0019097142,0.00070354086,0.64610106,0.13514806,0.06445224,0.12975557,0.00034597688],"about_ca_topic_score_codex":0.0020736982,"about_ca_topic_score_gemma":0.0030407682,"teacher_disagreement_score":0.015899444,"about_ca_system_score_codex":0.0008560997,"about_ca_system_score_gemma":0.0012498356,"threshold_uncertainty_score":0.05318898},"labels":[],"label_agreement":null},{"id":"W2150904363","doi":"10.1093/bioinformatics/btn239","title":"EM-random forest and new measures of variable importance for multi-locus quantitative trait linkage analysis","year":2008,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":21,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Hospital for Sick Children; Lunenfeld-Tanenbaum Research Institute; University of Toronto; SickKids Foundation; Mount Sinai Hospital","funders":"Canadian Institutes of Health Research; Mitacs; National Heart, Lung, and Blood Institute; School of Medicine, Boston University","keywords":"Linkage (software); Random forest; Statistics; Mathematics; Trait; Correlation; Regression; Tree (set theory); Computer science; Artificial intelligence; Biology; Genetics; Combinatorics","score_opus":0.06807023609068835,"score_gpt":0.2974211814087506,"score_spread":0.22935094531806227,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2150904363","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.008849657,0.0004435619,0.9895427,0.00011933247,0.000035248755,0.00006106123,0.0001543002,0.00042295942,0.00037111543],"genre_scores_gemma":[0.22612993,0.00037859677,0.77076143,0.00018988963,0.00023578074,0.0004248192,0.0010514245,0.00027540166,0.0005528162],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99292576,0.004592947,0.0003729637,0.00071921974,0.0011281992,0.0002610291],"domain_scores_gemma":[0.9599758,0.03261435,0.0018425115,0.0021459782,0.0028981012,0.00052326085],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.017418312,0.001349032,0.0016874317,0.0031638923,0.00071680016,0.0014038922,0.0029348652,0.0017030565,0.001848054],"category_scores_gemma":[0.052247643,0.00060657674,0.0011808067,0.002833959,0.0014288818,0.0036070682,0.001864579,0.003024385,0.00052499806],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0004389685,0.00017928067,0.013984789,0.0003302301,0.00038603833,0.00014239231,0.00014756033,0.72782534,0.0015295161,0.05484357,0.0038052646,0.19638698],"study_design_scores_gemma":[0.00003729438,0.000048465452,0.0009293696,0.000026006737,0.000022875083,0.000052460156,0.000009408602,0.95815766,0.0006407573,0.039301477,0.00075393176,0.000020348314],"about_ca_topic_score_codex":0.0016763741,"about_ca_topic_score_gemma":0.0017114963,"teacher_disagreement_score":0.017418312,"about_ca_system_score_codex":0.0011747971,"about_ca_system_score_gemma":0.001014224,"threshold_uncertainty_score":0.092117906},"labels":[],"label_agreement":null},{"id":"W2150937400","doi":"10.1093/bioinformatics/btr470","title":"A phylogenetic mixture model for the identification of functionally divergent protein residues","year":2011,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Protein Structure and Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":36,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Dalhousie University","funders":"Natural Sciences and Engineering Research Council of Canada; Nova Scotia Health Research Foundation","keywords":"False positive paradox; Phylogenetic tree; Classifier (UML); Divergence (linguistics); Computer science; Computational biology; Phylogenetics; Biology; Artificial intelligence; Pattern recognition (psychology); Evolutionary biology; Genetics; Gene","score_opus":0.017432395873248546,"score_gpt":0.22471220463730546,"score_spread":0.2072798087640569,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2150937400","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0812455,0.0015393307,0.9108052,0.0010614702,0.000117775024,0.00013463652,0.0010877217,0.0021305545,0.0018778254],"genre_scores_gemma":[0.7051099,0.00084208016,0.27937406,0.000688883,0.00023726367,0.00047292467,0.005856329,0.00046562462,0.0069529284],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99852717,0.00060421886,0.00007054465,0.00039127425,0.00024419592,0.00016260764],"domain_scores_gemma":[0.9945148,0.0039911075,0.00046476288,0.0002206263,0.00060360326,0.00020518478],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0063315257,0.0013948998,0.0017156524,0.0031851032,0.00098946,0.0017709446,0.0030749736,0.002641124,0.0040308884],"category_scores_gemma":[0.012241287,0.0006912506,0.0019005093,0.0020519218,0.0012045118,0.002483701,0.0016609107,0.002846386,0.0017552169],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00069971976,0.00019521387,0.014338037,0.00018575868,0.00023391287,0.00016105644,0.00018724347,0.87142026,0.002477607,0.018318253,0.005965788,0.08581715],"study_design_scores_gemma":[0.000006183259,0.00000918762,0.00029624713,0.0000068990225,0.0000065470676,0.000015035318,0.000004522574,0.99626833,0.0001326014,0.0030202384,0.00022847907,0.000005759546],"about_ca_topic_score_codex":0.010007001,"about_ca_topic_score_gemma":0.0066553024,"teacher_disagreement_score":0.010007001,"about_ca_system_score_codex":0.0019344747,"about_ca_system_score_gemma":0.0012720943,"threshold_uncertainty_score":0.033484697},"labels":[],"label_agreement":null},{"id":"W2151031196","doi":"10.1093/bioinformatics/btr525","title":"Computational prediction of eukaryotic phosphorylation sites","year":2011,"lang":"en","type":"review","venue":"Bioinformatics","topic":"Machine Learning in Bioinformatics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":157,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Saskatchewan","funders":"","keywords":"Phosphorylation; Kinase; Computational biology; Protein phosphorylation; Mechanism (biology); Resource (disambiguation); Sequence (biology); Posttranslational modification; Biology; Computer science; Bioinformatics; Protein kinase A; Biochemistry; Enzyme","score_opus":0.029212896477216817,"score_gpt":0.2834585629456036,"score_spread":0.2542456664683868,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2151031196","genre_codex":"methods","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"review","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.05318824,0.45074114,0.46189347,0.0029239745,0.00083997275,0.00014412483,0.0057529905,0.0041291434,0.020386998],"genre_scores_gemma":[0.21636204,0.3676453,0.39226246,0.000770142,0.00056336145,0.00026354106,0.014692631,0.00044154358,0.0069989115],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99978036,0.00005068988,0.00002074389,0.00006922238,0.000064749736,0.000014265077],"domain_scores_gemma":[0.99909186,0.0006210749,0.000078104196,0.000039591472,0.00014680612,0.000022538068],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007379685,0.000847286,0.0012609607,0.001357627,0.00017830532,0.00081196305,0.0014344178,0.00059149973,0.0015866222],"category_scores_gemma":[0.002056893,0.00040538807,0.0007722619,0.0017952663,0.00025224203,0.00082084554,0.00046462388,0.0007975146,0.0017161572],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00021096607,0.00010584635,0.0050781956,0.0057799243,0.00041223728,0.00021665716,0.000034172983,0.23146725,0.0051657665,0.009814816,0.0203312,0.7213829],"study_design_scores_gemma":[0.00010772828,0.00019726572,0.0065036947,0.0017297866,0.00044044614,0.001416962,0.000063237916,0.79837376,0.024084592,0.042471338,0.124490015,0.00012125333],"about_ca_topic_score_codex":0.0016379429,"about_ca_topic_score_gemma":0.001666649,"teacher_disagreement_score":0.0016379429,"about_ca_system_score_codex":0.00040900212,"about_ca_system_score_gemma":0.00063487166,"threshold_uncertainty_score":0.005307734},"labels":[],"label_agreement":null},{"id":"W2151043165","doi":"10.1093/bioinformatics/17.7.664","title":"T-REX: reconstructing and visualizing phylogenetic trees and reticulation networks","year":2001,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":156,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Phylogenetic tree; Phylogenetic network; Tree (set theory); Computer science; Inference; Tree structure; Software; Data mining; Artificial intelligence; Data structure; Biology; Combinatorics; Mathematics; Programming language","score_opus":0.013052082182155346,"score_gpt":0.23738933531856438,"score_spread":0.22433725313640904,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2151043165","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.013125177,0.00095484324,0.79672146,0.00094194314,0.00046775656,0.00020059106,0.018820088,0.15711215,0.011655957],"genre_scores_gemma":[0.06610426,0.00147041,0.8339392,0.00023293927,0.00020926232,0.00060295634,0.045216832,0.03750989,0.014714145],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99949205,0.00013445888,0.00003731158,0.00013216841,0.0001534939,0.00005053945],"domain_scores_gemma":[0.9982961,0.0008287596,0.00015985954,0.00025788374,0.00026322974,0.00019424675],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011655865,0.0015548954,0.0011566525,0.00224226,0.0010346984,0.002127124,0.0029352477,0.0016471377,0.075347915],"category_scores_gemma":[0.0044093407,0.0008745139,0.0010459268,0.002473578,0.0005860461,0.0021783174,0.0013043948,0.001845915,0.029553795],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009234534,0.0001356076,0.0026851676,0.0031715455,0.0001565515,0.0011747971,0.0013156349,0.015722629,0.08822586,0.02728687,0.38968837,0.4695135],"study_design_scores_gemma":[0.00032863065,0.00025799117,0.009492657,0.000569471,0.00017137162,0.0030190628,0.00047007683,0.2753011,0.10883721,0.04373319,0.55746275,0.00035653033],"about_ca_topic_score_codex":0.0013701787,"about_ca_topic_score_gemma":0.0013626646,"teacher_disagreement_score":0.075347915,"about_ca_system_score_codex":0.00037024927,"about_ca_system_score_gemma":0.00061466434,"threshold_uncertainty_score":0.25206387},"labels":[],"label_agreement":null},{"id":"W2151232515","doi":"10.1093/bioinformatics/btn611","title":"Sparse combinatorial inference with an application in cancer biology","year":2008,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene Regulatory Network Analysis","field":"Biochemistry, Genetics and Molecular Biology","cited_by":20,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Kinexus Bioinformatics Corporation (Canada)","funders":"Basic Energy Sciences; National Institutes of Health; National Cancer Institute; Office of Science; AstraZeneca; U.S. Department of Energy","keywords":"Inference; Computer science; Curse of dimensionality; Boolean function; Focus (optics); Statistical inference; Function (biology); Theoretical computer science; Boolean expression; Boolean network; Machine learning; Artificial intelligence; Algorithm; Mathematics; Biology; Statistics","score_opus":0.015231213001381477,"score_gpt":0.2626038295757963,"score_spread":0.24737261657441484,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2151232515","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0159589,0.0006459114,0.98045117,0.0013153076,0.000050979444,0.00005357304,0.0002795867,0.00035607402,0.00088843406],"genre_scores_gemma":[0.49930295,0.001110755,0.4932009,0.0010084661,0.00064032205,0.00035113882,0.0015264467,0.00018358599,0.0026754548],"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9961085,0.0024964793,0.00014487348,0.00049966405,0.0006307389,0.00011969429],"domain_scores_gemma":[0.9302941,0.064679265,0.0016265645,0.0015757733,0.0012295623,0.0005946895],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008716751,0.0010349463,0.0022445172,0.0025426997,0.0011153442,0.001702024,0.002152403,0.0018691587,0.0029534805],"category_scores_gemma":[0.050182875,0.0007228016,0.0016410789,0.0035944683,0.003045952,0.002369706,0.002402782,0.0031009153,0.00033229336],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00021910627,0.00015328852,0.0035530883,0.0003075424,0.0001745998,0.00041106317,0.00016186589,0.7837057,0.0012178387,0.15516846,0.002904568,0.052022878],"study_design_scores_gemma":[0.000031258263,0.000018701472,0.000238613,0.000009832565,0.0000119150955,0.00003981331,0.0000099464005,0.8586041,0.00024920463,0.14034228,0.00043371902,0.0000107578135],"about_ca_topic_score_codex":0.005190304,"about_ca_topic_score_gemma":0.004750373,"teacher_disagreement_score":0.008716751,"about_ca_system_score_codex":0.0017451606,"about_ca_system_score_gemma":0.0014769202,"threshold_uncertainty_score":0.046099126},"labels":[],"label_agreement":null},{"id":"W2151471947","doi":"10.1093/bioinformatics/btv219","title":"ESPRESSO: taking into account assessment errors on outcome and exposures in power analysis for association studies","year":2015,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Advanced Causal Inference Techniques","field":"Mathematics","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"FP7 Health; Canadian Institutes of Health Research; European Commission","keywords":"Association (psychology); Outcome (game theory); Computer science; Meta-analysis; Econometrics; Statistics; Psychology; Environmental health; Medicine; Mathematics; Internal medicine","score_opus":0.2878465843490817,"score_gpt":0.4961217734300007,"score_spread":0.20827518908091897,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2151471947","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":"methods","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":"methods","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0033610906,0.00055601005,0.955059,0.0025204923,0.00043587325,0.0009857652,0.0052764486,0.018113568,0.013691739],"genre_scores_gemma":[0.04320765,0.00041730664,0.92769194,0.0014428811,0.00029132032,0.0046047107,0.003214306,0.012902826,0.006227082],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9625031,0.029689407,0.002132621,0.0020906923,0.0032668752,0.00031727794],"domain_scores_gemma":[0.7271067,0.24573383,0.0059078066,0.015046517,0.005519036,0.0006861599],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.07365601,0.001506688,0.0012375583,0.0025052356,0.00080019224,0.0029075865,0.0028302032,0.0022383428,0.09908532],"category_scores_gemma":[0.30136657,0.0015873678,0.0028288108,0.0029569722,0.0019098386,0.003749296,0.0045270436,0.0029876984,0.024278333],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010833397,0.00016279257,0.011672919,0.005304643,0.0012109946,0.00061781757,0.0019971526,0.016783584,0.0023183818,0.14857924,0.28582934,0.5244398],"study_design_scores_gemma":[0.0016720621,0.0005330437,0.0129426215,0.003374276,0.00087194645,0.002141887,0.00027274922,0.1572784,0.010833216,0.40432325,0.40549225,0.0002643271],"about_ca_topic_score_codex":0.0014440193,"about_ca_topic_score_gemma":0.001858074,"teacher_disagreement_score":0.926344,"about_ca_system_score_codex":0.0010582241,"about_ca_system_score_gemma":0.002952481,"threshold_uncertainty_score":0.38953483},"labels":[],"label_agreement":null},{"id":"W2151519751","doi":"10.1093/bioinformatics/bti781","title":"Highly consistent patterns for inherited human diseases at the molecular level","year":2005,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Rare Diseases","field":"Biochemistry, Genetics and Molecular Biology","cited_by":53,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"Medical Research Council","keywords":"Biology; Gene; Functional genomics; Genetics; Disease; Identification (biology); Genomics; Computational biology; Phenotype; Human genetics; Human genome; Molecular genetics; Genome; Medicine","score_opus":0.020930450685390543,"score_gpt":0.25592990692722045,"score_spread":0.2349994562418299,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2151519751","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9379292,0.005773274,0.032073203,0.0010739948,0.00014354865,0.000080139755,0.0084087355,0.0011822063,0.013335638],"genre_scores_gemma":[0.97769165,0.00095659465,0.013286888,0.00026911285,0.000056346747,0.000057528054,0.0052872836,0.000121854486,0.0022726979],"study_design_codex":"bench_or_experimental","study_design_gemma":"observational","domain_scores_codex":[0.99910945,0.00013286913,0.00009433971,0.0003676389,0.00021655907,0.0000791261],"domain_scores_gemma":[0.9969699,0.0009678188,0.0007962507,0.00061857526,0.00041660693,0.00023092335],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006319456,0.00025284014,0.00031844128,0.0019299173,0.0005307305,0.0005851943,0.00028021928,0.0004314071,0.0050788634],"category_scores_gemma":[0.0025948812,0.00014447626,0.00030445826,0.0013277769,0.0004959125,0.00034732957,0.00067808514,0.0005714317,0.0015303844],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014521838,0.00017410533,0.2966867,0.0010356528,0.00072041224,0.0035140556,0.0011305349,0.0008633412,0.5743905,0.0064551635,0.00711176,0.10646559],"study_design_scores_gemma":[0.000042981454,0.00019866918,0.91451615,0.00005543488,0.00028776893,0.013614909,0.00043535803,0.0014680673,0.042194888,0.0058414205,0.021285096,0.00005934388],"about_ca_topic_score_codex":0.00045172748,"about_ca_topic_score_gemma":0.0010365646,"teacher_disagreement_score":0.0050788634,"about_ca_system_score_codex":0.00016780036,"about_ca_system_score_gemma":0.00019867005,"threshold_uncertainty_score":0.016990483},"labels":[],"label_agreement":null},{"id":"W2151743318","doi":"10.1093/bioinformatics/btn173","title":"Optimal pooling for genome re-sequencing with ultra-high-throughput short-read technologies","year":2008,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Algorithms and Data Compression","field":"Computer Science","cited_by":27,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University; BC Cancer Agency","funders":"","keywords":"Pooling; Multiplex; DNA sequencing; Computer science; Bacterial artificial chromosome; Genome; Deep sequencing; Reference genome; Computational biology; Throughput; Algorithm; Biology; Genetics; Artificial intelligence; Gene; Telecommunications","score_opus":0.037241946335152605,"score_gpt":0.24526871172737083,"score_spread":0.20802676539221823,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2151743318","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.01208224,0.00020116135,0.9863891,0.00007175002,0.000018316814,0.00008837338,0.000044712946,0.0007109549,0.00039334391],"genre_scores_gemma":[0.07267782,0.00016540506,0.92540044,0.00008625695,0.000021851853,0.00033501413,0.00034872352,0.0002014584,0.00076302694],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99587446,0.0020353931,0.00021517058,0.0007571304,0.0009308836,0.00018702904],"domain_scores_gemma":[0.9914349,0.0048657227,0.00067734625,0.002193631,0.00064397365,0.00018445405],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0062372587,0.0013562564,0.0018920233,0.0013012821,0.00069497275,0.0013493784,0.0025157556,0.0010401747,0.002732957],"category_scores_gemma":[0.011059752,0.00096484536,0.0013331432,0.0020277933,0.0011389002,0.002832426,0.0017105653,0.0013670335,0.0010708852],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010770735,0.00045416376,0.0022308843,0.00053220795,0.00019496852,0.0002585087,0.00026299537,0.34022793,0.10925559,0.03190239,0.002647197,0.51095605],"study_design_scores_gemma":[0.00008495102,0.00030162075,0.00097062276,0.000026256908,0.000064750224,0.00013100801,0.00007277841,0.8729265,0.074812695,0.04629155,0.004276923,0.000040317747],"about_ca_topic_score_codex":0.0011312901,"about_ca_topic_score_gemma":0.0017904292,"teacher_disagreement_score":0.0062372587,"about_ca_system_score_codex":0.0015816672,"about_ca_system_score_gemma":0.0012999338,"threshold_uncertainty_score":0.032986164},"labels":[],"label_agreement":null},{"id":"W2151781607","doi":"10.1093/bioinformatics/btu463","title":"Polytomy refinement for the correction of dubious duplications in gene trees","year":2014,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":24,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal; Simon Fraser University","funders":"Natural Sciences and Engineering Research Council of Canada; Fonds de recherche du Québec – Nature et technologies; Ministero dell’Istruzione, dell’Università e della Ricerca","keywords":"Source code; Tree (set theory); Computer science; Heuristic; Ensembl; Set (abstract data type); Bounded function; Gene duplication; Graph; Feature (linguistics); Binary tree; Code (set theory); Software; Theoretical computer science; Gene; Data mining; Algorithm; Artificial intelligence; Biology; Mathematics; Combinatorics; Genetics; Genomics; Programming language; Genome","score_opus":0.008654429824636958,"score_gpt":0.23043543154066712,"score_spread":0.22178100171603016,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2151781607","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.06009014,0.00033826302,0.93373615,0.00045342624,0.00007506816,0.00014426006,0.00019340208,0.0029688478,0.0020004828],"genre_scores_gemma":[0.15501465,0.000099113204,0.8421383,0.00017043306,0.000023581131,0.00012328969,0.00045238007,0.0008115653,0.0011666646],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9987797,0.00031553418,0.00006983759,0.00039640354,0.00029211352,0.0001463506],"domain_scores_gemma":[0.99263316,0.0044164145,0.00065202254,0.0014604535,0.00061765633,0.00022021685],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0024774333,0.00093298,0.0012744265,0.0012894088,0.0011756943,0.0008290364,0.0024885517,0.0015475883,0.0047927215],"category_scores_gemma":[0.0122274505,0.00062416063,0.0013318831,0.00122774,0.0019749873,0.0018013152,0.0020798114,0.0022395784,0.0013690967],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006056337,0.00024971185,0.008771572,0.0005925709,0.00014443291,0.0005230567,0.0009095615,0.58849674,0.02916142,0.03841831,0.009631176,0.32249576],"study_design_scores_gemma":[0.00008899706,0.00010079275,0.0010734394,0.000057238005,0.00006169827,0.00035718599,0.00014268668,0.92810744,0.012536315,0.052639097,0.0048052,0.000029918308],"about_ca_topic_score_codex":0.0035620625,"about_ca_topic_score_gemma":0.006846643,"teacher_disagreement_score":0.0047927215,"about_ca_system_score_codex":0.001264552,"about_ca_system_score_gemma":0.0018654739,"threshold_uncertainty_score":0.016033232},"labels":[],"label_agreement":null},{"id":"W2151790371","doi":"10.1093/bioinformatics/btg447","title":"Predicting subcellular localization of proteins using machine-learned classifiers","year":2004,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Machine Learning in Bioinformatics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":334,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Subcellular localization; Computer science; Artificial intelligence; Protein subcellular localization prediction; Machine learning; Computational biology; Pattern recognition (psychology); Chemistry; Biology; Biochemistry; Cytoplasm; Gene","score_opus":0.016360328509057477,"score_gpt":0.2565539395148381,"score_spread":0.24019361100578066,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2151790371","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.53514755,0.003599474,0.44152,0.0017143125,0.00028975008,0.00017075046,0.005740761,0.0062270835,0.0055903285],"genre_scores_gemma":[0.81001717,0.0009264064,0.17732978,0.00030200309,0.0001677151,0.00012532993,0.00926194,0.00010191603,0.00176776],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99927133,0.00018487805,0.00007172086,0.00018863648,0.000188052,0.000095294934],"domain_scores_gemma":[0.99455607,0.003391995,0.00047625246,0.00023728046,0.0011929589,0.00014542838],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016057312,0.0009999998,0.00076926436,0.0018278838,0.0005086212,0.0012248182,0.0008635198,0.0013011743,0.0013746911],"category_scores_gemma":[0.005892015,0.00020127228,0.0005536355,0.0012728906,0.00038843235,0.001383709,0.00039160118,0.0011370741,0.0017040045],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00067537755,0.00072257145,0.13028637,0.0007647161,0.00027356023,0.00068879966,0.00013304243,0.36421663,0.027941193,0.002958906,0.018975098,0.45236376],"study_design_scores_gemma":[0.000030022917,0.000083383326,0.006273776,0.00005533452,0.00007263832,0.00019892227,0.000042892632,0.9678016,0.019131072,0.004038227,0.002253691,0.000018458275],"about_ca_topic_score_codex":0.0030084017,"about_ca_topic_score_gemma":0.0025071967,"teacher_disagreement_score":0.0030084017,"about_ca_system_score_codex":0.000727628,"about_ca_system_score_gemma":0.00097432465,"threshold_uncertainty_score":0.008492053},"labels":[],"label_agreement":null},{"id":"W2151908672","doi":"10.1093/bioinformatics/btv318","title":"Computer vision-based automated peak picking applied to protein NMR spectra","year":2015,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Protein Structure and Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":28,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"University of Waterloo; King Abdullah University of Science and Technology","keywords":"Computer science; NMR spectra database; Identification (biology); Spectral line; Artificial intelligence; Algorithm; Pattern recognition (psychology); Automated method; Biological system; Physics; Biology","score_opus":0.008174705558512439,"score_gpt":0.24156195849310316,"score_spread":0.23338725293459073,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2151908672","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.009977541,0.00015966195,0.9839342,0.000058361133,0.00003007708,0.00006845574,0.00007534564,0.005051099,0.00064522895],"genre_scores_gemma":[0.11054981,0.00016153457,0.8868578,0.00011047989,0.000028509638,0.00012551168,0.00046149205,0.00050152134,0.0012033671],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9987489,0.00028886276,0.00005363197,0.00038085712,0.00041133532,0.00011648716],"domain_scores_gemma":[0.99813604,0.0007206739,0.00014049443,0.00026100568,0.00064745574,0.00009430803],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0017599403,0.0012025904,0.0010738755,0.0023026941,0.00057107396,0.0013359911,0.0020034676,0.0016418581,0.0034614527],"category_scores_gemma":[0.0041316524,0.00068969466,0.0011141304,0.0017995465,0.0009768575,0.0008939983,0.0013365301,0.001416669,0.0024098493],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00037455576,0.0002560255,0.0009795119,0.00041562787,0.0001083581,0.00019449068,0.00014524523,0.12807369,0.16910732,0.0036999919,0.005781127,0.690864],"study_design_scores_gemma":[0.000021845337,0.00010558181,0.0011731597,0.000021122363,0.000015706675,0.00019348868,0.000019145335,0.9241907,0.06781183,0.0037761221,0.002633443,0.000037972637],"about_ca_topic_score_codex":0.003231979,"about_ca_topic_score_gemma":0.0020351724,"teacher_disagreement_score":0.0034614527,"about_ca_system_score_codex":0.00086814683,"about_ca_system_score_gemma":0.0012832987,"threshold_uncertainty_score":0.011579692},"labels":[],"label_agreement":null},{"id":"W2152089378","doi":"10.1093/bioinformatics/18.10.1398","title":"Genquire: genome annotation browser/editor","year":2002,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"National Research Council Canada; Plant Biotechnology Institute","funders":"Genomic Health","keywords":"Perl; Annotation; Computer science; Software; Markup language; World Wide Web; Visualization; Gene Annotation; Plug-in; Download; Genome; Programming language; Information retrieval; XML; Biology; Data mining; Artificial intelligence; Gene; Genetics","score_opus":0.014212092307792955,"score_gpt":0.21323891375709833,"score_spread":0.19902682144930536,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2152089378","genre_codex":"dataset","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00081810064,0.001270378,0.13988143,0.0008835705,0.0014778026,0.00050654716,0.49171644,0.3438398,0.019605983],"genre_scores_gemma":[0.004672585,0.0011209114,0.19468722,0.0012464945,0.00020917221,0.0013392825,0.6915784,0.087844506,0.017301416],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9989961,0.000186431,0.00015389046,0.0002615312,0.00030163786,0.00010042179],"domain_scores_gemma":[0.9981833,0.00053027645,0.00018369906,0.00029728003,0.0006027742,0.00020254552],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0032206145,0.004191674,0.0029902093,0.0053356485,0.0012629812,0.0039329072,0.006207366,0.0021619094,0.20220062],"category_scores_gemma":[0.0050707115,0.0030160504,0.0018006251,0.005656236,0.0004712157,0.0026911178,0.002806221,0.004483806,0.14424439],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001848896,0.00003157805,0.00027114237,0.0013639929,0.00007211624,0.00018411734,0.000120363664,0.00038596737,0.0039655142,0.0021301124,0.96841305,0.022877077],"study_design_scores_gemma":[0.00015699358,0.000019158337,0.0006745468,0.0002535659,0.00003538737,0.00033047708,0.000050949548,0.002350513,0.0043552746,0.0050543975,0.986654,0.00006469474],"about_ca_topic_score_codex":0.004617578,"about_ca_topic_score_gemma":0.006672091,"teacher_disagreement_score":0.20220062,"about_ca_system_score_codex":0.00087416027,"about_ca_system_score_gemma":0.0029177761,"threshold_uncertainty_score":0.67642844},"labels":[],"label_agreement":null},{"id":"W2152433810","doi":"10.1093/bioinformatics/btv126","title":"GeLL: a generalized likelihood library for phylogenetic models","year":2015,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Evolution and Paleontology Studies","field":"Earth and Planetary Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Dalhousie University","funders":"Biotechnology and Biological Sciences Research Council","keywords":"Phylogenetic tree; Sequence (biology); Substitution (logic); Computer science; Java; License; Data mining; Theoretical computer science; Information retrieval; Algorithm; Artificial intelligence; Programming language; Biology","score_opus":0.0546363129713933,"score_gpt":0.23630129757967255,"score_spread":0.18166498460827926,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2152433810","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00047836194,0.00039212173,0.7988668,0.00042324373,0.00014156124,0.00011879909,0.016403722,0.17968294,0.0034925323],"genre_scores_gemma":[0.011790015,0.000715532,0.8354229,0.00072263245,0.00023870147,0.0014741117,0.034876306,0.10670013,0.008059733],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99734014,0.0010120092,0.0002930923,0.00045200155,0.00071300974,0.00018974005],"domain_scores_gemma":[0.9907672,0.005964958,0.000676741,0.001224345,0.0010085368,0.00035822272],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0046041715,0.0038054197,0.0029227438,0.004943567,0.0015270442,0.003590928,0.010462257,0.0036520236,0.15403767],"category_scores_gemma":[0.02107207,0.0032186985,0.0047595957,0.00405657,0.0011023586,0.005060452,0.005247153,0.00677531,0.09499194],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00051296793,0.00021722565,0.0026016573,0.003703734,0.0006479936,0.00095732725,0.00044831086,0.047518786,0.0050770873,0.059156217,0.5874467,0.29171193],"study_design_scores_gemma":[0.0006079273,0.00009028117,0.0010394931,0.0008547409,0.00019086878,0.001152658,0.00008352741,0.30384967,0.008023322,0.18470585,0.49902487,0.00037683133],"about_ca_topic_score_codex":0.0032688798,"about_ca_topic_score_gemma":0.0050298455,"teacher_disagreement_score":0.15403767,"about_ca_system_score_codex":0.0013188946,"about_ca_system_score_gemma":0.0030886491,"threshold_uncertainty_score":0.5153073},"labels":[],"label_agreement":null},{"id":"W2152852962","doi":"10.1093/bioinformatics/btt630","title":"ISCB/SPRINGER series in computational biology","year":2013,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetics, Bioinformatics, and Biomedical Research","field":"Biochemistry, Genetics and Molecular Biology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"National Institutes of Health","keywords":"Computer science; Computational genomics; Publication; Systems biology; Data science; Computational biology; Genomics; Biology; Genome; Genetics","score_opus":0.015310030544391024,"score_gpt":0.2721442721398233,"score_spread":0.2568342415954323,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2152852962","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.001203286,0.12155012,0.07967751,0.013789559,0.04106798,0.0005228424,0.023217825,0.014882165,0.7040887],"genre_scores_gemma":[0.009473898,0.1243245,0.07360521,0.006746522,0.011908848,0.0011737419,0.041301493,0.013494263,0.71797156],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99652183,0.00040556752,0.0003456345,0.0007359145,0.0018212505,0.00016978795],"domain_scores_gemma":[0.9942624,0.0016120347,0.00039287715,0.0013888163,0.00169306,0.00065077114],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0024904844,0.002334557,0.0027377633,0.0052613737,0.0010711659,0.012539079,0.0026351484,0.00307754,0.32657772],"category_scores_gemma":[0.011817001,0.0012662507,0.0016886089,0.010761603,0.0011785907,0.007703028,0.003880388,0.0065317173,0.3446293],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000029208211,0.00003261663,0.00014102155,0.00076688366,0.000043266056,0.000058644073,0.00005466542,0.000891398,0.0005132173,0.020982644,0.77667534,0.19981113],"study_design_scores_gemma":[0.0000082257075,0.000008192309,0.0001966749,0.00035419053,0.000009321969,0.00012294391,0.00003109582,0.0006682187,0.00015344161,0.011858576,0.98657167,0.000017460134],"about_ca_topic_score_codex":0.0019280629,"about_ca_topic_score_gemma":0.001650307,"teacher_disagreement_score":0.32657772,"about_ca_system_score_codex":0.0016325884,"about_ca_system_score_gemma":0.003895069,"threshold_uncertainty_score":0.96055514},"labels":[],"label_agreement":null},{"id":"W2153109441","doi":"10.1093/bioinformatics/btu494","title":"STAMP: statistical analysis of taxonomic and functional profiles","year":2014,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Statistical Methods in Clinical Trials","field":"Mathematics","cited_by":4314,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Dalhousie University","funders":"","keywords":"Python (programming language); Computer science; Graphical user interface; Software; R package; Exploratory data analysis; Statistical analysis; Data mining; Source code; Software package; Information retrieval; Programming language; Statistics; Mathematics","score_opus":0.36813942562374635,"score_gpt":0.47157612010780703,"score_spread":0.10343669448406068,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2153109441","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.022928745,0.0015017183,0.61907774,0.002430947,0.0019563886,0.0028138966,0.2913143,0.04659692,0.011379375],"genre_scores_gemma":[0.16967493,0.0011535285,0.61794955,0.0017593008,0.001247934,0.021874422,0.14857046,0.025013864,0.012756022],"study_design_codex":"not_applicable","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.97783214,0.007996794,0.0025146834,0.004227796,0.0064882557,0.00094027806],"domain_scores_gemma":[0.88725084,0.07546181,0.012392394,0.0139541,0.008773975,0.0021669897],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.020037623,0.0015200032,0.0031628895,0.008829535,0.0013557373,0.0037349302,0.003355234,0.0014193726,0.11439099],"category_scores_gemma":[0.14167549,0.0007712414,0.003383039,0.010243278,0.002401869,0.0034687861,0.0032365324,0.0040961555,0.023146661],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0038207846,0.0004108643,0.043526057,0.009883556,0.002933931,0.000636265,0.0015278058,0.011797553,0.009364958,0.028613856,0.49038333,0.397101],"study_design_scores_gemma":[0.0010121934,0.0021447393,0.10197986,0.003091398,0.002652654,0.0018354255,0.0011110357,0.11136561,0.017335363,0.17541915,0.58127576,0.0007767494],"about_ca_topic_score_codex":0.0016313329,"about_ca_topic_score_gemma":0.0019690122,"teacher_disagreement_score":0.11439099,"about_ca_system_score_codex":0.0013563425,"about_ca_system_score_gemma":0.0046116314,"threshold_uncertainty_score":0.382676},"labels":[],"label_agreement":null},{"id":"W2153167299","doi":"10.1093/bioinformatics/btr035","title":"OrthoNets: simultaneous visual analysis of orthologs and their interaction neighborhoods across different organisms","year":2011,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canada Research Chairs; Hospital for Sick Children; University of Toronto","funders":"Canadian Institutes of Health Research; Hospital for Sick Children","keywords":"Identification (biology); Plug-in; Organism; Computer science; Protein–protein interaction; Computational biology; Model organism; Protein Interaction Networks; Biology; Genetics; Programming language; Gene; Ecology","score_opus":0.012393129576010705,"score_gpt":0.2534325284217612,"score_spread":0.2410393988457505,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2153167299","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.24046761,0.0017806323,0.44514012,0.0010051213,0.00055499654,0.000467341,0.10190696,0.18846107,0.020216078],"genre_scores_gemma":[0.48815393,0.0013655387,0.3927111,0.00026095897,0.00012532005,0.0007278318,0.09243309,0.014272839,0.0099493675],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9997267,0.000040452815,0.000017961582,0.0000863144,0.000097549,0.000030911953],"domain_scores_gemma":[0.9991283,0.0003631911,0.0001030484,0.0001512342,0.000111014495,0.00014323017],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005700231,0.00082275126,0.00065618363,0.0030547483,0.0005600469,0.0012779308,0.0009953154,0.0004609052,0.020010076],"category_scores_gemma":[0.002052081,0.00040215827,0.0007861961,0.0016506191,0.00027995702,0.0019499813,0.002259084,0.00069428765,0.0026346482],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0037398872,0.00033132333,0.025407353,0.005017562,0.0010609942,0.0025945087,0.0019297107,0.017684892,0.34763727,0.02188759,0.23372439,0.33898455],"study_design_scores_gemma":[0.0006804045,0.0005085658,0.10058891,0.0006658357,0.00061499944,0.005049371,0.0014411755,0.27792457,0.16532129,0.0743968,0.37224725,0.0005608626],"about_ca_topic_score_codex":0.0021538322,"about_ca_topic_score_gemma":0.003855512,"teacher_disagreement_score":0.020010076,"about_ca_system_score_codex":0.00024060962,"about_ca_system_score_gemma":0.0005391049,"threshold_uncertainty_score":0.06694043},"labels":[],"label_agreement":null},{"id":"W2153369847","doi":"10.1093/bioinformatics/btq588","title":"Interactive microbial genome visualization with GView","year":2010,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":400,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Manitoba; Public Health Agency of Canada; University of Alberta","funders":"","keywords":"Bitmap; Computer science; Zoom; Application programming interface; Interface (matter); Java; Documentation; File format; Visualization; Context (archaeology); Source code; Scalable Vector Graphics; Programming language; Graphical user interface; MIT License; Scalability; User interface; Operating system; World Wide Web; Computer graphics (images); License; Data mining; Biology","score_opus":0.006221952275581206,"score_gpt":0.23270774913820616,"score_spread":0.22648579686262496,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2153369847","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0090901,0.00063426787,0.45569766,0.00074917375,0.00041907295,0.00020558493,0.032397676,0.47852957,0.022276947],"genre_scores_gemma":[0.13105257,0.001316108,0.6520041,0.0009339183,0.00034929177,0.0009889537,0.100255296,0.0882274,0.024872353],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99956053,0.00006894615,0.00003448588,0.000097835175,0.00017393085,0.00006426518],"domain_scores_gemma":[0.9990183,0.0004559599,0.000054402713,0.00014274073,0.00020786426,0.000120762976],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00095447566,0.0011489394,0.00077318965,0.0023331682,0.0004062486,0.0015761715,0.0019515696,0.00089637033,0.08995217],"category_scores_gemma":[0.0033720103,0.00059023005,0.0010982241,0.0014766002,0.00023218089,0.0016698574,0.0031830645,0.0015010135,0.0167938],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010826538,0.00014256583,0.002023908,0.0020758146,0.00025751712,0.00076058425,0.00095085416,0.0070870523,0.057151906,0.011808535,0.60271037,0.31394824],"study_design_scores_gemma":[0.00058551034,0.00013963672,0.00565584,0.0005239446,0.00013976473,0.0011734723,0.00026444366,0.10196552,0.06962352,0.026337689,0.79329634,0.00029438938],"about_ca_topic_score_codex":0.0019018378,"about_ca_topic_score_gemma":0.0017055586,"teacher_disagreement_score":0.08995217,"about_ca_system_score_codex":0.00032881365,"about_ca_system_score_gemma":0.0006892374,"threshold_uncertainty_score":0.30092},"labels":[],"label_agreement":null},{"id":"W2153508883","doi":"10.1093/bioinformatics/btp103","title":"An integrative scoring system for ranking SNPs by their potential deleterious effects","year":2009,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":96,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Queen's University","funders":"","keywords":"Single-nucleotide polymorphism; SNP genotyping; SNP; Genetics; Biology; Human genome; Computational biology; Genome-wide association study; Genotyping; Genetic association; Bioinformatics; Gene; Genome; Genotype","score_opus":0.005550570813302565,"score_gpt":0.24179419929864818,"score_spread":0.2362436284853456,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2153508883","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.063906066,0.00048461073,0.9250195,0.00046079638,0.00013408603,0.00038710967,0.001293569,0.0055504204,0.0027638883],"genre_scores_gemma":[0.34809446,0.00029500906,0.6442992,0.0003107216,0.00025165113,0.0005701332,0.0031410875,0.00046311028,0.00257465],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9974668,0.00054893224,0.0003054767,0.000510047,0.0010062711,0.0001624844],"domain_scores_gemma":[0.99091154,0.003648987,0.00066644023,0.0007794159,0.003522004,0.00047151762],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0054058875,0.0014527251,0.0016164067,0.005974508,0.00092755817,0.0019038419,0.0017892307,0.0010874259,0.0047650943],"category_scores_gemma":[0.015851824,0.00038545462,0.00097447034,0.003670015,0.0007021115,0.0016709714,0.0017604966,0.0011715106,0.0016125691],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00074251613,0.00039205354,0.050882578,0.00031188125,0.00043611368,0.00038835278,0.00046839507,0.062490538,0.020921826,0.010487831,0.020959085,0.8315189],"study_design_scores_gemma":[0.00022951525,0.00050545944,0.022909505,0.000076976285,0.00036292934,0.0014122553,0.00015943302,0.9304882,0.010912907,0.02265841,0.010069138,0.00021522219],"about_ca_topic_score_codex":0.0035914485,"about_ca_topic_score_gemma":0.003760825,"teacher_disagreement_score":0.005974508,"about_ca_system_score_codex":0.0009317416,"about_ca_system_score_gemma":0.0016313415,"threshold_uncertainty_score":0.028589427},"labels":[],"label_agreement":null},{"id":"W2153606072","doi":"10.1093/bioinformatics/bth079","title":"Automated ordering of fingerprinted clones","year":2004,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":20,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Genome British Columbia","funders":"","keywords":"Contig; Computer science; Fingerprint (computing); Software; clone (Java method); Sequence assembly; In silico; Data mining; Genome; Artificial intelligence; Biology; Genetics; DNA; Programming language; Gene","score_opus":0.0059434600040913456,"score_gpt":0.22119150124769435,"score_spread":0.215248041243603,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2153606072","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.20417692,0.0013268964,0.7629663,0.00040785212,0.00009583539,0.00046822996,0.0026524328,0.023432564,0.004472957],"genre_scores_gemma":[0.13897592,0.00045867718,0.85154516,0.00010643104,0.00002756527,0.00015209513,0.005073884,0.0009939949,0.0026662413],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99869186,0.00024367106,0.0001372782,0.000362485,0.00047399048,0.00009083795],"domain_scores_gemma":[0.98920095,0.0052742157,0.001165705,0.0016744663,0.0023391258,0.00034564236],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014427643,0.0008600943,0.00109819,0.002478474,0.0008007113,0.0012849265,0.0013863128,0.00078304054,0.0058252127],"category_scores_gemma":[0.009898933,0.00058558537,0.00050493795,0.0025322733,0.0005179827,0.0012495582,0.0008926878,0.0008955257,0.0033743638],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0012210567,0.0002042854,0.022050124,0.00085306034,0.00009196947,0.00061428605,0.00070858805,0.014761123,0.29704678,0.0037891562,0.008826308,0.6498332],"study_design_scores_gemma":[0.0002928673,0.0006220561,0.036474694,0.00021728326,0.00017812043,0.0028509384,0.0006137348,0.19193195,0.69379735,0.01579874,0.056996454,0.00022582943],"about_ca_topic_score_codex":0.0023922385,"about_ca_topic_score_gemma":0.0026194558,"teacher_disagreement_score":0.0058252127,"about_ca_system_score_codex":0.0006681755,"about_ca_system_score_gemma":0.0011224215,"threshold_uncertainty_score":0.019487321},"labels":[],"label_agreement":null},{"id":"W2154130239","doi":"10.1093/bioinformatics/btv217","title":"Genome-scale strain designs based on regulatory minimal cut sets","year":2015,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Microbial Metabolic Engineering and Bioproduction","field":"Biochemistry, Genetics and Molecular Biology","cited_by":31,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Constraint (computer-aided design); Metabolic engineering; Computer science; Range (aeronautics); Set (abstract data type); Genome; MATLAB; Flux (metallurgy); Strain (injury); Scale (ratio); Gene knockout; Simple (philosophy); Key (lock); Theoretical computer science; Mathematical optimization; Gene; Mathematics; Biology; Genetics; Physics; Chemistry; Engineering; Programming language","score_opus":0.02035519260935907,"score_gpt":0.2313966490671417,"score_spread":0.2110414564577826,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2154130239","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.06523624,0.0002717408,0.92715824,0.00015556677,0.00003478372,0.0001614553,0.00028474553,0.00064124167,0.006055995],"genre_scores_gemma":[0.46594024,0.00031809916,0.53053004,0.00007992319,0.000012594991,0.0005517796,0.0008362151,0.00020331232,0.001527795],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9996093,0.00011461733,0.000022333155,0.000076531374,0.00013658658,0.000040717045],"domain_scores_gemma":[0.9989298,0.00064563006,0.0001367379,0.000084372834,0.0001564234,0.00004698026],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008987268,0.0009398064,0.0005754744,0.0007451664,0.00028316595,0.00067660236,0.0009633482,0.0006679473,0.003566169],"category_scores_gemma":[0.0032211354,0.00041134568,0.000766936,0.00055205915,0.00062467804,0.00073079544,0.0007743303,0.0008111185,0.00048326317],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007388006,0.000049973492,0.00069810916,0.00018355988,0.000021851714,0.00007340677,0.000029838033,0.9315894,0.014273489,0.022947691,0.0004862306,0.029572465],"study_design_scores_gemma":[0.00003373505,0.000111039546,0.00018964216,0.000027606471,0.000018458735,0.00004039394,0.00001688944,0.96261287,0.011911658,0.02279773,0.00222668,0.000013225181],"about_ca_topic_score_codex":0.0012312984,"about_ca_topic_score_gemma":0.0011818801,"teacher_disagreement_score":0.003566169,"about_ca_system_score_codex":0.00074307487,"about_ca_system_score_gemma":0.00066364516,"threshold_uncertainty_score":0.0119300485},"labels":[],"label_agreement":null},{"id":"W2154164391","doi":"10.1093/bioinformatics/btt610","title":"A pathway-based data integration framework for prediction of disease progression","year":2013,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":86,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"BC Cancer Agency; Engineering and Physical Sciences Research Council; Medical Research Council; University of Bristol; Cancer Research UK","keywords":"Computer science; Machine learning; Feature selection; Artificial intelligence; Classifier (UML); Predictive modelling; Data mining; Data type","score_opus":0.02166951579529282,"score_gpt":0.26652677860525836,"score_spread":0.24485726280996553,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2154164391","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0060459683,0.00060015416,0.9833836,0.00066869846,0.00005058491,0.00012203999,0.0045553874,0.0037674818,0.0008060499],"genre_scores_gemma":[0.24419558,0.0010054935,0.734667,0.00032376757,0.00014099087,0.0006303338,0.015672985,0.00035922407,0.00300458],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99889934,0.00028168058,0.00010069477,0.00033066267,0.0003096885,0.00007789271],"domain_scores_gemma":[0.9982634,0.00072102883,0.00015398099,0.00022938418,0.0004380006,0.00019414576],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0031188442,0.0010132891,0.0012182994,0.0022923083,0.0005700521,0.002095131,0.002290563,0.0011729694,0.004112241],"category_scores_gemma":[0.006429397,0.0005554431,0.0018912454,0.0026608848,0.00066711573,0.0018214224,0.0023620045,0.0019419204,0.001792178],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000742972,0.00040364917,0.019613946,0.000656062,0.0005263282,0.00062554405,0.00039522818,0.63495463,0.004403351,0.059630744,0.018965509,0.25908196],"study_design_scores_gemma":[0.00002736575,0.000043750322,0.00093577604,0.00003283007,0.000031937892,0.000079182246,0.00002451469,0.95760125,0.0005514572,0.035974886,0.004675662,0.000021360896],"about_ca_topic_score_codex":0.015184092,"about_ca_topic_score_gemma":0.01756466,"teacher_disagreement_score":0.015184092,"about_ca_system_score_codex":0.0012216654,"about_ca_system_score_gemma":0.003087927,"threshold_uncertainty_score":0.030191422},"labels":[],"label_agreement":null},{"id":"W2154202919","doi":"10.1093/bioinformatics/bti057","title":"PSORTb v.2.0: Expanded prediction of bacterial protein subcellular localization and insights gained from comparative proteome analysis","year":2004,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Machine Learning in Bioinformatics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":723,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"Simon Fraser University; Michael Smith Health Research BC","keywords":"Proteome; Computer science; n-gram; Subsequence; Support vector machine; Computational biology; MIT License; Subcellular localization; Biology; Artificial intelligence; Software; Bioinformatics; Genetics; Mathematics; Programming language; Language model","score_opus":0.011121309572131761,"score_gpt":0.23112521259350607,"score_spread":0.22000390302137432,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2154202919","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.13809733,0.0023205688,0.66100556,0.000498088,0.00013913243,0.00024699062,0.030898321,0.16167913,0.0051148417],"genre_scores_gemma":[0.19036092,0.0012488678,0.7235301,0.0001860744,0.00006903095,0.0005025039,0.07147302,0.0093948115,0.0032347247],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9997168,0.00006561656,0.000025593,0.00007838855,0.00009099415,0.0000225043],"domain_scores_gemma":[0.9993617,0.00030396294,0.00008923634,0.00008446423,0.00011347698,0.000047101174],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011663679,0.0009815398,0.00094397215,0.0014785441,0.00026019587,0.000754475,0.0009998893,0.0004714882,0.005727987],"category_scores_gemma":[0.0029352058,0.00048320103,0.0006899817,0.0012007745,0.000158103,0.00071301823,0.0007933745,0.00061671674,0.0037326086],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0040745265,0.0003218019,0.015937295,0.0031992353,0.00045987536,0.0010666722,0.0003615284,0.042104993,0.33412695,0.005094295,0.15150896,0.44174385],"study_design_scores_gemma":[0.00057458214,0.0006620919,0.032356385,0.0003989735,0.00032644736,0.0037537813,0.00008648938,0.64309025,0.17815433,0.009226647,0.13116714,0.00020291586],"about_ca_topic_score_codex":0.000865289,"about_ca_topic_score_gemma":0.00091310946,"teacher_disagreement_score":0.005727987,"about_ca_system_score_codex":0.0002498478,"about_ca_system_score_gemma":0.00039185214,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W2154237723","doi":"10.1093/bioinformatics/btl538","title":"PROTERAN: animated terrain evolution for visual analysis of patterns in protein folding trajectory","year":2006,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Protein Structure and Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Concordia University","funders":"","keywords":"Computer science; Visualization; Energy landscape; Protein folding; Folding (DSP implementation); Data mining; Biology","score_opus":0.005096344942455624,"score_gpt":0.24225487652494349,"score_spread":0.23715853158248787,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2154237723","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.015220781,0.00017495504,0.8914594,0.00032853664,0.00007180271,0.00020592242,0.0029656861,0.0874881,0.0020847798],"genre_scores_gemma":[0.13284533,0.00037712432,0.8510694,0.00015631734,0.000030217354,0.0011944958,0.0040765954,0.008300439,0.0019501699],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9997873,0.00006889803,0.000017202066,0.00003444433,0.000069587695,0.000022690536],"domain_scores_gemma":[0.999356,0.00041871224,0.000036692614,0.000058868947,0.00006892625,0.000060752882],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010002914,0.0010990143,0.0007284199,0.0016442961,0.00057852414,0.0012183872,0.0019576997,0.00087500486,0.01643361],"category_scores_gemma":[0.0025018288,0.00059817685,0.0009046661,0.0009346227,0.00043187104,0.0010930253,0.0018296273,0.0015032085,0.0017201655],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0015759218,0.00039579882,0.004863035,0.0018861052,0.00047007855,0.0014700255,0.0027820212,0.20499718,0.06842749,0.046378344,0.14165828,0.5250957],"study_design_scores_gemma":[0.0002523321,0.00007601817,0.0012315895,0.000084703126,0.000028408514,0.00019856537,0.00013922191,0.93741554,0.01173685,0.01449345,0.034278,0.0000653537],"about_ca_topic_score_codex":0.0035822815,"about_ca_topic_score_gemma":0.0036158073,"teacher_disagreement_score":0.01643361,"about_ca_system_score_codex":0.0006325769,"about_ca_system_score_gemma":0.00066316716,"threshold_uncertainty_score":0.054975867},"labels":[],"label_agreement":null},{"id":"W2154348147","doi":"10.1093/bioinformatics/bts721","title":"Fragment recruitment on metabolic pathways: comparative metabolic profiling of metagenomes and metatranscriptomes","year":2013,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Dalhousie University","funders":"Leibniz-Gemeinschaft; Deutsche Forschungsgemeinschaft","keywords":"KEGG; Metagenomics; Profiling (computer programming); Perl; Metabolic pathway; Computational biology; Biology; Java; Computer science; Bioinformatics; Gene; Transcriptome; Genetics; World Wide Web; Programming language","score_opus":0.04910625613312687,"score_gpt":0.26873570459960555,"score_spread":0.2196294484664787,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2154348147","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.32133794,0.0016708652,0.5000729,0.00088176347,0.00019046217,0.00051433075,0.10022758,0.061324645,0.013779569],"genre_scores_gemma":[0.37230292,0.0011671796,0.5272436,0.00040485023,0.00006391443,0.0009560796,0.085855015,0.007457194,0.0045491755],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9993631,0.00010120075,0.000033973734,0.00026979248,0.00017484822,0.000057003053],"domain_scores_gemma":[0.9991574,0.00030434775,0.000109671295,0.00016485639,0.00013346785,0.00013021055],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012460705,0.0008164295,0.0006251786,0.0012944161,0.0005826634,0.0011414117,0.00064920116,0.00037955382,0.007965943],"category_scores_gemma":[0.001959248,0.00042285945,0.0008541047,0.0020748512,0.00032424385,0.0010076171,0.0016236594,0.0009263165,0.0027242105],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0019477529,0.00017037586,0.014527534,0.0016434761,0.00017002768,0.00030864237,0.00050365337,0.004528998,0.81520927,0.0036066314,0.017617526,0.13976608],"study_design_scores_gemma":[0.00018032282,0.00048118635,0.05644336,0.00020492461,0.00017898217,0.0008359146,0.00029830803,0.06187592,0.7929044,0.008936934,0.077507794,0.00015196759],"about_ca_topic_score_codex":0.0012505881,"about_ca_topic_score_gemma":0.001758192,"teacher_disagreement_score":0.007965943,"about_ca_system_score_codex":0.0005254416,"about_ca_system_score_gemma":0.00075878756,"threshold_uncertainty_score":0.02664876},"labels":[],"label_agreement":null},{"id":"W2154408165","doi":"10.1093/bioinformatics/17.6.557","title":"Compositional symmetries in complete genomes","year":2001,"lang":"en","type":"article","venue":"Bioinformatics","topic":"RNA and protein synthesis mechanisms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":69,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Hospital for Sick Children","funders":"","keywords":"Genome; Oligonucleotide; Homogeneous space; Basis (linear algebra); DNA; DNA sequencing; Computer science; Computational biology; Biology; Genetics; Mathematics; Gene","score_opus":0.018670551423125827,"score_gpt":0.23666133734492908,"score_spread":0.21799078592180327,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2154408165","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.16924916,0.004996352,0.7874366,0.0030348017,0.00046208082,0.00008973822,0.002537808,0.0012163282,0.030977193],"genre_scores_gemma":[0.7133509,0.0034439033,0.26980495,0.0006548525,0.0005904601,0.00015579423,0.00480446,0.00042486924,0.006769907],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99942774,0.00015096016,0.000040689236,0.0001509187,0.00017013015,0.00005943324],"domain_scores_gemma":[0.9987313,0.0004968727,0.00016205241,0.00034687493,0.0001780081,0.00008486927],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009685373,0.00032073923,0.0006299381,0.0009748283,0.00089056906,0.001415901,0.000530787,0.00044689988,0.003131775],"category_scores_gemma":[0.004033876,0.00035955597,0.0011952638,0.0011644319,0.0011094861,0.0032201293,0.0014209693,0.0012053709,0.0006271247],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00008213078,0.000018953868,0.0009900719,0.00012507885,0.000034470668,0.00013955763,0.00016065221,0.012506608,0.0032030672,0.91872543,0.0062477635,0.057766266],"study_design_scores_gemma":[0.000013856749,0.0000137968045,0.00036553328,0.000012475078,0.000009427502,0.000098555494,0.000039570285,0.017975627,0.0010398204,0.96723413,0.013182368,0.000014798733],"about_ca_topic_score_codex":0.0013444275,"about_ca_topic_score_gemma":0.0015193629,"teacher_disagreement_score":0.003131775,"about_ca_system_score_codex":0.000542039,"about_ca_system_score_gemma":0.0007452023,"threshold_uncertainty_score":0.010476828},"labels":[],"label_agreement":null},{"id":"W2154518589","doi":"10.1093/bioinformatics/btu181","title":"geiger v2.0: an expanded suite of methods for fitting macroevolutionary models to phylogenetic trees","year":2014,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Evolution and Paleontology Studies","field":"Earth and Planetary Sciences","cited_by":1222,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Natural Sciences and Engineering Research Council of Canada; National Science Foundation","keywords":"Geiger counter; Computer science; Source code; R package; Phylogenetic tree; Suite; Scope (computer science); Code (set theory); Data mining; Data science; Biology; Set (abstract data type); Programming language; Genetics; Physics; Geography","score_opus":0.06591234001703411,"score_gpt":0.3228357091837009,"score_spread":0.2569233691666668,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2154518589","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.005276646,0.0014985069,0.787264,0.00053697353,0.00039074227,0.00034246757,0.039460994,0.16087526,0.0043544853],"genre_scores_gemma":[0.021914905,0.0010682942,0.8013595,0.00056816294,0.00014754034,0.002028944,0.051116824,0.117441095,0.004354846],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99699223,0.001378086,0.00031664103,0.0006325529,0.0004939442,0.00018651574],"domain_scores_gemma":[0.99055755,0.0069885603,0.00054907793,0.0011188554,0.0005877671,0.00019813329],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008890259,0.004873794,0.003678935,0.005338884,0.0017613079,0.0035509686,0.007712415,0.002816166,0.05650549],"category_scores_gemma":[0.030336408,0.0038716064,0.00695231,0.004257701,0.0010835908,0.0033039232,0.005129053,0.0067610648,0.046351768],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000561489,0.00016802667,0.010795519,0.004648806,0.0036692738,0.0012868778,0.0014897935,0.1142331,0.0107161,0.02919447,0.5847337,0.23850282],"study_design_scores_gemma":[0.00045025768,0.0001983881,0.00722422,0.00080876827,0.00087583327,0.0021667718,0.000210216,0.45085868,0.009643701,0.0969108,0.43006384,0.0005885188],"about_ca_topic_score_codex":0.0044344324,"about_ca_topic_score_gemma":0.0073041087,"teacher_disagreement_score":0.05650549,"about_ca_system_score_codex":0.0010367893,"about_ca_system_score_gemma":0.0028776133,"threshold_uncertainty_score":0.1890297},"labels":[],"label_agreement":null},{"id":"W2154722286","doi":"10.1093/bioinformatics/btr556","title":"miREnvironment Database: providing a bridge for microRNAs, environmental factors and phenotypes","year":2011,"lang":"en","type":"article","venue":"Bioinformatics","topic":"MicroRNA in disease regulation","field":"Biochemistry, Genetics and Molecular Biology","cited_by":74,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Research Council Canada","keywords":"Phenotype; microRNA; Biology; Computational biology; Organism; Database; Bioinformatics; Genetics; Computer science; Gene","score_opus":0.023721835275175473,"score_gpt":0.22199157430721708,"score_spread":0.19826973903204162,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2154722286","genre_codex":"dataset","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.011422846,0.005075215,0.14384244,0.0011144059,0.0004082079,0.0007247547,0.695899,0.12984942,0.011663702],"genre_scores_gemma":[0.01945345,0.0018738641,0.1863423,0.00062104914,0.00010089567,0.0012034157,0.7809332,0.006669401,0.0028024276],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99837315,0.00029080047,0.00043160756,0.00046294267,0.00035116263,0.000090354064],"domain_scores_gemma":[0.99749964,0.00096844306,0.0005177864,0.00041496355,0.00031316295,0.00028598262],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0023056832,0.0026915132,0.0027326595,0.007069152,0.0011317676,0.0021378654,0.0027006292,0.0019244243,0.020273406],"category_scores_gemma":[0.0067521115,0.0012055213,0.0016107644,0.006646183,0.00039523953,0.0025734345,0.0036076445,0.0015255953,0.023476232],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0041607916,0.00056776684,0.01563509,0.020488715,0.00096542866,0.0036561228,0.001252139,0.005733408,0.07550127,0.019556807,0.5745215,0.27796102],"study_design_scores_gemma":[0.000440501,0.00018109442,0.012806713,0.0008887119,0.00038944115,0.0015503876,0.00026662197,0.012520154,0.020918231,0.00983215,0.9399897,0.00021632122],"about_ca_topic_score_codex":0.0012113238,"about_ca_topic_score_gemma":0.0017850004,"teacher_disagreement_score":0.020273406,"about_ca_system_score_codex":0.000684203,"about_ca_system_score_gemma":0.0021227368,"threshold_uncertainty_score":0.067821324},"labels":[],"label_agreement":null},{"id":"W2154865814","doi":"10.1093/bioinformatics/btr649","title":"Identifying quantitative trait loci via group-sparse multitask regression and feature selection: an imaging genetics study of the ADNI cohort","year":2011,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":181,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Institute on Aging; University of California, San Diego; National Institute of Biomedical Imaging and Bioengineering; Canadian Institutes of Health Research; University of California, Los Angeles; Genentech; National Institutes of Health; Servier; Eisai; Northern California Institute for Research and Education; Pfizer; Biogen; BioClinica; Alzheimer's Association; Amorfix Life Sciences; Dana Foundation; Bayer HealthCare; Alzheimer's Disease Neuroimaging Initiative; Meso Scale Diagnostics; National Center for Research Resources; F. Hoffmann-La Roche; Medpace; Bristol-Myers Squibb; Eli Lilly and Company; AstraZeneca; Novartis Pharmaceuticals Corporation; Synarc; Foundation for the National Institutes of Health; National Science Foundation","keywords":"Feature selection; Single-nucleotide polymorphism; Imaging genetics; Artificial intelligence; Computer science; Regression; Pairwise comparison; Computational biology; Interpretability; Neuroimaging; Machine learning; Pattern recognition (psychology); Biology; Genetics; Statistics; Mathematics; Genotype; Neuroscience","score_opus":0.03403405047364537,"score_gpt":0.29344337596239134,"score_spread":0.25940932548874596,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2154865814","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98112434,0.0005706086,0.01601112,0.0008770739,0.000033894972,0.00006824606,0.0008241488,0.00013484515,0.00035583356],"genre_scores_gemma":[0.97946334,0.0002500107,0.01771067,0.00015676588,0.000098177865,0.00008548082,0.0015076798,0.00003226232,0.00069565274],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9994252,0.0002833283,0.000022917233,0.0001714415,0.000061639876,0.000035562025],"domain_scores_gemma":[0.99772424,0.0011687219,0.00028183393,0.0004385797,0.00017248384,0.00021408696],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0055467784,0.00084443955,0.0007318088,0.00074977503,0.0007830383,0.00070224266,0.00087287184,0.00067338144,0.0007624788],"category_scores_gemma":[0.005458946,0.00027746058,0.0010234792,0.0007444598,0.0005385328,0.000259494,0.0005220233,0.00097590033,0.00018107521],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0036430568,0.0013512879,0.8367178,0.00020024052,0.0027246648,0.00264348,0.00073890085,0.04069993,0.010693369,0.002359904,0.009651552,0.088575736],"study_design_scores_gemma":[0.00097334414,0.0010342511,0.62134606,0.0000621013,0.0021079162,0.0019671498,0.0005369714,0.35012057,0.0046967138,0.012717357,0.0042498377,0.00018773673],"about_ca_topic_score_codex":0.012862415,"about_ca_topic_score_gemma":0.017339027,"teacher_disagreement_score":0.012862415,"about_ca_system_score_codex":0.00054658426,"about_ca_system_score_gemma":0.0009141716,"threshold_uncertainty_score":0.029334545},"labels":[],"label_agreement":null},{"id":"W2155201678","doi":"10.1093/bioinformatics/btt092","title":"EBARDenovo: highly accurate <i>de novo</i> assembly of RNA-Seq with efficient chimera-detection","year":2013,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":34,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"De Bruijn graph; Sequence assembly; De Bruijn sequence; Computer science; Computational biology; De novo transcriptome assembly; Amplicon; RNA-Seq; Software; RNA; Chimera (genetics); Biology; Graph; Genetics; Gene; Transcriptome; Theoretical computer science; Mathematics; Programming language; Gene expression; Combinatorics; Polymerase chain reaction","score_opus":0.0070750471589631535,"score_gpt":0.20596737948605845,"score_spread":0.1988923323270953,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2155201678","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.019364255,0.0010325727,0.86421895,0.0002691172,0.0002941084,0.0003474859,0.005009386,0.10482877,0.0046353354],"genre_scores_gemma":[0.039190583,0.00042237702,0.92403615,0.0002266233,0.0000367037,0.0006999962,0.010185608,0.021630937,0.0035711094],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.998021,0.00042879785,0.00018853652,0.00071073556,0.000507863,0.00014296095],"domain_scores_gemma":[0.9974733,0.0012121836,0.0002904636,0.0005243729,0.0003646144,0.00013499774],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0050411006,0.0022152385,0.0017569212,0.001355556,0.001863345,0.0029069064,0.0036650698,0.0016403877,0.008073058],"category_scores_gemma":[0.008544173,0.0020662537,0.0017353897,0.0010608429,0.0010208893,0.0018286096,0.0025055765,0.003866793,0.0070274333],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0026047523,0.00033610224,0.0077558164,0.005259264,0.0011013729,0.0009739479,0.0019731647,0.062619045,0.44767463,0.030848542,0.16791888,0.27093452],"study_design_scores_gemma":[0.0003117811,0.00028417868,0.0036188096,0.00039833662,0.00022579737,0.0010851844,0.00014752294,0.37051117,0.38520455,0.016228858,0.22159153,0.00039228625],"about_ca_topic_score_codex":0.0032728594,"about_ca_topic_score_gemma":0.0047115283,"teacher_disagreement_score":0.008073058,"about_ca_system_score_codex":0.001060897,"about_ca_system_score_gemma":0.0018592107,"threshold_uncertainty_score":0.027007103},"labels":[],"label_agreement":null},{"id":"W2155886619","doi":"10.1093/bioinformatics/btv065","title":"Where are we at regarding species translation? A review of the sbv IMPROVER challenge","year":2015,"lang":"en","type":"review","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Translation (biology); Computer science; Biology","score_opus":0.07257934998167527,"score_gpt":0.2985734266777903,"score_spread":0.22599407669611504,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2155886619","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000074684605,0.9934818,0.0004281606,0.0033815356,0.00083867874,0.0000042775896,0.000032242984,0.000021059654,0.0017375923],"genre_scores_gemma":[0.00094379083,0.99425334,0.0008208532,0.0021180585,0.00065343844,0.000008247999,0.00007953501,0.000019926218,0.0011028383],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9991491,0.00026616897,0.0001169432,0.000116089584,0.00027541752,0.00007633545],"domain_scores_gemma":[0.9948908,0.0031757914,0.00036126174,0.00016439242,0.0011410605,0.0002667355],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0049054776,0.0007574416,0.0012903314,0.0031300755,0.0005971791,0.002324468,0.0013711002,0.0019840298,0.008261195],"category_scores_gemma":[0.008648363,0.00038069827,0.00049341563,0.003193436,0.0016017336,0.0050478396,0.0020609442,0.0027188333,0.005295931],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00004591543,0.000023857463,0.00011810859,0.011603694,0.0000408782,0.00009265562,0.00020387201,0.0002407061,0.00050851237,0.014175803,0.0988484,0.8740976],"study_design_scores_gemma":[0.0000035798382,0.000019446128,0.00017521456,0.0050174017,0.000021283566,0.00021733744,0.00007701979,0.000034322315,0.00015627065,0.0033900635,0.9908778,0.000010384798],"about_ca_topic_score_codex":0.0019321498,"about_ca_topic_score_gemma":0.003793302,"teacher_disagreement_score":0.008261195,"about_ca_system_score_codex":0.0013976643,"about_ca_system_score_gemma":0.0029716499,"threshold_uncertainty_score":0.027636468},"labels":[],"label_agreement":null},{"id":"W2156124616","doi":"10.1093/bioinformatics/btm452","title":"Leveraging the structure of the Semantic Web to enhance information retrieval for proteomics","year":2007,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Biomedical Text Mining and Ontologies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Institutes of Health; National Institute of General Medical Sciences; McGill University","keywords":"Computer science; Information retrieval; Graph; Leverage (statistics); SPARQL; RDF; Subgraph isomorphism problem; Semantic Web; Artificial intelligence; Theoretical computer science","score_opus":0.009301903491705083,"score_gpt":0.26632088811460186,"score_spread":0.2570189846228968,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2156124616","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.15303357,0.005477612,0.809536,0.005016966,0.00027485302,0.000940699,0.0034504174,0.012366765,0.009903121],"genre_scores_gemma":[0.2867178,0.002414681,0.7035959,0.0006377786,0.0002010319,0.00025868797,0.0044201445,0.00045169843,0.0013023445],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.997387,0.0009877756,0.00029297415,0.00029260528,0.00092268497,0.0001169953],"domain_scores_gemma":[0.9920372,0.004657482,0.00056945597,0.0011660288,0.0014364081,0.0001334738],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0037535478,0.0008832243,0.00084979774,0.009231799,0.00074400136,0.0025207659,0.0008015461,0.0011279554,0.001760841],"category_scores_gemma":[0.018169725,0.00042673864,0.0010824885,0.007927314,0.0008529571,0.007457106,0.0023140989,0.00090730883,0.0018901817],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005471095,0.0006863999,0.010222214,0.0021760177,0.00025719285,0.00036177252,0.0007189561,0.04106189,0.07408066,0.028442556,0.017180905,0.8242642],"study_design_scores_gemma":[0.00019606792,0.0006516267,0.008238292,0.00032640548,0.00034590272,0.0010978717,0.00072240364,0.6542586,0.10126102,0.17737913,0.05531592,0.00020670658],"about_ca_topic_score_codex":0.0032391238,"about_ca_topic_score_gemma":0.0058002654,"teacher_disagreement_score":0.009231799,"about_ca_system_score_codex":0.0010083315,"about_ca_system_score_gemma":0.0014817577,"threshold_uncertainty_score":0.01985085},"labels":[],"label_agreement":null},{"id":"W2156216305","doi":"10.1093/bioinformatics/bts109","title":"Mirroring co-evolving trees in the light of their topologies","year":2012,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Network topology; Heuristics; Similarity (geometry); Computer science; Mirroring; Domain (mathematical analysis); Tree (set theory); Phylogenetic tree; Maximization; Distance matrix; Enhanced Data Rates for GSM Evolution; Theoretical computer science; Algorithm; Mathematics; Mathematical optimization; Artificial intelligence; Combinatorics; Biology","score_opus":0.019062205201531856,"score_gpt":0.25261054783164744,"score_spread":0.2335483426301156,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2156216305","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.26135176,0.0003706597,0.7322847,0.00044340416,0.00003484934,0.00015104018,0.00030022263,0.00238293,0.002680331],"genre_scores_gemma":[0.5483147,0.00017618033,0.44902,0.00008231631,0.00003143992,0.000103984465,0.00086742855,0.0003034089,0.0011005852],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99909186,0.00026292753,0.00006725128,0.00024906156,0.0002686731,0.000060159546],"domain_scores_gemma":[0.9938951,0.003345318,0.0007214924,0.0009817159,0.0008199853,0.00023641826],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0018868076,0.000584619,0.00092744373,0.0025791114,0.00078126707,0.0016465725,0.0017160533,0.0012450585,0.002099744],"category_scores_gemma":[0.016077502,0.00051521685,0.00068637816,0.0019415129,0.0006597118,0.0025792525,0.0014362241,0.0008551801,0.0007589402],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007171318,0.00030052135,0.03789325,0.0004528438,0.0002469123,0.00095566304,0.0011117674,0.35488692,0.041239478,0.046529617,0.005849048,0.5098168],"study_design_scores_gemma":[0.000022243637,0.00004985383,0.0013876186,0.00001942587,0.00002562142,0.0003331196,0.00009642736,0.9693652,0.005906472,0.021798143,0.000983635,0.000012198357],"about_ca_topic_score_codex":0.0016371966,"about_ca_topic_score_gemma":0.0036909464,"teacher_disagreement_score":0.0025791114,"about_ca_system_score_codex":0.000755086,"about_ca_system_score_gemma":0.0009188863,"threshold_uncertainty_score":0.009978533},"labels":[],"label_agreement":null},{"id":"W2156553312","doi":"10.1093/bioinformatics/btp028","title":"Gene network reconstruction from transcriptional dynamics under kinetic model uncertainty: a case for the second derivative","year":2009,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene Regulatory Network Analysis","field":"Biochemistry, Genetics and Molecular Biology","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Ottawa","funders":"","keywords":"Computer science; ENCODE; Bayesian probability; Expression (computer science); Bayesian network; Gene; Gene regulatory network; Representation (politics); Transcription (linguistics); Computational biology; Gene expression; Theoretical computer science; Data mining; Algorithm; Biology; Artificial intelligence; Genetics","score_opus":0.013357795792574708,"score_gpt":0.2265834792411285,"score_spread":0.2132256834485538,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2156553312","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.013227246,0.00022171345,0.98458916,0.0007364367,0.000021401413,0.000015892292,0.00010275118,0.00017006397,0.0009153464],"genre_scores_gemma":[0.6172301,0.0009961425,0.37495193,0.00062752207,0.000119881595,0.0001922916,0.0008295767,0.00047066066,0.004581875],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99811065,0.000894828,0.000079886544,0.00045665065,0.0003305094,0.00012753333],"domain_scores_gemma":[0.9845977,0.012359736,0.0007134925,0.0014828215,0.0005519332,0.00029428393],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008335715,0.00066811085,0.0013660489,0.0013153526,0.00061567314,0.0021058926,0.0017979025,0.0016709237,0.0025977956],"category_scores_gemma":[0.038305063,0.00070571015,0.0014397174,0.0013688423,0.002417965,0.00342854,0.0022809762,0.0029064007,0.000486162],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00018177422,0.000024034538,0.003097578,0.00012645716,0.00006258765,0.00029778774,0.0002073742,0.75146675,0.0017914091,0.20861003,0.0013124222,0.032821808],"study_design_scores_gemma":[0.00000957632,0.000008774343,0.0002714994,0.00001466461,0.000008500201,0.00005958342,0.000015405209,0.8618715,0.0005855836,0.13639322,0.00074886146,0.000012717646],"about_ca_topic_score_codex":0.0052594873,"about_ca_topic_score_gemma":0.0030294647,"teacher_disagreement_score":0.008335715,"about_ca_system_score_codex":0.0018751652,"about_ca_system_score_gemma":0.0014072345,"threshold_uncertainty_score":0.044083953},"labels":[],"label_agreement":null},{"id":"W2156733952","doi":"10.1093/bioinformatics/btn444","title":"Seeder: discriminative seeding DNA motif discovery","year":2008,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":48,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"McGill University","funders":"Natural Sciences and Engineering Research Council of Canada; Centre Sève; Fonds Québécois de la Recherche sur la Nature et les Technologies; Norges Teknisk-Naturvitenskapelige Universitet","keywords":"Discriminative model; Computational biology; Computer science; DNA binding site; Promoter; Benchmark (surveying); Transcription factor; Biology; Data mining; Artificial intelligence; Genetics; Gene; Gene expression","score_opus":0.012101206424195807,"score_gpt":0.2174705794309284,"score_spread":0.20536937300673258,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2156733952","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.033218186,0.00062800857,0.93622345,0.00016020198,0.000082265244,0.00019845512,0.002570164,0.025191844,0.0017274233],"genre_scores_gemma":[0.18216132,0.00021936087,0.8058501,0.00018193772,0.00005459865,0.00033993536,0.0065479255,0.0010284081,0.0036163975],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99879456,0.00032507305,0.00005809382,0.0003873252,0.00036832195,0.00006669802],"domain_scores_gemma":[0.99804294,0.001140713,0.0001571932,0.00028330635,0.00026525912,0.00011058815],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001816044,0.0011093139,0.001239463,0.0016643568,0.00050026766,0.00079495535,0.002207084,0.0010342603,0.007951148],"category_scores_gemma":[0.0061199646,0.00050782046,0.0006313461,0.0017614185,0.0006088185,0.0009903144,0.001241688,0.0010922676,0.0035707564],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0025670533,0.0004158858,0.008619151,0.0014551977,0.00027388652,0.00072968967,0.00018206553,0.121928476,0.085111625,0.022046652,0.0484628,0.70820755],"study_design_scores_gemma":[0.0002777715,0.00021465111,0.0010296387,0.000037854617,0.000036169182,0.00053229363,0.000025895337,0.92483073,0.04695476,0.015832944,0.010188959,0.000038206057],"about_ca_topic_score_codex":0.0011637377,"about_ca_topic_score_gemma":0.0020919752,"teacher_disagreement_score":0.007951148,"about_ca_system_score_codex":0.00044966853,"about_ca_system_score_gemma":0.0010757365,"threshold_uncertainty_score":0.026599228},"labels":[],"label_agreement":null},{"id":"W2156847802","doi":"10.1093/bioinformatics/btl094","title":"Combining multi-species genomic data for microRNA identification using a Naïve Bayes classifier","year":2006,"lang":"en","type":"article","venue":"Bioinformatics","topic":"MicroRNA in disease regulation","field":"Biochemistry, Genetics and Molecular Biology","cited_by":181,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Queen's University","funders":"National Cancer Institute; National Center for Research Resources; Natural Sciences and Engineering Research Council of Canada; Pennsylvania Department of Health; National Science Foundation","keywords":"False positive paradox; Bayes' theorem; Naive Bayes classifier; Classifier (UML); Computer science; Artificial intelligence; Gene prediction; Genome; Computational biology; Machine learning; Data mining; Gene; Biology; Support vector machine; Bayesian probability; Genetics","score_opus":0.05732768182420857,"score_gpt":0.29172140772989985,"score_spread":0.2343937259056913,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2156847802","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.06947459,0.00045995894,0.9263757,0.00027421725,0.00008243205,0.00026297377,0.0003585136,0.0017156224,0.0009960064],"genre_scores_gemma":[0.28189167,0.0001777614,0.71623206,0.00019950565,0.00007812254,0.00019486886,0.00075572263,0.000055056567,0.000415258],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9961908,0.0015045337,0.00046580462,0.00068226754,0.0010196405,0.00013701318],"domain_scores_gemma":[0.9889599,0.008050322,0.000625285,0.0005621378,0.0016619234,0.00014035591],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0069842697,0.0011014258,0.0014986293,0.0031580164,0.0006901448,0.0013664627,0.0012644726,0.0015974424,0.0014876348],"category_scores_gemma":[0.018703504,0.00048592762,0.0011196203,0.0016243382,0.0004599421,0.0018068573,0.00066376216,0.001016582,0.0014255084],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013493433,0.000831361,0.055836648,0.00072317675,0.0005555422,0.0009223924,0.00031119835,0.10937973,0.041459437,0.0029760941,0.0035280064,0.7821271],"study_design_scores_gemma":[0.000053476528,0.00020819453,0.004463321,0.00007419122,0.00013372848,0.0006702913,0.00010557649,0.9655777,0.019909987,0.007179095,0.001562682,0.000061776394],"about_ca_topic_score_codex":0.0020457571,"about_ca_topic_score_gemma":0.0020933135,"teacher_disagreement_score":0.0069842697,"about_ca_system_score_codex":0.00059896446,"about_ca_system_score_gemma":0.0007998252,"threshold_uncertainty_score":0.03693676},"labels":[],"label_agreement":null},{"id":"W2156923845","doi":"10.1093/bioinformatics/btn146","title":"Guided genome halving: hardness, heuristics and the history of the Hemiascomycetes","year":2008,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genome Rearrangement Algorithms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":41,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Ottawa","funders":"Canada Research Chairs","keywords":"Genome; Ancestor; Heuristics; Computer science; Event (particle physics); Genome evolution; Heuristic; Biology; Evolutionary biology; Computational biology; Genetics; Gene; Artificial intelligence; Physics","score_opus":0.018734386632463865,"score_gpt":0.20116576835242958,"score_spread":0.1824313817199657,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2156923845","genre_codex":"empirical","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.66807586,0.0013369756,0.31648028,0.0010395761,0.00004951643,0.00010838718,0.0010211143,0.00209208,0.009796258],"genre_scores_gemma":[0.67033094,0.00062451855,0.32319707,0.00025726936,0.000019898831,0.00013438753,0.002560941,0.0006241551,0.002250837],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9996761,0.00011348248,0.0000198769,0.00008352709,0.00006451395,0.000042511234],"domain_scores_gemma":[0.9971343,0.0020709212,0.00022016144,0.00026786755,0.00018690203,0.00011987962],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014091992,0.00053870986,0.0006737949,0.00083922956,0.001046429,0.0018193403,0.0014414438,0.0010328597,0.0036225356],"category_scores_gemma":[0.0075305407,0.00040251188,0.0007847692,0.0009843165,0.00095159095,0.0016816437,0.0009784078,0.0012913309,0.00053665735],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006123274,0.00016098535,0.02109708,0.00044472422,0.00008142998,0.0003207112,0.0006071667,0.808757,0.01644,0.04187852,0.0044955155,0.10510445],"study_design_scores_gemma":[0.00008026784,0.00009622087,0.005358957,0.00008604841,0.000056712364,0.0002258009,0.00036534196,0.9121782,0.01496639,0.062282566,0.0042606234,0.000042815842],"about_ca_topic_score_codex":0.0036137823,"about_ca_topic_score_gemma":0.004994892,"teacher_disagreement_score":0.0036225356,"about_ca_system_score_codex":0.0014148826,"about_ca_system_score_gemma":0.0008912312,"threshold_uncertainty_score":0.012118578},"labels":[],"label_agreement":null},{"id":"W2157460239","doi":"10.1093/bioinformatics/btt178","title":"Assembling the 20 Gb white spruce (<i>Picea glauca</i>) genome from whole-genome shotgun sequencing data","year":2013,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":393,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ministry of Forests; Government of British Columbia; Simon Fraser University; University of British Columbia","funders":"BC Cancer Agency; Canada's Michael Smith Genome Sciences Centre","keywords":"Shotgun sequencing; Sequence assembly; Genome; Shotgun; Hybrid genome assembly; Biology; Computational biology; Whole genome sequencing; Illumina dye sequencing; Genomics; DNA sequencing; Reference genome; Genome project; Genetics; Gene; Transcriptome","score_opus":0.03217634457678138,"score_gpt":0.2396316445186623,"score_spread":0.20745529994188094,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2157460239","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.5811459,0.0034640068,0.25072193,0.0011276123,0.00059581496,0.00080320926,0.14052188,0.011543912,0.010075765],"genre_scores_gemma":[0.25555786,0.0025404645,0.35591164,0.0003065867,0.00014880602,0.0003877671,0.3781309,0.0013985466,0.0056174262],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99968123,0.00003770834,0.000036749985,0.00011652642,0.000083307146,0.000044596334],"domain_scores_gemma":[0.99912506,0.00020254626,0.000158117,0.00013036601,0.00024880064,0.00013508725],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008666149,0.00083596486,0.00061616587,0.001517992,0.00065130234,0.0010397389,0.0007928949,0.0005862614,0.0036965057],"category_scores_gemma":[0.0019742365,0.0003807701,0.0009790224,0.0020335403,0.00029290305,0.00071195525,0.00065415865,0.0008298098,0.0032846997],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00092405867,0.0001953109,0.015906833,0.0014420708,0.00019248747,0.000866849,0.00061404286,0.0031163082,0.88518876,0.0013647687,0.011001316,0.07918723],"study_design_scores_gemma":[0.00029160283,0.0012371069,0.23251933,0.00072318944,0.0011079491,0.0026666352,0.0011380892,0.032395218,0.53252566,0.0054502827,0.18969949,0.00024540367],"about_ca_topic_score_codex":0.005677532,"about_ca_topic_score_gemma":0.007789088,"teacher_disagreement_score":0.005677532,"about_ca_system_score_codex":0.00055271806,"about_ca_system_score_gemma":0.0010710873,"threshold_uncertainty_score":0.012366056},"labels":[],"label_agreement":null},{"id":"W2157673856","doi":"10.1093/bioinformatics/btu677","title":"flowDensity: reproducing manual gating of flow cytometry data by automated density-based cell population identification","year":2014,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Single-cell and spatial transcriptomics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":139,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia; BC Cancer Agency","funders":"National Institute of Biomedical Imaging and Bioengineering; National Institute of Allergy and Infectious Diseases; National Institutes of Health","keywords":"Bioconductor; Computer science; Bivariate analysis; Gating; Source code; Data set; Identification (biology); Population; Data mining; Set (abstract data type); Code (set theory); Artificial intelligence; Programming language; Machine learning; Biology","score_opus":0.014861539705209686,"score_gpt":0.2517141235168281,"score_spread":0.23685258381161842,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2157673856","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0049227956,0.00024242059,0.8841349,0.00032694227,0.00039600217,0.00049336266,0.015729917,0.09064436,0.003109373],"genre_scores_gemma":[0.047633402,0.00048894284,0.8644285,0.0005400807,0.0002567928,0.0035832243,0.038742226,0.037775066,0.006551795],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9959751,0.0007680163,0.00038148428,0.0011791842,0.0013607685,0.00033549702],"domain_scores_gemma":[0.99468094,0.002013616,0.00030783622,0.0017886853,0.00096690946,0.00024201938],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0057218904,0.0022669642,0.0019776262,0.003252479,0.0010907068,0.004174401,0.00471599,0.001686294,0.0389518],"category_scores_gemma":[0.016929122,0.0018955013,0.0018766669,0.0023760928,0.0010469331,0.0025086948,0.0035176103,0.0035205889,0.030974213],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013008659,0.00029760879,0.008573847,0.0024948453,0.00063427386,0.00044599044,0.0007862479,0.015239437,0.14370091,0.030677091,0.44813186,0.34771705],"study_design_scores_gemma":[0.0006093495,0.00025784445,0.010033393,0.0003331512,0.00023120479,0.0012528457,0.00012743816,0.30603966,0.26984718,0.050217826,0.36040607,0.00064396753],"about_ca_topic_score_codex":0.002712665,"about_ca_topic_score_gemma":0.0032437195,"teacher_disagreement_score":0.0389518,"about_ca_system_score_codex":0.0013987969,"about_ca_system_score_gemma":0.0024561712,"threshold_uncertainty_score":0.13030672},"labels":[],"label_agreement":null},{"id":"W2157748357","doi":"10.1093/bioinformatics/bti054","title":"Circular genome visualization and exploration using CGView","year":2004,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1122,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Genome Prairie; Genome Canada","keywords":"Scalable Vector Graphics; Visualization; Computer science; Java applet; Genome; Graphics; Documentation; Sample (material); World Wide Web; XML; Computer graphics; Computer graphics (images); Java; Information retrieval; Data mining; Biology; Programming language; Genetics","score_opus":0.029482022421761275,"score_gpt":0.26261835932282573,"score_spread":0.23313633690106444,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2157748357","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.012125899,0.0009200104,0.64643013,0.0012701206,0.0004288826,0.0002840069,0.050571296,0.25382,0.03414961],"genre_scores_gemma":[0.11089414,0.0018019772,0.7065318,0.0006227009,0.00021428127,0.0010671002,0.10049085,0.051879853,0.026497198],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9996697,0.000051241925,0.000020938689,0.00008330706,0.00012992334,0.000044941473],"domain_scores_gemma":[0.99920565,0.0003098928,0.000052702657,0.0001232069,0.00021099205,0.000097706004],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008895539,0.0009208921,0.0006209763,0.0024699066,0.0005700934,0.001468642,0.001489706,0.00067739794,0.08298483],"category_scores_gemma":[0.0021543948,0.00049203157,0.0007810002,0.002307437,0.00027177422,0.0011866784,0.0014695149,0.0013838831,0.018251935],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010547825,0.00015348167,0.0028589969,0.0016967896,0.00015801849,0.000624767,0.0007361923,0.0055718794,0.09714168,0.014732031,0.47287437,0.40239698],"study_design_scores_gemma":[0.0003446296,0.00011513746,0.0068381317,0.0004331246,0.00009498148,0.0012571409,0.0002896643,0.08705574,0.1275545,0.015469414,0.76030284,0.00024463044],"about_ca_topic_score_codex":0.002530193,"about_ca_topic_score_gemma":0.0025006118,"teacher_disagreement_score":0.08298483,"about_ca_system_score_codex":0.0004298769,"about_ca_system_score_gemma":0.00084467453,"threshold_uncertainty_score":0.2776119},"labels":[],"label_agreement":null},{"id":"W2158026707","doi":"10.1093/bioinformatics/btt348","title":"PING 2.0: an R/Bioconductor package for nucleosome positioning using next-generation sequencing data","year":2013,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":19,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Montreal Clinical Research Institute","funders":"National Human Genome Research Institute","keywords":"Bioconductor; R package; Computer science; DNA sequencing; Nucleosome; Ping (video games); Software; Software package; Computational biology; Data mining; Biology; Computational science; Programming language; Genetics; Computer network; DNA; Histone","score_opus":0.10728624577053644,"score_gpt":0.2843603002387506,"score_spread":0.17707405446821414,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2158026707","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.004235097,0.0016684744,0.2961167,0.0010479214,0.0009835177,0.00053229986,0.22945033,0.4596169,0.0063486667],"genre_scores_gemma":[0.03298266,0.0015617787,0.45965287,0.0016154532,0.00032311294,0.0051937248,0.27110147,0.2159878,0.0115811],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99634546,0.0011081246,0.0003345648,0.0010589459,0.0008353131,0.00031752203],"domain_scores_gemma":[0.9923807,0.0036598851,0.0006915625,0.0013378229,0.0014314265,0.00049858546],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007469441,0.004323824,0.0046368013,0.0046207197,0.0014232716,0.0028281733,0.008085232,0.0018241215,0.08253074],"category_scores_gemma":[0.022108821,0.0027972683,0.0030745424,0.0050156927,0.0011619146,0.0025329576,0.0040957998,0.0044613243,0.090478584],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011481281,0.000101003956,0.004229354,0.0062483903,0.0018916484,0.000510167,0.00049732625,0.0057299174,0.007172965,0.010094238,0.9190362,0.043340687],"study_design_scores_gemma":[0.0011305974,0.00029914922,0.014053153,0.0011502174,0.0016394943,0.0011601605,0.00020750929,0.09122774,0.025259895,0.059563022,0.80370337,0.0006057787],"about_ca_topic_score_codex":0.0052474192,"about_ca_topic_score_gemma":0.0053128502,"teacher_disagreement_score":0.08253074,"about_ca_system_score_codex":0.0012185809,"about_ca_system_score_gemma":0.0048091225,"threshold_uncertainty_score":0.27609283},"labels":[],"label_agreement":null},{"id":"W2158376448","doi":"10.1093/bioinformatics/btp045","title":"QMSim: a large-scale genome simulator for livestock","year":2009,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic and phenotypic traits in livestock","field":"Biochemistry, Genetics and Molecular Biology","cited_by":399,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Guelph","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Executable; Computer science; Population; Scale (ratio); Pedigree chart; Programming language; Biology; Genetics; Cartography; Geography","score_opus":0.010195448025228832,"score_gpt":0.2439689393708104,"score_spread":0.23377349134558156,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2158376448","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.09020126,0.00054361025,0.7987743,0.00088827836,0.0003382442,0.00043647745,0.018095488,0.06377664,0.026945699],"genre_scores_gemma":[0.5454764,0.0007506418,0.4106079,0.0004990269,0.00008802974,0.0018996926,0.019964496,0.009189866,0.011523978],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9997515,0.00008885522,0.000016788996,0.000031628955,0.00007691523,0.00003427232],"domain_scores_gemma":[0.99890614,0.00062686077,0.00005560767,0.000092848284,0.00020057143,0.00011800716],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010578749,0.00074057374,0.0006735374,0.000447305,0.0005473911,0.0007405279,0.0024156785,0.0010053733,0.01768017],"category_scores_gemma":[0.0028076714,0.0005641026,0.0007690665,0.0007894905,0.0003548282,0.00093833014,0.0008894645,0.0013261024,0.0024479786],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00032085594,0.00011318222,0.003934341,0.00029105062,0.00014037639,0.00022541755,0.00020469456,0.92755723,0.004483797,0.012769435,0.029089011,0.020870559],"study_design_scores_gemma":[0.00007776424,0.000021214679,0.00023337874,0.000010179837,0.0000121603025,0.000020909154,0.000013777798,0.98778373,0.0013108444,0.0021904362,0.008314315,0.000011222098],"about_ca_topic_score_codex":0.0076450966,"about_ca_topic_score_gemma":0.0043930896,"teacher_disagreement_score":0.01768017,"about_ca_system_score_codex":0.0007502232,"about_ca_system_score_gemma":0.0012972307,"threshold_uncertainty_score":0.059146047},"labels":[],"label_agreement":null},{"id":"W2158404603","doi":"10.1093/bioinformatics/bts082","title":"Early immunologic correlates of HIV protection can be identified from computational analysis of complex multivariate T-cell flow cytometry assays","year":2012,"lang":"en","type":"article","venue":"Bioinformatics","topic":"HIV Research and Treatment","field":"Immunology and Microbiology","cited_by":72,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Michael Smith Health Research BC; University of British Columbia; BC Cancer Agency","funders":"National Cancer Institute; Natural Sciences and Engineering Research Council of Canada; Walter Reed National Military Medical Center; National Institute of Biomedical Imaging and Bioengineering; Michael Smith Health Research BC; U.S. Public Health Service; Compute Canada; Canadian Institutes of Health Research; Uniformed Services University of the Health Sciences; Vaccine Research Center; National Institute of Allergy and Infectious Diseases; Terry Fox Research Institute; Western Canada Research Grid; National Institutes of Health; U.S. Department of Health and Human Services","keywords":"Bioconductor; Multivariate statistics; Multivariate analysis; Flow cytometry; Identification (biology); Computer science; Computational biology; Documentation; Biology; Data mining; Immunology; Machine learning; Genetics","score_opus":0.035279313773383635,"score_gpt":0.2667180177372091,"score_spread":0.23143870396382546,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2158404603","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.3851899,0.0004249909,0.5930283,0.00101837,0.00005320085,0.0002382637,0.009714909,0.0078197615,0.0025123162],"genre_scores_gemma":[0.7245801,0.0002666487,0.263091,0.00027457104,0.000053686897,0.00038405324,0.00994926,0.000547481,0.00085324666],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99973184,0.0000949678,0.000019906312,0.00007302937,0.000054312026,0.000025894675],"domain_scores_gemma":[0.9975546,0.0019038378,0.00017907053,0.00015256913,0.00012007241,0.00008982464],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0017440552,0.0008565554,0.0007147688,0.0010583684,0.00029583665,0.0009079942,0.0008597072,0.00046623545,0.0031607763],"category_scores_gemma":[0.0060115797,0.00023968573,0.0008467707,0.0006545865,0.00042683157,0.00043099932,0.0006500915,0.0008437865,0.0006601882],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013557323,0.00041641996,0.15585567,0.0007634159,0.0003871673,0.00055618316,0.00031105257,0.62306446,0.03033668,0.008579097,0.016704913,0.1616692],"study_design_scores_gemma":[0.000044587185,0.00005424032,0.015429234,0.000028492557,0.00004335817,0.00013815014,0.000034548637,0.9713425,0.00482619,0.0060927444,0.0019445891,0.000021442573],"about_ca_topic_score_codex":0.0029821156,"about_ca_topic_score_gemma":0.00439706,"teacher_disagreement_score":0.0031607763,"about_ca_system_score_codex":0.00053175463,"about_ca_system_score_gemma":0.0011885014,"threshold_uncertainty_score":0.010573864},"labels":[],"label_agreement":null},{"id":"W2158576281","doi":"10.1093/bioinformatics/bti829","title":"Quality classification of tandem mass spectrometry data","year":2005,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Advanced Proteomics Techniques and Applications","field":"Chemistry","cited_by":44,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Canadian Institute for Theoretical Astrophysics","keywords":"Mascot; Identification (biology); Computer science; Bottleneck; Tandem mass spectrometry; Process (computing); Data mining; Replicate; Quality (philosophy); Mass spectrometry; Artificial intelligence; Information retrieval; Chemistry; Chromatography; Mathematics; Biology; Statistics","score_opus":0.0656602963842446,"score_gpt":0.34696959766278473,"score_spread":0.28130930127854015,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2158576281","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.62660456,0.0032370947,0.35972095,0.0011430309,0.00034877885,0.0005353017,0.003214226,0.003037514,0.0021584893],"genre_scores_gemma":[0.85148424,0.0006328537,0.13876398,0.0002525926,0.00026022532,0.00019534823,0.007331034,0.000291231,0.0007885292],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9929162,0.0010676746,0.0010872738,0.00082397275,0.0037953749,0.00030951976],"domain_scores_gemma":[0.96600217,0.012875679,0.0058316323,0.0037726646,0.01088864,0.0006292751],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0073161125,0.0006948241,0.0010470672,0.00489853,0.0006874933,0.0020285181,0.0014917672,0.0016148138,0.0016947405],"category_scores_gemma":[0.025059871,0.00016598022,0.0007579185,0.0030484174,0.0010478497,0.0013855961,0.00089845713,0.0007279536,0.00074731297],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0050544683,0.0005793719,0.17830558,0.0019476891,0.00054643466,0.0012609734,0.0004955517,0.018671896,0.15540253,0.0024398575,0.007298581,0.62799704],"study_design_scores_gemma":[0.00036080665,0.0016864036,0.25689265,0.00042834182,0.00073582155,0.004648505,0.00071224815,0.42047885,0.289582,0.0106266355,0.013618544,0.00022915899],"about_ca_topic_score_codex":0.00071001064,"about_ca_topic_score_gemma":0.0006259498,"teacher_disagreement_score":0.0073161125,"about_ca_system_score_codex":0.00086823746,"about_ca_system_score_gemma":0.0006127371,"threshold_uncertainty_score":0.03869176},"labels":[],"label_agreement":null},{"id":"W2158957954","doi":"10.1093/bioinformatics/bts395","title":"Side-chain rotamer changes upon ligand binding: common, crucial, correlate with entropy and rearrange hydrogen bonding","year":2012,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Protein Structure and Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":83,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Sherbrooke","funders":"","keywords":"Hydrogen bond; Conformational isomerism; Chemistry; Conformational entropy; Ligand (biochemistry); Entropy (arrow of time); Computational chemistry; Thermodynamics; Biochemistry; Physics; Molecule; Organic chemistry","score_opus":0.008453271641621773,"score_gpt":0.2213110976668868,"score_spread":0.21285782602526504,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2158957954","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9689426,0.0041255443,0.005099143,0.00012342856,0.000017961893,0.000024686795,0.019331165,0.00063508016,0.0017003206],"genre_scores_gemma":[0.94057935,0.0010267849,0.005900309,0.00010781613,0.00002845651,0.000049859213,0.05172118,0.00012081054,0.0004653794],"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9990631,0.00013285232,0.00007927106,0.00040473332,0.00020665182,0.00011343312],"domain_scores_gemma":[0.99841726,0.000388822,0.0006087167,0.0002691339,0.00016746127,0.00014860347],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006518485,0.00073547685,0.0010973185,0.00156269,0.00046648594,0.00069332455,0.0005608107,0.00055074785,0.001753868],"category_scores_gemma":[0.0022011956,0.00021874435,0.0007542806,0.0016105145,0.00053655147,0.0005650435,0.00077806,0.000566809,0.0012538735],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0023261874,0.00042684365,0.6648766,0.003907275,0.0012475919,0.0014132094,0.00037537285,0.023594143,0.22193477,0.0007833699,0.015460529,0.063654155],"study_design_scores_gemma":[0.00008010157,0.00035809272,0.8928958,0.00014773679,0.0004414288,0.0031305978,0.00027523903,0.046347864,0.034805343,0.0024633757,0.018961018,0.00009328517],"about_ca_topic_score_codex":0.0015813878,"about_ca_topic_score_gemma":0.0024595265,"teacher_disagreement_score":0.001753868,"about_ca_system_score_codex":0.00038658248,"about_ca_system_score_gemma":0.00033785184,"threshold_uncertainty_score":0.005867243},"labels":[],"label_agreement":null},{"id":"W2159227212","doi":"10.1093/bioinformatics/16.10.899","title":"USAGE: a web-based approach towards the analysis of SAGE data","year":2000,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":45,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Pediatric Oncology Group","funders":"","keywords":"Computer science; Flexibility (engineering); SAGE; Relational database; The Internet; Web application; Software; Set (abstract data type); Database; Data sharing; World Wide Web; Data science; Information retrieval; Data mining","score_opus":0.0382471465092211,"score_gpt":0.2795155096130832,"score_spread":0.2412683631038621,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2159227212","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0027468938,0.0006164479,0.85260445,0.0008701898,0.00030636563,0.0012975822,0.00838505,0.12244706,0.010725921],"genre_scores_gemma":[0.014493536,0.0014005593,0.9176063,0.00086501247,0.00045288308,0.0025993225,0.025960414,0.0257808,0.010841201],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99226344,0.0027507725,0.0010986815,0.00094520795,0.0025896856,0.00035223537],"domain_scores_gemma":[0.9817368,0.009437546,0.0008869213,0.0040226486,0.0026719463,0.0012441758],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008893886,0.0026817573,0.0023211355,0.00801875,0.0015638951,0.006251764,0.0041550146,0.0019197928,0.030452358],"category_scores_gemma":[0.019458985,0.0019153339,0.002832973,0.0083042765,0.0015821966,0.0054881633,0.0053430684,0.0041680885,0.027916627],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011691273,0.00058048184,0.004562718,0.003625473,0.0005602026,0.002582131,0.00217598,0.003252265,0.051881056,0.03735146,0.2745323,0.61772686],"study_design_scores_gemma":[0.00029642353,0.0005699946,0.008997697,0.0012538938,0.00031150997,0.005700373,0.00066618726,0.045278132,0.034332108,0.11487778,0.7871777,0.00053814496],"about_ca_topic_score_codex":0.0012077654,"about_ca_topic_score_gemma":0.0012023945,"teacher_disagreement_score":0.030452358,"about_ca_system_score_codex":0.0005191178,"about_ca_system_score_gemma":0.0020346988,"threshold_uncertainty_score":0.10187334},"labels":[],"label_agreement":null},{"id":"W2159531362","doi":"10.1093/bioinformatics/bts658","title":"A system for exact and approximate genetic linkage analysis of SNP data in large pedigrees","year":2012,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":51,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Ottawa","funders":"National Institute of General Medical Sciences; National Human Genome Research Institute","keywords":"Tag SNP; Computer science; SNP; Markov chain Monte Carlo; Workflow; Linkage (software); Data mining; Pedigree chart; Theoretical computer science; Haplotype; Single-nucleotide polymorphism; Bayesian probability; Genetics; Artificial intelligence; Database; Biology","score_opus":0.031441021981738086,"score_gpt":0.2962077704679798,"score_spread":0.2647667484862417,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2159531362","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0051385267,0.00021668136,0.6752806,0.0003316235,0.00015947034,0.00031353137,0.008890882,0.3037281,0.0059406236],"genre_scores_gemma":[0.10562056,0.0005402713,0.80172783,0.0010820444,0.00028653652,0.0018703996,0.042413983,0.028458443,0.018000027],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9975599,0.00058834226,0.00020407958,0.0006993657,0.0007856051,0.00016275873],"domain_scores_gemma":[0.9946366,0.0019478984,0.000340942,0.0016589918,0.0009691826,0.00044639193],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0032712624,0.0013728451,0.0011788326,0.0023168747,0.001138379,0.0025989718,0.0038120016,0.0014147593,0.056699257],"category_scores_gemma":[0.01037124,0.0010450175,0.0009920114,0.0026707887,0.00077454356,0.002972403,0.0045587444,0.0018754596,0.030818868],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0036108564,0.00039360928,0.010578264,0.0009553129,0.00038020237,0.0013196047,0.0007997512,0.023876976,0.02604938,0.03091601,0.42882943,0.4722907],"study_design_scores_gemma":[0.001590424,0.00038685172,0.00676506,0.00022501637,0.00019294416,0.0015275263,0.00015607328,0.56205845,0.03614772,0.06916621,0.32143858,0.00034515987],"about_ca_topic_score_codex":0.0034725005,"about_ca_topic_score_gemma":0.002950658,"teacher_disagreement_score":0.056699257,"about_ca_system_score_codex":0.00083587965,"about_ca_system_score_gemma":0.0024819851,"threshold_uncertainty_score":0.18967795},"labels":[],"label_agreement":null},{"id":"W2159640103","doi":"10.1093/bioinformatics/btq216","title":"Next-generation VariationHunter: combinatorial algorithms for transposon insertion discovery","year":2010,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genome Rearrangement Algorithms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":221,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"Natural Sciences and Engineering Research Council of Canada; National Institutes of Health; Genome British Columbia; National Human Genome Research Institute; Howard Hughes Medical Institute","keywords":"Transposable element; Algorithm; Genome; Computer science; Human genome; Structural variation; Scope (computer science); Computational biology; Biology; Genetics; Gene","score_opus":0.024851415321221256,"score_gpt":0.2506231169510057,"score_spread":0.22577170162978444,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2159640103","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.012495964,0.00063186954,0.9779158,0.0004894799,0.00010056111,0.0001588792,0.0005444766,0.005494369,0.0021684442],"genre_scores_gemma":[0.10455511,0.00021517351,0.890883,0.00036658943,0.00008799665,0.00036974627,0.0013157716,0.0007758319,0.0014308126],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9989127,0.0004318581,0.00006118237,0.000250875,0.0002594671,0.000083857034],"domain_scores_gemma":[0.99537164,0.0037341295,0.00024201059,0.0002643063,0.0002734724,0.00011438947],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002240058,0.0013667373,0.0011556443,0.0016439909,0.0007545757,0.0013262831,0.0029492613,0.0016921164,0.007899821],"category_scores_gemma":[0.008525358,0.00060249504,0.0012970467,0.0022226665,0.0008961478,0.0013938268,0.0013634533,0.0020668004,0.0014994594],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00049319013,0.00024529037,0.003360318,0.00046224627,0.0002542602,0.00024746874,0.00011704878,0.62690324,0.003015629,0.020604493,0.015256006,0.3290408],"study_design_scores_gemma":[0.00007902255,0.00003092237,0.0001418372,0.000014212431,0.000019799792,0.00005773303,0.000009832466,0.98248994,0.0010667872,0.013729711,0.0023483648,0.000011839695],"about_ca_topic_score_codex":0.0035603973,"about_ca_topic_score_gemma":0.0044311183,"teacher_disagreement_score":0.007899821,"about_ca_system_score_codex":0.0009269212,"about_ca_system_score_gemma":0.0016568579,"threshold_uncertainty_score":0.026427507},"labels":[],"label_agreement":null},{"id":"W2159739447","doi":"10.1093/bioinformatics/btv341","title":"JSBML 1.0: providing a smorgasbord of options to encode systems biology models","year":2015,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene Regulatory Network Analysis","field":"Biochemistry, Genetics and Molecular Biology","cited_by":42,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"National Institute of Allergy and Infectious Diseases; Biotechnology and Biological Sciences Research Council; Seventh Framework Programme; Danmarks Tekniske Universitet; National Institute of General Medical Sciences; Novo Nordisk; Bundesministerium für Bildung und Forschung; European Commission; Deutsche Forschungsgemeinschaft; Novo Nordisk Fonden; National Science Foundation; Google; National Institutes of Health","keywords":"ENCODE; Computer science; Computational biology; Systems biology; Biology; Data science; Genetics; Gene","score_opus":0.034921473298470486,"score_gpt":0.2662321145306797,"score_spread":0.2313106412322092,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2159739447","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0032147123,0.0007452255,0.6268924,0.0014190746,0.00071069365,0.00029202824,0.0622993,0.29156414,0.012862459],"genre_scores_gemma":[0.042246558,0.0018961107,0.65968704,0.0017457405,0.00026449506,0.0023464842,0.18222292,0.093619406,0.015971323],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99866307,0.0002819504,0.00021189067,0.00026355663,0.00045166453,0.00012787378],"domain_scores_gemma":[0.9959765,0.002214218,0.00039548552,0.0006405464,0.0004678322,0.00030540503],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0035594483,0.0024223237,0.0011696251,0.003249582,0.0013179453,0.0039826124,0.0043476466,0.0027677761,0.05483528],"category_scores_gemma":[0.010402732,0.0024053783,0.0031447369,0.0027000485,0.0014589535,0.0040187337,0.004755085,0.0036232758,0.031523988],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014793626,0.00029424956,0.004926907,0.0045695817,0.0002483111,0.0013474587,0.0014474365,0.033622734,0.02074614,0.1455777,0.6072506,0.17848961],"study_design_scores_gemma":[0.000311292,0.00006625212,0.00089077564,0.0010504713,0.00010347048,0.00039055868,0.00013770384,0.08721931,0.017117895,0.093575545,0.79893756,0.00019907704],"about_ca_topic_score_codex":0.0038734514,"about_ca_topic_score_gemma":0.006131039,"teacher_disagreement_score":0.05483528,"about_ca_system_score_codex":0.0014536497,"about_ca_system_score_gemma":0.0028025564,"threshold_uncertainty_score":0.1834423},"labels":[],"label_agreement":null},{"id":"W2160004708","doi":"10.1093/bioinformatics/btl259","title":"Distance based algorithms for small biomolecule classification and structural similarity search","year":2006,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Computational Drug Discovery Methods","field":"Computer Science","cited_by":35,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia; Simon Fraser University","funders":"","keywords":"Minkowski distance; Classifier (UML); Nearest neighbor search; Computer science; Similarity (geometry); Artificial intelligence; Pruning; Pattern recognition (psychology); Data mining; Focus (optics); k-nearest neighbors algorithm; Machine learning; Algorithm; Mathematics; Euclidean distance","score_opus":0.054116687373180636,"score_gpt":0.31771800108442666,"score_spread":0.263601313711246,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2160004708","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0029231363,0.0012023677,0.9939731,0.0001259602,0.00003863496,0.0000777164,0.00008969346,0.00056753465,0.0010019126],"genre_scores_gemma":[0.05997569,0.0011754821,0.93519276,0.00012754413,0.00010696604,0.00041145235,0.0007692461,0.00016580375,0.002075015],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9978289,0.00053296634,0.00022765195,0.00039955458,0.0009371897,0.00007362195],"domain_scores_gemma":[0.9963852,0.00219632,0.00033738062,0.00041446736,0.00058477366,0.00008195351],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0019178706,0.0010465642,0.0018334002,0.0030285749,0.0005579989,0.0016563311,0.0024484955,0.0017874899,0.00395726],"category_scores_gemma":[0.009251792,0.00051575975,0.0010204911,0.004141043,0.0010226833,0.00258161,0.0018185352,0.0019576363,0.0023228575],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00014334735,0.00013883489,0.00080748205,0.00047627222,0.00009643005,0.00006397161,0.000089231624,0.27781874,0.0029933662,0.06365641,0.005998992,0.647717],"study_design_scores_gemma":[0.000038212464,0.000064923224,0.00029489317,0.000043324108,0.000017530181,0.00010881058,0.000024206074,0.90415674,0.0023416805,0.085373655,0.0075113787,0.000024582903],"about_ca_topic_score_codex":0.0022418534,"about_ca_topic_score_gemma":0.0015701439,"teacher_disagreement_score":0.00395726,"about_ca_system_score_codex":0.0013408164,"about_ca_system_score_gemma":0.0010598303,"threshold_uncertainty_score":0.01323837},"labels":[],"label_agreement":null},{"id":"W2160265768","doi":"10.1093/bioinformatics/btr208","title":"Error correction of high-throughput sequencing datasets with non-uniform coverage","year":2011,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Cancer Genomics and Diagnostics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":123,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"National Center for Research Resources; National Institutes of Health","keywords":"Computer science; Error detection and correction; Throughput; Simple (philosophy); Hamming graph; Algorithm; Hamming distance; Hamming code; Data mining; Decoding methods; Telecommunications","score_opus":0.015894315644210757,"score_gpt":0.22750690572631177,"score_spread":0.211612590082101,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2160265768","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.046043567,0.0006094792,0.94777817,0.0003351693,0.00013575814,0.00008386315,0.00090664596,0.003784359,0.0003229621],"genre_scores_gemma":[0.27484515,0.00038096507,0.7164823,0.00048846094,0.00010212655,0.00021665455,0.0054741143,0.00088663324,0.0011236895],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.98594844,0.0045321127,0.0015012332,0.0037668364,0.003835179,0.00041624697],"domain_scores_gemma":[0.93942344,0.03374427,0.005288516,0.01575376,0.005212532,0.0005774558],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.013680268,0.0012270678,0.0017799456,0.0028718275,0.0013870919,0.001592368,0.0031143096,0.002040406,0.0010187421],"category_scores_gemma":[0.062032085,0.00082374277,0.0018321597,0.0046351156,0.0022195089,0.0029032226,0.0039273854,0.0030247534,0.00076320104],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0015627171,0.00024696402,0.074504845,0.0021438187,0.0015713746,0.0014403546,0.0012790282,0.33929995,0.099827506,0.025154429,0.013335012,0.439634],"study_design_scores_gemma":[0.00011744739,0.00021170855,0.017888503,0.00013696059,0.00026847428,0.00153563,0.0002671789,0.6983899,0.18247274,0.088074505,0.010481948,0.00015504888],"about_ca_topic_score_codex":0.0019022117,"about_ca_topic_score_gemma":0.002838563,"teacher_disagreement_score":0.013680268,"about_ca_system_score_codex":0.0013015821,"about_ca_system_score_gemma":0.0016755196,"threshold_uncertainty_score":0.07234901},"labels":[],"label_agreement":null},{"id":"W2160615485","doi":"10.1093/bioinformatics/bti120","title":"A Java tool for dynamic web-based 3D visualization of anatomy and overlapping gene or protein expression patterns","year":2004,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Biomedical Text Mining and Ontologies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"","keywords":"Java; Visualization; Computer science; Computational biology; Expression (computer science); Web application; Gene expression; Computer graphics (images); Gene; Biology; Anatomy; Programming language; World Wide Web; Data mining; Genetics","score_opus":0.011974473710474309,"score_gpt":0.2844673868023079,"score_spread":0.27249291309183354,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2160615485","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0009726354,0.00019181025,0.6925007,0.00016676242,0.00012336342,0.00025365586,0.014810496,0.28590643,0.005074177],"genre_scores_gemma":[0.017261004,0.0008098364,0.81694704,0.0006834875,0.00012566008,0.0019572126,0.057289183,0.084035955,0.020890737],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99923456,0.000085918604,0.000108409804,0.00015636253,0.0003430331,0.00007172219],"domain_scores_gemma":[0.9983859,0.00079791644,0.00012779732,0.0002506014,0.00023157023,0.00020619099],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0017236859,0.001996765,0.001447535,0.0031949559,0.00070249534,0.0024557854,0.003265068,0.0014111828,0.08236214],"category_scores_gemma":[0.0030043942,0.0018393332,0.0016742646,0.0016733268,0.0005786491,0.0023528545,0.003597781,0.0024287354,0.040066384],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00053670024,0.00038486664,0.0018593312,0.0019424146,0.0002552904,0.0010850724,0.00044952816,0.004484866,0.05592012,0.0113702,0.5138743,0.4078374],"study_design_scores_gemma":[0.0007512148,0.00014595373,0.00679949,0.00056717324,0.00020218903,0.002770056,0.0001538154,0.10151542,0.054981228,0.033090264,0.7986494,0.0003738536],"about_ca_topic_score_codex":0.0024351834,"about_ca_topic_score_gemma":0.004203144,"teacher_disagreement_score":0.08236214,"about_ca_system_score_codex":0.00054577837,"about_ca_system_score_gemma":0.0011307175,"threshold_uncertainty_score":0.27552885},"labels":[],"label_agreement":null},{"id":"W2160972508","doi":"10.1093/bioinformatics/bts411","title":"From phenotype to genotype: an association study of longitudinal phenotypic markers to Alzheimer's disease relevant SNPs","year":2012,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":71,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"U.S. National Library of Medicine; National Institute on Aging; Genentech; National Institutes of Health; Canadian Institutes of Health Research; University of California, Los Angeles; Servier; Eisai; Northern California Institute for Research and Education; University of California, San Diego; Pfizer; Biogen; BioClinica; Amorfix Life Sciences; Dana Foundation; Bayer HealthCare; Meso Scale Diagnostics; National Center for Research Resources; F. Hoffmann-La Roche; Medpace; Bristol-Myers Squibb; Eli Lilly and Company; AstraZeneca; Novartis Pharmaceuticals Corporation; Synarc; Foundation for the National Institutes of Health; National Science Foundation","keywords":"Single-nucleotide polymorphism; SNP; Genetic association; Genotype; Computational biology; Biology; Imaging genetics; Phenotype; Genetics; Genome-wide association study; Longitudinal study; Neuroimaging; Medicine; Pathology; Neuroscience; Gene","score_opus":0.022760359593949547,"score_gpt":0.2850777229348396,"score_spread":0.26231736334089,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2160972508","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7352806,0.0018974781,0.24388699,0.0073527982,0.00021445737,0.00014410281,0.007632266,0.0011167248,0.002474587],"genre_scores_gemma":[0.90991455,0.0007340092,0.078841776,0.00082742213,0.00020447123,0.00022222973,0.005814152,0.00019708168,0.0032442543],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9986518,0.0007825342,0.000046605954,0.0003729312,0.000098408826,0.0000477798],"domain_scores_gemma":[0.98781174,0.008921428,0.0010027642,0.0012138663,0.00063124805,0.00041882915],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0054610353,0.0006576809,0.00066157326,0.00071576156,0.0005764389,0.0008249572,0.0010015406,0.0010432911,0.0039442866],"category_scores_gemma":[0.019701347,0.0004073053,0.00092643336,0.0014344519,0.00057624996,0.0008424393,0.0012237261,0.0017738609,0.0007856712],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0041627907,0.0009783119,0.6444163,0.00043937526,0.0015098657,0.0021274548,0.0010335536,0.07333438,0.0066871364,0.0096625285,0.028944058,0.22670433],"study_design_scores_gemma":[0.0005462161,0.00087205874,0.20152208,0.00016592786,0.0012127294,0.002062225,0.00040528012,0.7146318,0.0034826274,0.06425257,0.010700279,0.00014619835],"about_ca_topic_score_codex":0.007077538,"about_ca_topic_score_gemma":0.008331683,"teacher_disagreement_score":0.007077538,"about_ca_system_score_codex":0.00042681792,"about_ca_system_score_gemma":0.00096584787,"threshold_uncertainty_score":0.028881073},"labels":[],"label_agreement":null},{"id":"W2161186860","doi":"10.1093/bioinformatics/bth148","title":"Haplotypes histories as pathways of recombinations","year":2004,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic Neurodegenerative Diseases","field":"Neuroscience","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Centre Hospitalier Universitaire Sainte-Justine; Université de Montréal","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Haplotype; Biology; Recombination; Genetics; Genome; Evolutionary biology; Sequence (biology); Computational biology; Gene; Allele","score_opus":0.03892729957076401,"score_gpt":0.25214078638459925,"score_spread":0.21321348681383523,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2161186860","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.43660802,0.0010457603,0.5502064,0.0004444407,0.000027430398,0.00020074952,0.0034071724,0.0011470548,0.006912984],"genre_scores_gemma":[0.74495506,0.00072911277,0.24814841,0.00005665573,0.000026112795,0.00013938072,0.0032865664,0.00022157382,0.0024371033],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9994444,0.00014214059,0.000049843868,0.0001608133,0.00014580571,0.00005693141],"domain_scores_gemma":[0.995672,0.0029156515,0.000504929,0.00029849893,0.00032514177,0.00028388028],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010079046,0.00028891614,0.0002765621,0.0022750662,0.00070102105,0.0015058045,0.0007626641,0.00067826995,0.0042776503],"category_scores_gemma":[0.008002861,0.0004299773,0.0005183262,0.0018795854,0.0008080243,0.0021082729,0.0009998397,0.0006754533,0.0005840445],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009590153,0.00017187106,0.11902357,0.0010566658,0.00016255502,0.0011694852,0.002969898,0.3651122,0.020972626,0.15526018,0.004295007,0.32884684],"study_design_scores_gemma":[0.000073192365,0.00019398607,0.02339092,0.00024631264,0.0001191273,0.0020213225,0.00047437454,0.5813897,0.015588085,0.3570145,0.019395191,0.00009331941],"about_ca_topic_score_codex":0.001467184,"about_ca_topic_score_gemma":0.0017405959,"teacher_disagreement_score":0.0042776503,"about_ca_system_score_codex":0.00067898433,"about_ca_system_score_gemma":0.00052887975,"threshold_uncertainty_score":0.014310181},"labels":[],"label_agreement":null},{"id":"W2161241698","doi":"10.1093/bioinformatics/bts060","title":"Predicting kinase substrates using conservation of local motif density","year":2012,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":20,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Kinase; Computer science; Motif (music); Computational biology; Training set; Artificial intelligence; Biology; Machine learning; Genetics; Physics","score_opus":0.018419556843068275,"score_gpt":0.23568615793772005,"score_spread":0.21726660109465176,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2161241698","genre_codex":"empirical","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8917675,0.0010001841,0.1024813,0.0001481642,0.000013964707,0.000059995025,0.0019733093,0.0010257872,0.0015298199],"genre_scores_gemma":[0.964661,0.00024852643,0.03149599,0.000024834575,0.000024255305,0.000025800955,0.0029917017,0.00004744214,0.00048040238],"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9996687,0.00006904916,0.000025972775,0.00012548604,0.000081309845,0.000029484723],"domain_scores_gemma":[0.99866045,0.0006359732,0.0003113704,0.00010163235,0.00015233575,0.00013818941],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006970463,0.0005900243,0.00072835194,0.0024352274,0.0003009447,0.0007016767,0.00048648258,0.00045824848,0.0015738357],"category_scores_gemma":[0.0030316547,0.00025659852,0.00034970389,0.0014035974,0.00032971677,0.0008690693,0.00066223845,0.00041035848,0.0006845353],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.002180723,0.00043566295,0.33577818,0.0008558043,0.00037945822,0.0009420965,0.00019919492,0.3005299,0.095704935,0.0039186906,0.00593276,0.2531426],"study_design_scores_gemma":[0.000084360276,0.00019763233,0.034983393,0.000024759622,0.000077301134,0.00077398325,0.00007108206,0.9231166,0.033098396,0.0060649714,0.0014690969,0.00003851132],"about_ca_topic_score_codex":0.001558988,"about_ca_topic_score_gemma":0.0025369097,"teacher_disagreement_score":0.0024352274,"about_ca_system_score_codex":0.0003652202,"about_ca_system_score_gemma":0.0004422315,"threshold_uncertainty_score":0.0052649975},"labels":[],"label_agreement":null},{"id":"W2161481677","doi":"10.1093/bioinformatics/btq255","title":"Pathgroups, a dynamic data structure for genome reconstruction problems","year":2010,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genome Rearrangement Algorithms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":18,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal; Computer Research Institute of Montréal","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Heuristics; Computer science; Heuristic; Greedy algorithm; Genome; Set (abstract data type); Data structure; Sequence (biology); Algorithm; Scheme (mathematics); Theoretical computer science; Data mining; Computational biology; Gene; Artificial intelligence; Biology; Mathematics; Genetics","score_opus":0.012882705248054703,"score_gpt":0.24099728298190282,"score_spread":0.2281145777338481,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2161481677","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.019101942,0.00073915033,0.93906546,0.0018430746,0.00018125607,0.0004847002,0.018199144,0.0150044905,0.005380711],"genre_scores_gemma":[0.06757856,0.00043309398,0.89951885,0.00025065485,0.000093541916,0.00077718525,0.027476823,0.0014869154,0.0023842535],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9985403,0.00036664886,0.00015398831,0.00040775086,0.00043714273,0.0000942225],"domain_scores_gemma":[0.9946208,0.002751745,0.00037949407,0.001433843,0.00053644297,0.00027763663],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0024073562,0.001052406,0.000952931,0.001968866,0.0016197074,0.0023054092,0.0025981402,0.0013342573,0.007888397],"category_scores_gemma":[0.011016272,0.00084613543,0.0012807724,0.0038728607,0.0010468549,0.004164982,0.0027892445,0.0034111412,0.0027237907],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014246915,0.00051921787,0.004214623,0.001473963,0.000103806335,0.00038571088,0.0006843174,0.20979539,0.009708568,0.17345354,0.13725677,0.4609794],"study_design_scores_gemma":[0.0004467982,0.00030825692,0.00071293704,0.00011535413,0.00003980621,0.0005129386,0.00018729645,0.5571028,0.0060241213,0.33001864,0.1044684,0.000062700325],"about_ca_topic_score_codex":0.002593694,"about_ca_topic_score_gemma":0.0044001755,"teacher_disagreement_score":0.007888397,"about_ca_system_score_codex":0.0013527876,"about_ca_system_score_gemma":0.003854677,"threshold_uncertainty_score":0.02638936},"labels":[],"label_agreement":null},{"id":"W2161666955","doi":"10.1093/bioinformatics/btm489","title":"BiasViz: visualization of amino acid biased regions in protein alignments","year":2007,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Ottawa","funders":"Biotechnology and Biological Sciences Research Council","keywords":"Sequence (biology); Source code; Computer science; Visualization; Java; Java applet; Protein sequencing; Sequence alignment; Protein superfamily; Computational biology; Amino acid; Peptide sequence; Code (set theory); Amino acid residue; Programming language; Biology; Data mining; Set (abstract data type); Genetics","score_opus":0.019369109772539153,"score_gpt":0.2649673212057758,"score_spread":0.24559821143323668,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2161666955","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.106435284,0.0017460117,0.45752758,0.0013643736,0.00042529585,0.0004698446,0.0410307,0.37629825,0.014702643],"genre_scores_gemma":[0.3652638,0.002248269,0.54658043,0.0005274645,0.00012976151,0.0012381334,0.034533445,0.041755807,0.0077228555],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99967706,0.00006705601,0.000026891315,0.00005995816,0.00012416253,0.000044807595],"domain_scores_gemma":[0.9990496,0.0005611494,0.00011194938,0.00008342293,0.00011026329,0.00008363148],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008915111,0.001122723,0.0006859368,0.0029545536,0.0005185758,0.0016369871,0.0010131737,0.00094970426,0.02749608],"category_scores_gemma":[0.0033184008,0.00045584486,0.0005808798,0.0022746096,0.00021700807,0.0015198035,0.0018521979,0.0012088105,0.0039469907],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0031975335,0.00037510786,0.013528372,0.0049049286,0.00062213605,0.0020613144,0.003135368,0.010108334,0.22311495,0.020968389,0.24383414,0.47414944],"study_design_scores_gemma":[0.0014211243,0.00047700817,0.05026532,0.0012089174,0.00030762062,0.002581866,0.001315593,0.29437461,0.17987394,0.039081726,0.42867658,0.00041574673],"about_ca_topic_score_codex":0.0013180005,"about_ca_topic_score_gemma":0.001459091,"teacher_disagreement_score":0.02749608,"about_ca_system_score_codex":0.00047185805,"about_ca_system_score_gemma":0.00053567305,"threshold_uncertainty_score":0.0919835},"labels":[],"label_agreement":null},{"id":"W2161722223","doi":"10.1093/bioinformatics/btv541","title":"markophylo: Markov chain analysis on phylogenetic trees","year":2015,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University","funders":"Innovative Research Group Project of the National Natural Science Foundation of China","keywords":"Markov chain; R package; Phylogenetic tree; Computer science; Tree (set theory); Markov chain Monte Carlo; Markov model; Data mining; Theoretical computer science; Bayesian probability; Mathematics; Artificial intelligence; Biology; Machine learning; Combinatorics; Genetics; Programming language; Gene","score_opus":0.015612745329582954,"score_gpt":0.2358507888474713,"score_spread":0.22023804351788834,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2161722223","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0012138742,0.00036353598,0.9431102,0.0002903687,0.00014123895,0.00012749719,0.007819917,0.04475247,0.0021808997],"genre_scores_gemma":[0.029325273,0.00063465856,0.92109,0.00039427335,0.0001885124,0.0015478336,0.011797817,0.032016102,0.0030055598],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9974941,0.001238784,0.00021449693,0.0005138007,0.00039791153,0.00014098143],"domain_scores_gemma":[0.99083626,0.0067087743,0.0007105021,0.0010353915,0.0004307716,0.00027830587],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005591741,0.0021045732,0.0018538118,0.002638843,0.0011262413,0.002796904,0.0039726114,0.0014251759,0.05172444],"category_scores_gemma":[0.024549255,0.0020326367,0.0030938808,0.0022154946,0.0014257053,0.0033369944,0.004159656,0.0049378565,0.020294083],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00077322795,0.00016896406,0.005877869,0.004005439,0.0015308016,0.0011244738,0.0014228413,0.12712516,0.008485046,0.23292252,0.32740462,0.28915912],"study_design_scores_gemma":[0.00023910083,0.000075818476,0.001626699,0.00041880144,0.0002181319,0.00051409926,0.000082319566,0.49592605,0.0046239337,0.28538135,0.21069695,0.00019670186],"about_ca_topic_score_codex":0.0029278717,"about_ca_topic_score_gemma":0.0032474804,"teacher_disagreement_score":0.05172444,"about_ca_system_score_codex":0.000894894,"about_ca_system_score_gemma":0.0027172263,"threshold_uncertainty_score":0.17303544},"labels":[],"label_agreement":null},{"id":"W2161724277","doi":"10.1093/bioinformatics/btq040","title":"SNVMix: predicting single nucleotide variants from next-generation sequencing of tumors","year":2010,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Cancer Genomics and Diagnostics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":218,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Provincial Health Services Authority; BC Cancer Agency","funders":"Canadian Institutes of Health Research; Michael Smith Health Research BC","keywords":"Genome; Computational biology; Computer science; Probabilistic logic; Biology; Inference; DNA sequencing; Software; Exome sequencing; Data mining; Genetics; Artificial intelligence; Mutation; Gene","score_opus":0.02670086111264113,"score_gpt":0.2255201977902274,"score_spread":0.19881933667758625,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2161724277","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.12763616,0.0013793241,0.8289291,0.0008420795,0.00018765086,0.0003561638,0.01140059,0.027114535,0.0021542867],"genre_scores_gemma":[0.39678958,0.0005861063,0.5759452,0.0006228835,0.00014020105,0.0006823214,0.020863602,0.0016044428,0.0027656483],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9987417,0.0004269412,0.00006989569,0.00040954023,0.00026142696,0.00009045684],"domain_scores_gemma":[0.99746025,0.0017806153,0.00021677763,0.00022308691,0.0001962961,0.00012290516],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0064783576,0.0013758636,0.0009914846,0.0016855908,0.0006056937,0.0014013634,0.0026539112,0.0017413369,0.0027209257],"category_scores_gemma":[0.009507225,0.0009494663,0.0015826757,0.0011093266,0.00053278176,0.0011514456,0.0016341693,0.0014947597,0.0013123733],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011845098,0.0002297035,0.0755632,0.0006162166,0.0011361977,0.00042492113,0.00024783373,0.6943657,0.012993183,0.0063571227,0.023019813,0.18386158],"study_design_scores_gemma":[0.000028266815,0.00004328278,0.0016123189,0.000020626376,0.00003280875,0.000092868235,0.000013461013,0.98790735,0.0042157373,0.0045755072,0.001440752,0.00001702336],"about_ca_topic_score_codex":0.0065066554,"about_ca_topic_score_gemma":0.00793579,"teacher_disagreement_score":0.0065066554,"about_ca_system_score_codex":0.0013388456,"about_ca_system_score_gemma":0.0014196578,"threshold_uncertainty_score":0.034261227},"labels":[],"label_agreement":null},{"id":"W2161921787","doi":"10.1093/bioinformatics/btv630","title":"Fast Optimized Cluster Algorithm for Localizations (FOCAL): a spatial cluster analysis for super-resolved microscopy","year":2015,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Advanced Fluorescence Microscopy Techniques","field":"Biochemistry, Genetics and Molecular Biology","cited_by":59,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"DBSCAN; Focus (optics); Cluster analysis; Computer science; Scaling; Algorithm; Cluster (spacecraft); Software; Microscopy; Artificial intelligence; Cardinal point; Pattern recognition (psychology); Mathematics; Physics; Optics; Geometry; Fuzzy clustering","score_opus":0.015587870179144573,"score_gpt":0.298134211063087,"score_spread":0.2825463408839424,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2161921787","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0018404871,0.00012255892,0.99175185,0.00009310924,0.000025115854,0.000087208,0.00037804377,0.0052236924,0.0004778757],"genre_scores_gemma":[0.021893794,0.00013264928,0.9734038,0.00006119527,0.000015396929,0.00047875632,0.0012314906,0.0017779686,0.0010048719],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9989309,0.00029081895,0.000052728927,0.00020457787,0.00043937087,0.00008156036],"domain_scores_gemma":[0.998309,0.0006127607,0.000127054,0.00026982577,0.0005807231,0.00010047809],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0021140038,0.0017833344,0.0015095485,0.0019231195,0.0014199025,0.0016736797,0.0038968802,0.0016588871,0.008904817],"category_scores_gemma":[0.0060543017,0.00094362104,0.0015199273,0.0024108307,0.0009096061,0.0015770497,0.0027367182,0.0026829767,0.003714077],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006039816,0.00015308462,0.0016510168,0.00086234236,0.00039095592,0.00021855725,0.00039907006,0.46874043,0.019828547,0.049875364,0.060254436,0.3970222],"study_design_scores_gemma":[0.000047464615,0.000025947915,0.0002499835,0.000019291867,0.000014902723,0.00005192308,0.000031674965,0.9767581,0.00461298,0.011634307,0.006521489,0.000031773954],"about_ca_topic_score_codex":0.013246335,"about_ca_topic_score_gemma":0.014498217,"teacher_disagreement_score":0.013246335,"about_ca_system_score_codex":0.0019494606,"about_ca_system_score_gemma":0.0036563133,"threshold_uncertainty_score":0.029789567},"labels":[],"label_agreement":null},{"id":"W2161938868","doi":"10.1093/bioinformatics/btv192","title":"Gene selection for the reconstruction of stem cell differentiation trees: a linear programming approach","year":2015,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Wilfrid Laurier University; Ottawa Hospital; University of Ottawa","funders":"","keywords":"Hierarchy; Hierarchical clustering; Gene; Computational biology; Metric (unit); Tree (set theory); Euclidean distance; Cluster analysis; Biology; Computer science; Mathematics; Genetics; Artificial intelligence; Combinatorics","score_opus":0.03283974096337588,"score_gpt":0.25013622598395635,"score_spread":0.21729648502058047,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2161938868","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.015491206,0.00012788053,0.9830494,0.00025073206,0.000010152934,0.00007080117,0.00010671081,0.00032308867,0.0005700322],"genre_scores_gemma":[0.1678112,0.00016912812,0.8282267,0.00029306274,0.000055874643,0.00057364703,0.00087540667,0.00020959118,0.0017854813],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9991937,0.00035883888,0.00003139895,0.00016797728,0.00015520716,0.00009286962],"domain_scores_gemma":[0.997008,0.0023510489,0.000156882,0.000054401564,0.00034293087,0.000086824664],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002196974,0.001298499,0.0014558772,0.0015996485,0.0005730194,0.0009651795,0.0014428591,0.0012836936,0.0030416718],"category_scores_gemma":[0.005595165,0.0008829529,0.0012885935,0.0015000185,0.00108914,0.0008876671,0.0009248657,0.0018947312,0.00061343535],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00011593917,0.00012386632,0.0015038648,0.00012448296,0.00004296358,0.00008911765,0.00010266333,0.912621,0.0032223344,0.00881888,0.0018952858,0.07133957],"study_design_scores_gemma":[0.000008523685,0.000020661526,0.000054874177,0.0000061398655,0.0000042487163,0.000008204992,0.000011242288,0.9950039,0.0004713302,0.0041998806,0.00020786167,0.000003171429],"about_ca_topic_score_codex":0.0048896796,"about_ca_topic_score_gemma":0.004872429,"teacher_disagreement_score":0.0048896796,"about_ca_system_score_codex":0.0013445037,"about_ca_system_score_gemma":0.0017711349,"threshold_uncertainty_score":0.011618793},"labels":[],"label_agreement":null},{"id":"W2162156442","doi":"10.1093/bioinformatics/btg378","title":"GRIL: genome rearrangement and inversion locator","year":2003,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":35,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"U.S. National Library of Medicine; National Institute of General Medical Sciences; National Institutes of Health; Groupe de recherche interuniversitaire en limnologie; Mashhad University of Medical Sciences","keywords":"Correctness; Computer science; Genome; Sequence (biology); Basis (linear algebra); Set (abstract data type); Data mining; Computational biology; Algorithm; Theoretical computer science; Biology; Genetics; Mathematics; Programming language; Gene","score_opus":0.010679139315246573,"score_gpt":0.20592708579076494,"score_spread":0.19524794647551835,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2162156442","genre_codex":"software","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0043400126,0.00020039373,0.21301097,0.00033895904,0.00022401378,0.00041205462,0.043388113,0.7332515,0.0048338505],"genre_scores_gemma":[0.058751628,0.0003088747,0.5691445,0.0007902198,0.00024275076,0.0018679589,0.21641566,0.13858013,0.013898233],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99834406,0.0002637341,0.00014250189,0.0006030392,0.00045717345,0.00018957148],"domain_scores_gemma":[0.99717927,0.0011613452,0.0005306866,0.00059332704,0.00032243226,0.00021307608],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003691385,0.002677012,0.0016700967,0.0051302714,0.0012005423,0.0022197457,0.005307035,0.0017839745,0.0779147],"category_scores_gemma":[0.008648077,0.002031515,0.0027669442,0.0022977479,0.0008991551,0.002667434,0.003428478,0.0030803035,0.060812257],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001961205,0.00027509924,0.005027526,0.0024191092,0.00023962077,0.0013206099,0.0005963529,0.008032983,0.036980517,0.012913105,0.68410563,0.24612819],"study_design_scores_gemma":[0.0019687635,0.0006577548,0.009032557,0.00057678676,0.00024729365,0.0022575248,0.0002935534,0.119860984,0.13027926,0.02481038,0.7093622,0.00065301993],"about_ca_topic_score_codex":0.0014494723,"about_ca_topic_score_gemma":0.0015203933,"teacher_disagreement_score":0.0779147,"about_ca_system_score_codex":0.0013481928,"about_ca_system_score_gemma":0.0012421457,"threshold_uncertainty_score":0.26065063},"labels":[],"label_agreement":null},{"id":"W2162350421","doi":"10.1093/bioinformatics/btt707","title":"A phylogenetic Kalman filter for ancestral trait reconstruction using molecular data","year":2013,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":16,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"","keywords":"Phylogenetic tree; Trait; Bayesian probability; Regression; Biology; Phylogenetics; Evolutionary biology; Algorithm; Computer science; Statistics; Mathematics; Genetics; Gene","score_opus":0.04485866981425683,"score_gpt":0.26827159190331573,"score_spread":0.22341292208905889,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2162350421","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0013440797,0.000026559661,0.99805224,0.000030182093,0.000007865342,0.000009595751,0.00003841534,0.00033434207,0.00015663319],"genre_scores_gemma":[0.08202805,0.00018839217,0.91501045,0.00007730464,0.00006190673,0.00021500376,0.00054454495,0.00018876228,0.0016857196],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99952376,0.00013867088,0.000032763786,0.00012921197,0.00012694695,0.000048753136],"domain_scores_gemma":[0.9981292,0.0013533926,0.00012345529,0.00010449613,0.00022905956,0.000060368697],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0021417027,0.00062539405,0.0010026451,0.00093531143,0.0007274828,0.00085766776,0.001387047,0.0013910466,0.0044628894],"category_scores_gemma":[0.007424865,0.0006903747,0.00092227553,0.00097549823,0.00084671774,0.0013428746,0.001137366,0.0015356372,0.001439413],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00016244242,0.000061386716,0.0023348972,0.00020218175,0.00010598189,0.0001453157,0.00020782367,0.732708,0.0053741955,0.06413746,0.002247756,0.19231254],"study_design_scores_gemma":[0.000015332733,0.000013209779,0.00013546177,0.000013114633,0.000009527232,0.000020334026,0.0000070496444,0.9878401,0.00069969427,0.010189379,0.0010464195,0.000010371322],"about_ca_topic_score_codex":0.011638461,"about_ca_topic_score_gemma":0.011424843,"teacher_disagreement_score":0.011638461,"about_ca_system_score_codex":0.0011188361,"about_ca_system_score_gemma":0.0026672552,"threshold_uncertainty_score":0.023141444},"labels":[],"label_agreement":null},{"id":"W2162473677","doi":"10.1093/bioinformatics/btv361","title":"MetaPathways v2.5: quantitative functional, taxonomic and usability improvements","year":2015,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":53,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Genome British Columbia; University of British Columbia","funders":"Pacific Northwest National Laboratory; Natural Sciences and Engineering Research Council of Canada; Genome Alberta; University of British Columbia; Genome British Columbia; Genome Canada; Tula Foundation; Canadian Institute for Advanced Research; Western Canada Research Grid; Compute Canada","keywords":"Computer science; Usability; Pipeline (software); Annotation; Software; Data mining; Information retrieval; Artificial intelligence; Human–computer interaction; Programming language","score_opus":0.056242950902730474,"score_gpt":0.25802429057536885,"score_spread":0.2017813396726384,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2162473677","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.021392656,0.0005744569,0.17053631,0.00090532744,0.0005329332,0.0007082324,0.08498173,0.7151906,0.0051777624],"genre_scores_gemma":[0.080727294,0.0008055627,0.4447661,0.0009704744,0.00018074938,0.0027646667,0.26344866,0.20038278,0.005953739],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99279344,0.0017159745,0.0007786042,0.0016516694,0.0024255728,0.00063469936],"domain_scores_gemma":[0.98854476,0.0040271254,0.0009888439,0.003656522,0.0019783715,0.0008043646],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.012702012,0.0056418204,0.0019831709,0.004986025,0.002064745,0.004281958,0.00832044,0.0018299251,0.03150194],"category_scores_gemma":[0.023910753,0.0031193632,0.00419792,0.006044284,0.0008584988,0.0054781917,0.006286115,0.0051208325,0.023654783],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.007898731,0.00090713816,0.027452018,0.0054501477,0.0019420984,0.0008000086,0.001508257,0.007211024,0.084644675,0.007246064,0.59755903,0.25738078],"study_design_scores_gemma":[0.0015091334,0.0011559782,0.027498377,0.0013710244,0.0010985531,0.0015844387,0.0008072222,0.09301652,0.16972941,0.02034932,0.6805209,0.0013591537],"about_ca_topic_score_codex":0.0051424406,"about_ca_topic_score_gemma":0.006743498,"teacher_disagreement_score":0.03150194,"about_ca_system_score_codex":0.001668123,"about_ca_system_score_gemma":0.0038333682,"threshold_uncertainty_score":0.10538453},"labels":[],"label_agreement":null},{"id":"W2162671544","doi":"10.1093/bioinformatics/btt261","title":"pyGenClean: efficient tool for genetic data clean up before association testing","year":2013,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":20,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal; Montreal Heart Institute","funders":"","keywords":"Python (programming language); Computer science; Genotyping; Data mining; Software; Source code; Pipeline (software); Documentation; Data quality; Operating system; Genotype; Biology; Engineering","score_opus":0.029356549159342018,"score_gpt":0.2703399993806277,"score_spread":0.24098345022128567,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2162671544","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0032349003,0.0002599282,0.59480894,0.00064134493,0.00030122523,0.00028276426,0.023527836,0.374859,0.0020841756],"genre_scores_gemma":[0.034617834,0.00047145897,0.8058417,0.0015150559,0.00026707535,0.0020610185,0.05505762,0.09339986,0.006768352],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9940925,0.0013964408,0.0006657686,0.0013927858,0.0020858455,0.00036662628],"domain_scores_gemma":[0.98527247,0.007855254,0.0015523712,0.0028268017,0.0017723683,0.0007207427],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.009667646,0.0028546345,0.002326467,0.0040326826,0.0019060121,0.0031938977,0.0055390117,0.0015558478,0.051080797],"category_scores_gemma":[0.029892389,0.0021397492,0.0027230552,0.004547769,0.0015359685,0.003423293,0.005491232,0.00484938,0.032796003],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0016621685,0.00036209403,0.018056093,0.0020555875,0.0007095592,0.0010357348,0.0008324966,0.008472307,0.019413965,0.011545835,0.59522396,0.34063017],"study_design_scores_gemma":[0.0011530307,0.00032050707,0.01899364,0.00061999983,0.0003859218,0.0023302247,0.00025795857,0.17890899,0.09447143,0.09477449,0.6069675,0.0008162745],"about_ca_topic_score_codex":0.0025240302,"about_ca_topic_score_gemma":0.002443202,"teacher_disagreement_score":0.051080797,"about_ca_system_score_codex":0.0009493048,"about_ca_system_score_gemma":0.0055440334,"threshold_uncertainty_score":0.17088228},"labels":[],"label_agreement":null},{"id":"W2162792752","doi":"10.1093/bioinformatics/btq249","title":"PSORTb 3.0: improved protein subcellular localization prediction with refined localization subcategories and predictive capabilities for all prokaryotes","year":2010,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Machine Learning in Bioinformatics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":2574,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University; University of British Columbia","funders":"Natural Sciences and Engineering Research Council of Canada; Canadian Institutes of Health Research; Simon Fraser University; Michael Smith Health Research BC; Cystic Fibrosis Foundation","keywords":"Proteome; Protein subcellular localization prediction; Archaea; Computer science; Subcellular localization; Precision and recall; Computational biology; Interface (matter); Proteomics; Software; Metagenomics; Biology; Bioinformatics; Data mining; Artificial intelligence; Bacteria; Cytoplasm; Genetics; Gene; Programming language","score_opus":0.004537457218943714,"score_gpt":0.21131141928287536,"score_spread":0.20677396206393164,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2162792752","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.25007883,0.0018730409,0.5257825,0.0013382726,0.00021256247,0.00033135345,0.031996958,0.18451531,0.0038711354],"genre_scores_gemma":[0.27718467,0.0006646733,0.6211372,0.0004525877,0.00008731647,0.00059640507,0.08416393,0.0111011015,0.004612062],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9991978,0.00016804496,0.00007287906,0.00023856162,0.00025486219,0.00006780422],"domain_scores_gemma":[0.99848676,0.000557114,0.00019989136,0.00020850256,0.00041014733,0.00013757804],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0023019197,0.0020183923,0.0015196754,0.0010953607,0.000555147,0.0014426359,0.0014396303,0.0006951142,0.0048347325],"category_scores_gemma":[0.005074054,0.00078153703,0.0015316805,0.0012009735,0.00029765806,0.0012180135,0.0014560181,0.0016244868,0.0043951985],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.007678094,0.0007391932,0.060842797,0.0028500704,0.0014124272,0.0010816719,0.00045685418,0.10624388,0.31236136,0.0033905304,0.16924104,0.33370218],"study_design_scores_gemma":[0.00054618885,0.00048939907,0.019694379,0.00013851853,0.00040125265,0.0009964992,0.00007256969,0.8062905,0.13225658,0.003538151,0.035355803,0.00022009111],"about_ca_topic_score_codex":0.0030625477,"about_ca_topic_score_gemma":0.0043097376,"teacher_disagreement_score":0.0048347325,"about_ca_system_score_codex":0.0005243601,"about_ca_system_score_gemma":0.0010098312,"threshold_uncertainty_score":0.01617384},"labels":[],"label_agreement":null},{"id":"W2162797435","doi":"10.1093/bioinformatics/btl233","title":"BNTagger: improved tagging SNP selection using Bayesian networks","year":2006,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Machine Learning in Bioinformatics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":56,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Queen's University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Selection (genetic algorithm); Computer science; SNP; Bayesian probability; Artificial intelligence; Biology; Genetics; Single-nucleotide polymorphism","score_opus":0.006642357560736971,"score_gpt":0.24105347208302355,"score_spread":0.23441111452228658,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2162797435","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0119043635,0.0007288791,0.97848094,0.00028301374,0.00008838083,0.000107495944,0.00096458907,0.006310992,0.001131333],"genre_scores_gemma":[0.17564982,0.00086306175,0.8113166,0.0006881336,0.00020143954,0.00039363102,0.0057372283,0.0008204214,0.0043296507],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9973347,0.001102371,0.00013217362,0.000533407,0.00074301846,0.00015430235],"domain_scores_gemma":[0.99618787,0.00277116,0.00023261295,0.00024916534,0.00043441574,0.00012480367],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0048744273,0.0014020156,0.0017653416,0.0024057867,0.00071363617,0.0011782392,0.0022027944,0.0013641963,0.0036221296],"category_scores_gemma":[0.008763281,0.0008210302,0.0013413989,0.0019132757,0.00046179377,0.0018274132,0.0014648617,0.0018402752,0.0019996671],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011977111,0.00041162272,0.0144677395,0.00039083016,0.00055178104,0.00044689022,0.0001894739,0.25890657,0.010929918,0.010576358,0.025692545,0.67623854],"study_design_scores_gemma":[0.00010475356,0.000038225626,0.0012157764,0.00003229313,0.000061437364,0.00012028704,0.000012176444,0.97873586,0.0028703462,0.012585734,0.0041839513,0.00003920459],"about_ca_topic_score_codex":0.014150385,"about_ca_topic_score_gemma":0.019177169,"teacher_disagreement_score":0.014150385,"about_ca_system_score_codex":0.0009363269,"about_ca_system_score_gemma":0.0020616376,"threshold_uncertainty_score":0.028136015},"labels":[],"label_agreement":null},{"id":"W2163059805","doi":"10.1093/bioinformatics/btv047","title":"KAPPA, a simple algorithm for discovery and clustering of proteins defined by a key amino acid pattern: a case study of the cysteine-rich proteins","year":2015,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Machine Learning in Bioinformatics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Executable; Computer science; Cluster analysis; Computational biology; Kappa; Data mining; Pairwise comparison; Cysteine; Set (abstract data type); Sequence (biology); Protein superfamily; Protein sequencing; Peptide sequence; Bioinformatics; Theoretical computer science; Artificial intelligence; Biology; Mathematics; Genetics; Biochemistry; Programming language","score_opus":0.014589410809665969,"score_gpt":0.25502361307559557,"score_spread":0.2404342022659296,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2163059805","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.028160736,0.00081061077,0.93618363,0.000334402,0.00013217218,0.00047718745,0.002483579,0.028570794,0.0028469258],"genre_scores_gemma":[0.050038397,0.00031616705,0.94166404,0.0001364645,0.000027705335,0.00035353217,0.003504702,0.0020857009,0.0018732259],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9987431,0.0002433315,0.00014148858,0.00040893236,0.0003495588,0.00011365652],"domain_scores_gemma":[0.9982723,0.00079417194,0.00024648447,0.00021619115,0.00031472463,0.00015610727],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00232113,0.002008013,0.0013161747,0.004947561,0.001647605,0.0024107115,0.0023825772,0.0014574364,0.0063578477],"category_scores_gemma":[0.0074699884,0.0009284127,0.0015934889,0.003727408,0.0008371147,0.002144867,0.0020627298,0.0012833592,0.0035603482],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0015857575,0.00042291338,0.021844586,0.0029796353,0.00084965245,0.0010205894,0.0019841325,0.03626695,0.056026775,0.023154734,0.07842453,0.77543974],"study_design_scores_gemma":[0.00042689085,0.0004040288,0.007498662,0.00037431286,0.00028622485,0.002288095,0.0012326713,0.80953866,0.029391073,0.06300107,0.0854008,0.00015745408],"about_ca_topic_score_codex":0.003379939,"about_ca_topic_score_gemma":0.0057381387,"teacher_disagreement_score":0.0063578477,"about_ca_system_score_codex":0.000776754,"about_ca_system_score_gemma":0.0023072546,"threshold_uncertainty_score":0.021269143},"labels":[],"label_agreement":null},{"id":"W2163262592","doi":"10.1093/bioinformatics/btp701","title":"Pandora, a PAthway and Network DiscOveRy Approach based on common biological evidence","year":2009,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia; Ontario Institute for Cancer Research","funders":"Michael Smith Health Research BC; Canadian Institutes of Health Research; Government of Ontario; Ontario Institute for Cancer Research","keywords":"Biological network; Biological pathway; KEGG; Computer science; Computational biology; Biological data; Systems biology; Biological database; Perl; Gene ontology; Ontology; UniProt; Domain (mathematical analysis); Biology; Bioinformatics; Gene; Genetics; World Wide Web","score_opus":0.021563019755216208,"score_gpt":0.2361732331037945,"score_spread":0.2146102133485783,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2163262592","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.042974636,0.0019942613,0.9202984,0.00061904365,0.00009424639,0.0006335915,0.013040418,0.015357244,0.0049881404],"genre_scores_gemma":[0.15809342,0.0008055171,0.82374203,0.00011698457,0.000063180705,0.00054666953,0.014473689,0.0003738255,0.0017848152],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.999172,0.00019312584,0.00004904534,0.00032840224,0.00020833546,0.000049186656],"domain_scores_gemma":[0.9978009,0.0012808454,0.00022966941,0.000224301,0.00032237213,0.0001418634],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014835064,0.0012168196,0.0009092319,0.009888769,0.0011982026,0.0018728131,0.0012632194,0.0006922646,0.0071905022],"category_scores_gemma":[0.0054462985,0.0005958286,0.0019196768,0.0034934378,0.00053172326,0.001713912,0.0018989777,0.0009003498,0.0008747064],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0012849708,0.0005040377,0.042035103,0.003778463,0.001906641,0.0024444659,0.0005571803,0.11079972,0.022998169,0.05732106,0.026190437,0.73017967],"study_design_scores_gemma":[0.0001671975,0.00020110846,0.009374093,0.00027555617,0.00072238455,0.0017295983,0.00022990402,0.8641571,0.008344559,0.076739684,0.03796766,0.00009121208],"about_ca_topic_score_codex":0.0043889596,"about_ca_topic_score_gemma":0.009396606,"teacher_disagreement_score":0.009888769,"about_ca_system_score_codex":0.0007124718,"about_ca_system_score_gemma":0.0025905482,"threshold_uncertainty_score":0.024054587},"labels":[],"label_agreement":null},{"id":"W2163587256","doi":"10.1093/bioinformatics/btr079","title":"Automated workflows for accurate mass-based putative metabolite identification in LC/MS-derived metabolomic datasets","year":2011,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Metabolomics and Mass Spectrometry Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":193,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Biotechnology and Biological Sciences Research Council; Directorate for Biological Sciences; National Institute for Health and Care Research; Cancer Research UK; Wellcome Trust","keywords":"Metabolomics; Workflow; Computer science; Identification (biology); Matching (statistics); Metabolite; Annotation; Software; Artifact (error); Data mining; False positive paradox; Limiting; Computational biology; Chemistry; Database; Artificial intelligence; Chromatography; Biology; Mathematics","score_opus":0.030480199805048347,"score_gpt":0.28736586061543684,"score_spread":0.2568856608103885,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2163587256","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.014317691,0.00026918665,0.770788,0.0004966593,0.00010314006,0.0019394845,0.021729173,0.18907471,0.0012818781],"genre_scores_gemma":[0.030283125,0.00021444645,0.9242318,0.00026672057,0.000057015026,0.002249615,0.037118066,0.004537349,0.0010418119],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99427754,0.0012589891,0.0011841763,0.0012918478,0.0016527256,0.00033468698],"domain_scores_gemma":[0.98151577,0.0071007176,0.0023641896,0.003974572,0.0041872403,0.0008574429],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0145661775,0.002377869,0.0016419151,0.005114336,0.0017299503,0.0049424674,0.0029071427,0.0012141707,0.009612179],"category_scores_gemma":[0.024739273,0.0011711279,0.002116091,0.00393865,0.0011089251,0.0026331262,0.0031790305,0.0019537823,0.010197954],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.007220695,0.0011627756,0.017600818,0.0034740563,0.0007723984,0.0018385769,0.0016387805,0.020056099,0.20728067,0.010124561,0.12296865,0.60586196],"study_design_scores_gemma":[0.002198021,0.00067356485,0.03277755,0.00067574147,0.0003088611,0.0016145689,0.00061695126,0.36153287,0.45030004,0.03826266,0.110084906,0.0009542569],"about_ca_topic_score_codex":0.0037129086,"about_ca_topic_score_gemma":0.003708544,"teacher_disagreement_score":0.0145661775,"about_ca_system_score_codex":0.0020513039,"about_ca_system_score_gemma":0.0049412264,"threshold_uncertainty_score":0.077034235},"labels":[],"label_agreement":null},{"id":"W2163649997","doi":"10.1093/bioinformatics/btl080","title":"MACGT: multi-dimensional automated clustering genotyping tool for analysis of microarray-based mini-sequencing data","year":2006,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"St. Paul's Hospital; University of British Columbia","funders":"Heart and Stroke Foundation of British Columbia and Yukon; National Sanitarium Association; Heart and Stroke Foundation of Canada","keywords":"Genotyping; Cluster analysis; SNP; SNP genotyping; Computer science; DNA microarray; Single-nucleotide polymorphism; Computational biology; Genotype; SNP array; Data mining; Biology; Artificial intelligence; Genetics","score_opus":0.04516803501496724,"score_gpt":0.3026467413629859,"score_spread":0.25747870634801867,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2163649997","genre_codex":"software","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.006948662,0.00028074003,0.40189278,0.0003069493,0.00030977756,0.0006182289,0.15527119,0.43132412,0.003047556],"genre_scores_gemma":[0.025009153,0.00022331985,0.6968625,0.00064021675,0.00011063393,0.003986951,0.20043065,0.065413356,0.007323169],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99753076,0.000454628,0.0002922353,0.00076039095,0.0007627896,0.00019917163],"domain_scores_gemma":[0.9957176,0.0018160637,0.0006772228,0.0007697162,0.0007538477,0.0002656609],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004546317,0.0024063762,0.001800926,0.003946925,0.0011110139,0.0018158228,0.00327473,0.0013119596,0.06787442],"category_scores_gemma":[0.0095178,0.0015816657,0.0017708341,0.0027069396,0.0004600085,0.0015998428,0.0018528133,0.0021244658,0.03595518],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014712847,0.00015190808,0.0045095584,0.0018237005,0.00040913044,0.00045551767,0.00042971654,0.0049224445,0.056435548,0.0040652365,0.79973114,0.1255948],"study_design_scores_gemma":[0.0014268763,0.0006019543,0.027310412,0.0006728026,0.00040935387,0.0019846316,0.0002471503,0.17867549,0.19396336,0.026494266,0.56736964,0.00084408553],"about_ca_topic_score_codex":0.0022689656,"about_ca_topic_score_gemma":0.0040199705,"teacher_disagreement_score":0.06787442,"about_ca_system_score_codex":0.0010835801,"about_ca_system_score_gemma":0.001837408,"threshold_uncertainty_score":0.22706252},"labels":[],"label_agreement":null},{"id":"W2163822209","doi":"10.1093/bioinformatics/bth913","title":"Striped sheets and protein contact prediction","year":2004,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Protein Structure and Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":89,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Knut och Alice Wallenbergs Stiftelse; Stiftelsen för Strategisk Forskning; Physicians' Services Incorporated Foundation","keywords":"Genetic programming; Sequence (biology); Computer science; Set (abstract data type); Source code; Algorithm; Protein methods; Sequence analysis; Artificial intelligence; Data mining; Biology; Genetics","score_opus":0.004783633876989148,"score_gpt":0.20641157875567803,"score_spread":0.20162794487868888,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2163822209","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7547561,0.0030713824,0.2322363,0.00066404784,0.00005494932,0.00014295458,0.0012729021,0.0034789185,0.004322445],"genre_scores_gemma":[0.92870504,0.00061791437,0.06768749,0.00007743556,0.000051981675,0.000064349995,0.0016889335,0.00008767455,0.0010190392],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9993424,0.00023846602,0.000036061745,0.00013604175,0.00020699506,0.000039999562],"domain_scores_gemma":[0.9944326,0.003614125,0.00075773516,0.000376983,0.0005926441,0.00022593506],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013726393,0.000557386,0.00054837234,0.001941754,0.00044696147,0.00071267213,0.0005453902,0.0005511464,0.0018082871],"category_scores_gemma":[0.006655135,0.00020721322,0.0002469485,0.0017141993,0.0006473609,0.00090965733,0.0007300197,0.0005566741,0.0009184827],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014015598,0.00027073608,0.14010689,0.0007157859,0.00021118212,0.0007249676,0.00033828683,0.35931498,0.023354452,0.0058082123,0.006553545,0.46119937],"study_design_scores_gemma":[0.000035625555,0.00020695468,0.025884088,0.000067399356,0.000027676922,0.00067589147,0.00008894577,0.9279247,0.019639775,0.023114031,0.0023078183,0.000027130885],"about_ca_topic_score_codex":0.00088105525,"about_ca_topic_score_gemma":0.0006303756,"teacher_disagreement_score":0.001941754,"about_ca_system_score_codex":0.00024610257,"about_ca_system_score_gemma":0.00032667487,"threshold_uncertainty_score":0.0072592497},"labels":[],"label_agreement":null},{"id":"W2163980860","doi":"10.1093/bioinformatics/bti493","title":"Discovering patterns to extract protein–protein interactions from the literature: Part II","year":2005,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Biomedical Text Mining and Ontologies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":96,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"Natural Sciences and Engineering Research Council of Canada; Killam Trusts","keywords":"Merge (version control); Computer science; Recall; Protein–protein interaction; Generalization; Precision and recall; Protein expression; Artificial intelligence; Task (project management); Machine learning; Data mining; Computational biology; Information retrieval; Biology; Genetics; Mathematics","score_opus":0.015103357633382478,"score_gpt":0.26207536494041134,"score_spread":0.24697200730702887,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2163980860","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.15318106,0.0035386435,0.82659143,0.0022594861,0.00012559339,0.0011320775,0.005823164,0.0039294343,0.0034191243],"genre_scores_gemma":[0.16063589,0.0018094871,0.82468253,0.00029072815,0.00016040151,0.00087215164,0.009035481,0.00019808135,0.0023152656],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9985436,0.00030309812,0.00029742194,0.0004024203,0.00037082753,0.00008267336],"domain_scores_gemma":[0.99719834,0.0014979693,0.00038027964,0.00042109683,0.00037260354,0.00012972135],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016087858,0.0010632095,0.00082314375,0.0069341017,0.0007705575,0.0019603923,0.0012879303,0.0009373401,0.0027673116],"category_scores_gemma":[0.00756898,0.00054148363,0.0011379122,0.0064156163,0.0009238071,0.002793149,0.0015196963,0.00083912956,0.0015565257],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00023105262,0.0005541327,0.022902839,0.0015960323,0.00018224887,0.0007138824,0.0003979099,0.007193329,0.046551745,0.0049326466,0.008889494,0.90585476],"study_design_scores_gemma":[0.00039714776,0.00095811236,0.08505093,0.0006219078,0.00092431426,0.008364484,0.0016649974,0.5490792,0.19139782,0.087517135,0.07377168,0.00025230722],"about_ca_topic_score_codex":0.0015243152,"about_ca_topic_score_gemma":0.002191204,"teacher_disagreement_score":0.0069341017,"about_ca_system_score_codex":0.00049525616,"about_ca_system_score_gemma":0.0016714146,"threshold_uncertainty_score":0.009257615},"labels":[],"label_agreement":null},{"id":"W2163982822","doi":"10.1093/bioinformatics/btu198","title":"Inferring rare disease risk variants based on exact probabilities of sharing by multiple affected relatives","year":2014,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":32,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Institut Universitaire en Santé Mentale de Québec","funders":"National Institute of Dental and Craniofacial Research; National Institute of General Medical Sciences; National Human Genome Research Institute","keywords":"Kinship; Genetics; Exome sequencing; Population; Biology; Computer science; Mutation; Medicine; Gene","score_opus":0.010034335280250776,"score_gpt":0.2302433048776615,"score_spread":0.22020896959741074,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2163982822","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.25922054,0.0004178762,0.7368685,0.0002159795,0.000025549656,0.000089314264,0.0010940857,0.00092139776,0.0011467681],"genre_scores_gemma":[0.8363743,0.00038522368,0.160437,0.00013751,0.00006412354,0.00019054484,0.0013826501,0.00021233008,0.00081633014],"study_design_codex":"observational","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9950375,0.0029343849,0.00032111866,0.00118872,0.00036585814,0.00015246365],"domain_scores_gemma":[0.94774795,0.046137515,0.002768072,0.0025599673,0.0004353011,0.0003513015],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.009441359,0.00073167076,0.0014184478,0.0020247677,0.0005916587,0.0010917026,0.0015833526,0.0011901497,0.0044727325],"category_scores_gemma":[0.058725607,0.0006514231,0.0012978246,0.0021561156,0.0014830797,0.0014597449,0.0015138729,0.0011299761,0.00075666135],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013175759,0.00015937281,0.4791268,0.0007748566,0.0015163355,0.0039905454,0.0014819626,0.34679565,0.008738396,0.024762154,0.0024706246,0.12886575],"study_design_scores_gemma":[0.0001970287,0.00026482056,0.07355613,0.00020802561,0.0006067679,0.005463407,0.00032278933,0.7838728,0.006331066,0.12611209,0.0029404478,0.00012470562],"about_ca_topic_score_codex":0.004708996,"about_ca_topic_score_gemma":0.0027455806,"teacher_disagreement_score":0.009441359,"about_ca_system_score_codex":0.0004798556,"about_ca_system_score_gemma":0.00067411305,"threshold_uncertainty_score":0.049931288},"labels":[],"label_agreement":null},{"id":"W2164555431","doi":"10.1093/bioinformatics/btm532","title":"Uniformization for sampling realizations of Markov processes: applications to Bayesian implementations of codon substitution models","year":2007,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":70,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal; Canadian Institute for Advanced Research","funders":"Canadian Institutes of Health Research; Centre National de la Recherche Scientifique; Canadian Institute for Advanced Research","keywords":"Computer science; Markov chain Monte Carlo; Uniformization (probability theory); Substitution (logic); Tree (set theory); Bayesian probability; Algorithm; Probabilistic logic; Theoretical computer science; Pruning; Range (aeronautics); Markov chain; Markov model; Mathematics; Artificial intelligence; Variable-order Markov model; Machine learning; Combinatorics; Programming language","score_opus":0.033327972746494273,"score_gpt":0.31336914141959965,"score_spread":0.28004116867310536,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2164555431","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.004841287,0.000064511136,0.99448353,0.000055490655,0.0000066674916,0.0000333027,0.000020715637,0.00019689137,0.00029755945],"genre_scores_gemma":[0.20779209,0.0002667347,0.78995746,0.00011531052,0.000062868006,0.00047645948,0.0002905693,0.00023752061,0.0008009356],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9967892,0.0021740359,0.00016305913,0.00031330425,0.00042197143,0.00013854071],"domain_scores_gemma":[0.9826188,0.013965361,0.0007087942,0.0014871998,0.0008890453,0.00033075933],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.010106636,0.00067907624,0.0009668666,0.0011240772,0.0008134867,0.0013606228,0.0023604196,0.0014108751,0.0038617584],"category_scores_gemma":[0.042734016,0.0008693231,0.0008588953,0.0014605834,0.0017660006,0.0026765857,0.0025269026,0.0023185937,0.0007802269],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002494801,0.00012594863,0.0033548991,0.0001353411,0.000073685085,0.00015537339,0.00045830262,0.52208424,0.002883429,0.35656515,0.0015383288,0.112375885],"study_design_scores_gemma":[0.000030150444,0.000015853599,0.0001463986,0.000017271246,0.0000056806007,0.00003119644,0.000010244748,0.9438704,0.00066533056,0.05462309,0.0005728784,0.000011373023],"about_ca_topic_score_codex":0.0037381575,"about_ca_topic_score_gemma":0.0042785495,"teacher_disagreement_score":0.010106636,"about_ca_system_score_codex":0.0013886833,"about_ca_system_score_gemma":0.0016546249,"threshold_uncertainty_score":0.05344963},"labels":[],"label_agreement":null},{"id":"W2164620224","doi":"10.1093/bioinformatics/btm115","title":"SherLoc: high-accuracy prediction of protein subcellular localization by integrating text and protein sequence data","year":2007,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Machine Learning in Bioinformatics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":127,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Queen's University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Computer science; Sequence (biology); Protein sequencing; Artificial intelligence; Computational biology; Protein subcellular localization prediction; Data mining; Pattern recognition (psychology); Peptide sequence; Biology; Genetics; Gene","score_opus":0.015547919242480267,"score_gpt":0.25874349043564987,"score_spread":0.2431955711931696,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2164620224","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.33141416,0.0045454875,0.416853,0.0016227255,0.000380685,0.00035605472,0.089418136,0.15120523,0.0042045303],"genre_scores_gemma":[0.4526826,0.0010915435,0.40556142,0.0003850624,0.00026405507,0.00038311703,0.13390301,0.0015596813,0.0041695912],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9992455,0.00018219133,0.0000674243,0.00023099266,0.00022068711,0.00005312826],"domain_scores_gemma":[0.99505156,0.00267356,0.00067061867,0.00062482664,0.00075221993,0.00022723275],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0017720155,0.0017587987,0.0009770418,0.0054067513,0.00053872017,0.0011423837,0.0014408188,0.0015333256,0.0035455127],"category_scores_gemma":[0.006913758,0.00031712037,0.0007025228,0.0033675279,0.00038880965,0.002176547,0.0013488067,0.0008123473,0.0042564846],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0024183828,0.0006540655,0.06667293,0.0021546488,0.0006075591,0.0013007956,0.00027353902,0.08322464,0.046765227,0.0024499092,0.1727264,0.620752],"study_design_scores_gemma":[0.0002471506,0.00036443784,0.016279304,0.000085588756,0.00013358,0.0010082393,0.0001248814,0.9176593,0.039861448,0.006826465,0.01730588,0.000103802486],"about_ca_topic_score_codex":0.0018029208,"about_ca_topic_score_gemma":0.003486579,"teacher_disagreement_score":0.0054067513,"about_ca_system_score_codex":0.0005050951,"about_ca_system_score_gemma":0.0007353626,"threshold_uncertainty_score":0.011860967},"labels":[],"label_agreement":null},{"id":"W2164803230","doi":"10.1093/bioinformatics/bth142","title":"Fuzzy J-Means and VNS methods for clustering genes from microarray data","year":2004,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":70,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"National Research Council Canada","funders":"Université de Moncton","keywords":"Cluster analysis; Microarray analysis techniques; Computer science; Fuzzy logic; Data mining; Microarray; Computational biology; Gene; Artificial intelligence; Biology; Genetics; Gene expression","score_opus":0.048461004433985974,"score_gpt":0.3445911649265163,"score_spread":0.2961301604925303,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2164803230","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0032681667,0.00039713318,0.9951025,0.00011169803,0.000053583793,0.00007535146,0.00011468947,0.00034469785,0.0005321837],"genre_scores_gemma":[0.04184311,0.00037477433,0.95578,0.00007803616,0.000088480876,0.00036932732,0.00041004468,0.00010908345,0.00094707485],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9974915,0.00080653874,0.00021212564,0.00044681825,0.00095953466,0.00008352318],"domain_scores_gemma":[0.9969434,0.001785167,0.00023251225,0.0002560324,0.0007184175,0.0000644372],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0040010894,0.0009824473,0.0014018187,0.0037485764,0.00086790445,0.0012621592,0.0020441718,0.0011794296,0.0020157201],"category_scores_gemma":[0.010082007,0.000517504,0.0015648304,0.0040608337,0.0010954322,0.0008901173,0.0007773823,0.0013415745,0.0008827561],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002161151,0.00007603882,0.0013212097,0.00040584506,0.0002390151,0.000066134526,0.00029616678,0.45340884,0.0032473286,0.02760524,0.0045544393,0.5085637],"study_design_scores_gemma":[0.000025169811,0.00002663215,0.0006491464,0.00003498811,0.000021557049,0.00004512533,0.00004341205,0.9710296,0.0010638067,0.024230812,0.0028016886,0.000028083401],"about_ca_topic_score_codex":0.0103967,"about_ca_topic_score_gemma":0.011024499,"teacher_disagreement_score":0.0103967,"about_ca_system_score_codex":0.0014047206,"about_ca_system_score_gemma":0.0018956601,"threshold_uncertainty_score":0.021160066},"labels":[],"label_agreement":null},{"id":"W2165260680","doi":"10.1093/bioinformatics/bts228","title":"Identifying disease sensitive and quantitative trait-relevant biomarkers from multidimensional heterogeneous imaging genetics data via sparse multimodal multitask learning","year":2012,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":159,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Institute on Aging; University of California, San Diego; National Institute of Biomedical Imaging and Bioengineering; Canadian Institutes of Health Research; University of California, Los Angeles; Genentech; National Institutes of Health; Servier; Eisai; Northern California Institute for Research and Education; Pfizer; Biogen; BioClinica; Alzheimer's Association; Amorfix Life Sciences; Dana Foundation; Bayer HealthCare; Alzheimer's Disease Neuroimaging Initiative; National Center for Advancing Translational Sciences; Meso Scale Diagnostics; National Center for Research Resources; F. Hoffmann-La Roche; Medpace; Bristol-Myers Squibb; Eli Lilly and Company; AstraZeneca; Novartis Pharmaceuticals Corporation; Synarc; Foundation for the National Institutes of Health; National Science Foundation","keywords":"Imaging genetics; Neuroimaging; Cognition; Computer science; Machine learning; Artificial intelligence; Alzheimer's Disease Neuroimaging Initiative; Pairwise comparison; Multi-task learning; Psychology; Neuroscience; Cognitive impairment; Task (project management)","score_opus":0.0339192501247622,"score_gpt":0.2984632222230629,"score_spread":0.26454397209830066,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2165260680","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.032289654,0.0003508821,0.96255386,0.0013742681,0.000024400939,0.000064453234,0.001404157,0.001351615,0.00058670144],"genre_scores_gemma":[0.52836657,0.0006169167,0.46286646,0.0005202411,0.00015897765,0.00037568458,0.0044942857,0.00028267683,0.0023181331],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99937457,0.0002948011,0.000029811274,0.00016482965,0.000096317184,0.00003972402],"domain_scores_gemma":[0.9974101,0.0017763915,0.00026938287,0.00018476948,0.00023595354,0.00012335152],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0030137654,0.0008957991,0.00079250627,0.0012682484,0.00034768638,0.0008119212,0.0010622704,0.0009176903,0.0020365482],"category_scores_gemma":[0.009165399,0.00032878292,0.0006579383,0.0011546216,0.0007263203,0.00077042315,0.0013211734,0.0012162003,0.00081514457],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005439592,0.0004321693,0.022654215,0.00057651155,0.00033402687,0.0005507902,0.00021571516,0.6764852,0.013203458,0.022607984,0.018909523,0.24348651],"study_design_scores_gemma":[0.000027078071,0.000024893394,0.0020919675,0.0000161078,0.000022701388,0.00007303964,0.000017880382,0.96800494,0.0016412524,0.026947265,0.0011176822,0.000015194463],"about_ca_topic_score_codex":0.0040245955,"about_ca_topic_score_gemma":0.0043133385,"teacher_disagreement_score":0.0040245955,"about_ca_system_score_codex":0.00060410023,"about_ca_system_score_gemma":0.0010653442,"threshold_uncertainty_score":0.01593846},"labels":[],"label_agreement":null},{"id":"W2165474022","doi":"10.1093/bioinformatics/btn165","title":"Designing succinct structural alphabets","year":2008,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Protein Structure and Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"Chinese Government Scholarship; Natural Sciences and Engineering Research Council of Canada; Mitacs; National Natural Science Foundation of China","keywords":"Sequence (biology); Computer science; Construct (python library); State (computer science); Algorithm; Programming language; Chemistry","score_opus":0.009531432337122883,"score_gpt":0.22692665459427547,"score_spread":0.21739522225715258,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2165474022","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.08898649,0.0006348155,0.8904444,0.0005548652,0.00020669062,0.00027778855,0.0021754429,0.004917451,0.011802096],"genre_scores_gemma":[0.3568824,0.0007284094,0.6314175,0.00041495345,0.00010603071,0.00059408223,0.0046694065,0.00097356294,0.0042137173],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9984395,0.00041334872,0.00021834244,0.0002991391,0.0004521719,0.00017761326],"domain_scores_gemma":[0.9941936,0.002974597,0.00049212977,0.0012288709,0.0008385448,0.00027233976],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001508624,0.00063910615,0.0011265465,0.0017092956,0.00088173134,0.0016638214,0.0017985043,0.0009680557,0.0058655385],"category_scores_gemma":[0.008761628,0.000830665,0.00072879146,0.001910405,0.0010771111,0.0036545158,0.0018656123,0.0014493611,0.002036327],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010107575,0.00052240444,0.0064783343,0.001569749,0.000079442056,0.0009499985,0.0009338503,0.24654894,0.051469784,0.24043478,0.020591155,0.42941085],"study_design_scores_gemma":[0.0001302148,0.00030781113,0.00061149854,0.0002332532,0.00006351975,0.00050300546,0.00031047128,0.7170136,0.0484871,0.19789654,0.03437093,0.00007205134],"about_ca_topic_score_codex":0.0005936821,"about_ca_topic_score_gemma":0.0013449404,"teacher_disagreement_score":0.0058655385,"about_ca_system_score_codex":0.0008476707,"about_ca_system_score_gemma":0.0014419897,"threshold_uncertainty_score":0.019622207},"labels":[],"label_agreement":null},{"id":"W2165567081","doi":"10.1093/bioinformatics/16.5.484","title":"Viral Genome DataBase: storing and analyzing genes and proteins from complete viral genomes","year":2000,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Animal Virus Infections Studies","field":"Agricultural and Biological Sciences","cited_by":30,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Victoria","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"GenBank; Genome; Gene; Biology; Database; Codon usage bias; Genetics; DNA sequencing; Computational biology; Computer science","score_opus":0.03443914197597459,"score_gpt":0.22560039761033884,"score_spread":0.19116125563436426,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2165567081","genre_codex":"dataset","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.02404136,0.0015951063,0.09237754,0.00048949255,0.00017508,0.0004848473,0.77541184,0.094528005,0.010896707],"genre_scores_gemma":[0.016226508,0.0005545444,0.053872738,0.00011226741,0.000032582342,0.00022239804,0.9243123,0.0025119479,0.0021547677],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99953103,0.00006160327,0.00008065551,0.00014295611,0.00012138453,0.000062315405],"domain_scores_gemma":[0.9991732,0.0001480444,0.000101610174,0.00018522078,0.00023428103,0.00015763048],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011903719,0.0011652436,0.0011589343,0.003427626,0.0007589044,0.0018569573,0.0022903876,0.0007707392,0.016650628],"category_scores_gemma":[0.0025197961,0.00063342357,0.00057623425,0.005630591,0.00024308565,0.0017777297,0.0015811883,0.00095074036,0.020483015],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.003274041,0.00047096916,0.010617148,0.004037226,0.00028622962,0.00054262264,0.0004583392,0.004113362,0.07195298,0.010627466,0.70777726,0.18584244],"study_design_scores_gemma":[0.0009795482,0.00040498393,0.026005976,0.00048466143,0.00034222068,0.001550246,0.00035281532,0.026375053,0.087077186,0.016759379,0.8394602,0.00020778501],"about_ca_topic_score_codex":0.003483964,"about_ca_topic_score_gemma":0.0023667724,"teacher_disagreement_score":0.016650628,"about_ca_system_score_codex":0.0005257166,"about_ca_system_score_gemma":0.0013943067,"threshold_uncertainty_score":0.05570185},"labels":[],"label_agreement":null},{"id":"W2166098531","doi":"10.1093/bioinformatics/bts701","title":"Characterization of disordered proteins with ENSEMBLE","year":2012,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Protein Structure and Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":157,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"SickKids Foundation; Hospital for Sick Children; University of Toronto","funders":"Canadian Institutes of Health Research","keywords":"Perl; Computer science; Characterization (materials science); Interface (matter); Set (abstract data type); Ensemble learning; Intrinsically disordered proteins; Data mining; Programming language; Artificial intelligence; Physics; Operating system; Nanotechnology; Materials science","score_opus":0.005304317679989394,"score_gpt":0.2036163241529141,"score_spread":0.1983120064729247,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2166098531","genre_codex":"empirical","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6480546,0.0010170025,0.3398669,0.00019298904,0.000072061586,0.00008286166,0.00382688,0.002719139,0.004167647],"genre_scores_gemma":[0.90002507,0.00033186522,0.091276065,0.00007021551,0.00004332606,0.00007520333,0.0069054277,0.0004425302,0.0008302808],"study_design_codex":"simulation_or_modeling","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9995951,0.000058534282,0.000029420144,0.00013086417,0.00013924985,0.000046806705],"domain_scores_gemma":[0.9989073,0.00031354302,0.00017807817,0.00022914984,0.00023203435,0.00013988143],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00072639994,0.00043550122,0.00056351203,0.0014408715,0.00047978794,0.0007719013,0.00063748675,0.00043364658,0.0021091376],"category_scores_gemma":[0.0025803605,0.00019860452,0.0005108322,0.0011333158,0.00038598283,0.001089675,0.0007219844,0.0005230615,0.0005856725],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0018108787,0.0003254559,0.10935914,0.0010375198,0.000503982,0.0011411989,0.0006501218,0.34171176,0.2661797,0.027978564,0.011466793,0.23783483],"study_design_scores_gemma":[0.00001444391,0.000083922954,0.010866525,0.000025058864,0.000042968757,0.00043429036,0.00006491313,0.9372254,0.03712763,0.010310304,0.0037715735,0.00003308048],"about_ca_topic_score_codex":0.000712843,"about_ca_topic_score_gemma":0.00089803134,"teacher_disagreement_score":0.0021091376,"about_ca_system_score_codex":0.00037394013,"about_ca_system_score_gemma":0.00027556272,"threshold_uncertainty_score":0.0070557594},"labels":[],"label_agreement":null},{"id":"W2166268906","doi":"10.1093/bioinformatics/btp212","title":"A partition function algorithm for interacting nucleic acid strands","year":2009,"lang":"en","type":"article","venue":"Bioinformatics","topic":"DNA and Nucleic Acid Chemistry","field":"Biochemistry, Genetics and Molecular Biology","cited_by":104,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Algorithm; Nucleic acid; Partition (number theory); Computer science; Function (biology); Mathematics; Combinatorics; Chemistry; Biology; Biochemistry; Genetics","score_opus":0.00877055586134036,"score_gpt":0.24114521534579478,"score_spread":0.2323746594844544,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2166268906","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.028283762,0.0002416691,0.96814775,0.0002890873,0.000043387685,0.00009291025,0.00013721545,0.00057464186,0.002189572],"genre_scores_gemma":[0.28171784,0.00021430117,0.71099585,0.00017361341,0.00008471965,0.0006692494,0.0008025939,0.00049451255,0.004847356],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.999481,0.0001903542,0.000022978611,0.00011419046,0.00011610291,0.00007536969],"domain_scores_gemma":[0.99683917,0.002429477,0.0001491833,0.00011042444,0.0003557732,0.00011606379],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011388243,0.0011171023,0.0010857907,0.0013447604,0.00082897604,0.0011883362,0.002186221,0.0018585057,0.0063042687],"category_scores_gemma":[0.0054730265,0.00044701365,0.000875635,0.0010071343,0.0010619335,0.0015499634,0.0015986062,0.0017753738,0.0011583484],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00015743665,0.000101606,0.00096353114,0.00009470034,0.000048358284,0.0000742994,0.00013560489,0.87735,0.0013641343,0.03077279,0.0034007565,0.085536875],"study_design_scores_gemma":[0.000009410398,0.0000108872255,0.000041760046,0.0000047951853,0.0000024873395,0.000009759246,0.000006267093,0.98728704,0.00016128586,0.012164062,0.00029922687,0.000002940876],"about_ca_topic_score_codex":0.0035388295,"about_ca_topic_score_gemma":0.0029403975,"teacher_disagreement_score":0.0063042687,"about_ca_system_score_codex":0.0014807435,"about_ca_system_score_gemma":0.001319335,"threshold_uncertainty_score":0.021089852},"labels":[],"label_agreement":null},{"id":"W2166281646","doi":"10.1093/bioinformatics/bth019","title":"Transposable element annotation of the rice genome","year":2004,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Chromosomal and Genetic Variations","field":"Agricultural and Biological Sciences","cited_by":62,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"","keywords":"Annotation; Transposable element; Genome; Computer science; Perl; Set (abstract data type); Computational biology; Biology; Sequence alignment; Genetics; Programming language; Gene","score_opus":0.013399751487314138,"score_gpt":0.19352560113012673,"score_spread":0.18012584964281259,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2166281646","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.5860684,0.0041007935,0.17895044,0.0005674859,0.00024732965,0.00051613076,0.19499019,0.014395874,0.020163313],"genre_scores_gemma":[0.32277712,0.0014130444,0.3643044,0.0001725087,0.00006390393,0.00025411,0.2976627,0.0016263025,0.011725928],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9998099,0.0000163195,0.000018908251,0.00008834867,0.00004031568,0.000026163905],"domain_scores_gemma":[0.99961036,0.000055114655,0.000036498914,0.00005264798,0.00019529111,0.000050180308],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00030263362,0.00076902396,0.00045171496,0.0012619038,0.00044031907,0.00048367892,0.0004588191,0.00026139087,0.004665012],"category_scores_gemma":[0.00084110897,0.00016575183,0.00056645577,0.0020320418,0.0001594185,0.00031467248,0.0004568913,0.0005122447,0.0042007477],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0020866932,0.00015639933,0.025495293,0.0021811028,0.00009391386,0.0012188939,0.0011422974,0.0070922226,0.70622367,0.0029237375,0.016129138,0.23525661],"study_design_scores_gemma":[0.00024432718,0.0006521266,0.21074596,0.00040252905,0.00038728456,0.0032101062,0.0007478309,0.041605983,0.41198662,0.005250835,0.32460186,0.00016448733],"about_ca_topic_score_codex":0.008155257,"about_ca_topic_score_gemma":0.007051131,"teacher_disagreement_score":0.008155257,"about_ca_system_score_codex":0.0006393742,"about_ca_system_score_gemma":0.0017486188,"threshold_uncertainty_score":0.016215563},"labels":[],"label_agreement":null},{"id":"W2166335393","doi":"10.1093/bioinformatics/btp222","title":"Constrained mixture estimation for analysis and robust classification of clinical time series","year":2009,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Time Series Analysis and Forecasting","field":"Computer Science","cited_by":37,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"Division of Mathematical Sciences; Coordenação de Aperfeiçoamento de Pessoal de Nível Superior","keywords":"Computer science; Robustness (evolution); Personalized medicine; Machine learning; Data mining; Artificial intelligence; Missing data; Pattern recognition (psychology); Bioinformatics; Gene; Biology","score_opus":0.04136910599703127,"score_gpt":0.29435331925133573,"score_spread":0.2529842132543045,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2166335393","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.001613542,0.00018822463,0.9968405,0.00013069696,0.000026468633,0.0000407429,0.00016266866,0.0008558007,0.00014141337],"genre_scores_gemma":[0.18664384,0.000735618,0.8043367,0.00036683935,0.0002566668,0.0006392026,0.0028579875,0.00072264654,0.003440369],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.997898,0.0010074883,0.00013587532,0.00043669407,0.00039835938,0.00012358534],"domain_scores_gemma":[0.9931265,0.0051030824,0.0005023101,0.0005382401,0.00056655833,0.00016323464],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005241317,0.001482906,0.0014599611,0.0016709929,0.00052423193,0.0013622475,0.0020539283,0.0020472335,0.006121751],"category_scores_gemma":[0.021088468,0.0010432687,0.0021116352,0.0012912847,0.0011155875,0.0011825631,0.0021524008,0.0028768457,0.0034453818],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00053477,0.00014753762,0.0031637743,0.00035461393,0.00034947455,0.000205824,0.00017043551,0.72164816,0.006533308,0.022845842,0.007073568,0.2369727],"study_design_scores_gemma":[0.000007982631,0.000014715108,0.00023799132,0.000013871833,0.0000097462125,0.000018213117,0.000003654485,0.99166054,0.0005840474,0.0066086613,0.00082816154,0.000012478562],"about_ca_topic_score_codex":0.00880102,"about_ca_topic_score_gemma":0.006622676,"teacher_disagreement_score":0.00880102,"about_ca_system_score_codex":0.0009524576,"about_ca_system_score_gemma":0.0017362143,"threshold_uncertainty_score":0.02771908},"labels":[],"label_agreement":null},{"id":"W2166372712","doi":"10.1093/bioinformatics/btu544","title":"Figmop: a profile HMM to identify genes and bypass troublesome gene models in draft genomes","year":2014,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"","keywords":"Gene; Genome; Computational biology; Hidden Markov model; Biology; Genetics; Computer science; Artificial intelligence","score_opus":0.0148272516669068,"score_gpt":0.2479729135171589,"score_spread":0.2331456618502521,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2166372712","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.012991643,0.0004252103,0.61844504,0.00039657252,0.0002745355,0.00028721837,0.053449035,0.3105892,0.0031416153],"genre_scores_gemma":[0.07956033,0.000905865,0.76346046,0.00045732237,0.00010916599,0.0010992491,0.10125573,0.049471512,0.0036803475],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9993807,0.00016875076,0.00005561838,0.00021291364,0.000116718795,0.00006520406],"domain_scores_gemma":[0.9979073,0.001224595,0.00016164422,0.0003873234,0.0002037354,0.000115345334],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001989793,0.0015833041,0.0010186746,0.002162927,0.00095023867,0.001716729,0.0025547238,0.0018239147,0.030449077],"category_scores_gemma":[0.009321917,0.0014429649,0.0018328164,0.0021792068,0.00047808528,0.0028823442,0.0023339354,0.001781885,0.01715514],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0032748654,0.000309138,0.013986971,0.0064588496,0.00083842507,0.0016920379,0.0015613597,0.024180211,0.059478287,0.013783314,0.5208088,0.35362768],"study_design_scores_gemma":[0.00064293784,0.0004415102,0.011930776,0.0009240089,0.00046832435,0.0023243057,0.0006616385,0.4036697,0.06878282,0.04595287,0.4638168,0.0003844026],"about_ca_topic_score_codex":0.0019280647,"about_ca_topic_score_gemma":0.0035589477,"teacher_disagreement_score":0.030449077,"about_ca_system_score_codex":0.0005975754,"about_ca_system_score_gemma":0.0012902532,"threshold_uncertainty_score":0.10186225},"labels":[],"label_agreement":null},{"id":"W2166393666","doi":"10.1093/bioinformatics/btm145","title":"An efficient method for the detection and elimination of systematic error in high-throughput screening","year":2007,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Computational Drug Discovery Methods","field":"Computer Science","cited_by":98,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University and Génome Québec Innovation Centre; Université du Québec à Montréal","funders":"","keywords":"Throughput; Computer science; Error detection and correction; Systematic error; High-throughput screening; Statistics; Algorithm; Mathematics; Biology; Bioinformatics; Operating system","score_opus":0.03858466843007075,"score_gpt":0.34658664766903596,"score_spread":0.3080019792389652,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2166393666","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0028433842,0.00010226967,0.99524045,0.00006393839,0.000044229106,0.00006280908,0.00005151267,0.0014296138,0.00016185617],"genre_scores_gemma":[0.06362192,0.00014874872,0.93417186,0.00012094453,0.000065895765,0.00028710754,0.00031188302,0.0002644868,0.0010071069],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99100167,0.002415038,0.00037283712,0.0009959908,0.0049327463,0.00028173445],"domain_scores_gemma":[0.97611964,0.01244858,0.002141732,0.004557874,0.0042802407,0.0004518352],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005898529,0.0015833178,0.0019777054,0.0026817927,0.0007984518,0.0010786966,0.0032388591,0.0017910678,0.0021536045],"category_scores_gemma":[0.026584292,0.0007503912,0.0010113746,0.0024779956,0.001497488,0.0016568177,0.002214075,0.0026249404,0.001429968],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005850294,0.00039212682,0.006193343,0.00059891585,0.00035382097,0.0004922126,0.00024421996,0.11603242,0.10661651,0.02478859,0.007876412,0.73582643],"study_design_scores_gemma":[0.00010722621,0.0001926269,0.0021729602,0.000039921048,0.00008184769,0.0007421176,0.000020446962,0.9016894,0.07464921,0.013350622,0.0068387813,0.000114767696],"about_ca_topic_score_codex":0.0010574593,"about_ca_topic_score_gemma":0.0014407124,"teacher_disagreement_score":0.005898529,"about_ca_system_score_codex":0.0010042939,"about_ca_system_score_gemma":0.0026817124,"threshold_uncertainty_score":0.031194806},"labels":[],"label_agreement":null},{"id":"W2166542841","doi":"10.1093/bioinformatics/btq562","title":"GeneMANIA Cytoscape plugin: fast gene function predictions on the desktop","year":2010,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":661,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"Ontario Genomics; Ontario Genomics Institute; Genome Canada","keywords":"Plug-in; Computer science; Function (biology); Complement (music); Java; Set (abstract data type); Gene; Programming language; Biology; Genetics","score_opus":0.007931295299022024,"score_gpt":0.20052849950022197,"score_spread":0.19259720420119994,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2166542841","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0068485565,0.001205661,0.2302029,0.0010432061,0.0008122255,0.00031050385,0.09388879,0.6519182,0.01376996],"genre_scores_gemma":[0.102604404,0.002119033,0.43390164,0.0017320198,0.00059173163,0.002916627,0.31145084,0.09961162,0.045072168],"study_design_codex":"not_applicable","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99929035,0.00009848188,0.000034702854,0.00018467575,0.0003176934,0.000074154785],"domain_scores_gemma":[0.9981109,0.0007864132,0.00011417876,0.00032934936,0.00040858073,0.00025057682],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010578647,0.0020085683,0.0012907555,0.003010206,0.00094695744,0.0013874612,0.0037869792,0.0016233495,0.083523124],"category_scores_gemma":[0.0060010296,0.0011252756,0.001161151,0.0020619486,0.0004044086,0.0016526931,0.0015358223,0.0024278401,0.05946521],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00065193506,0.000110382396,0.0016979281,0.0012184572,0.00019441084,0.00039543578,0.00012964962,0.0029419507,0.01623827,0.00492838,0.82199186,0.14950132],"study_design_scores_gemma":[0.00052002753,0.000114850314,0.0057979994,0.00036930974,0.00019556552,0.0009801065,0.000075152566,0.09967504,0.097510606,0.036356058,0.7579823,0.00042280156],"about_ca_topic_score_codex":0.005217165,"about_ca_topic_score_gemma":0.007091139,"teacher_disagreement_score":0.083523124,"about_ca_system_score_codex":0.0006586026,"about_ca_system_score_gemma":0.0016762839,"threshold_uncertainty_score":0.2794127},"labels":[],"label_agreement":null},{"id":"W2166603052","doi":"10.1093/bioinformatics/btl513","title":"THOR: targeted high-throughput ortholog reconstructor","year":2006,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"BC Cancer Agency","funders":"Michael Smith Health Research BC; Genome Canada","keywords":"Sequence (biology); Genome; Computational biology; Throughput; Biology; Set (abstract data type); Whole genome sequencing; Computer science; Genetics; Gene","score_opus":0.0070013370118372584,"score_gpt":0.20645833400386795,"score_spread":0.1994569969920307,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2166603052","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.015668435,0.00021943213,0.5726398,0.0001275023,0.0001193363,0.00019523429,0.0129184695,0.39490104,0.0032106836],"genre_scores_gemma":[0.12828095,0.00039247295,0.770943,0.0003183919,0.00006602165,0.0007344611,0.05686252,0.0320276,0.010374649],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9993267,0.00010607374,0.000050858784,0.00021416105,0.00023367992,0.000068537636],"domain_scores_gemma":[0.9990163,0.00037412334,0.0001073537,0.00030790677,0.00012176829,0.00007260743],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015063466,0.001430457,0.0009454992,0.0013634287,0.0006552506,0.0012907493,0.0018627589,0.0010390689,0.022395719],"category_scores_gemma":[0.002541886,0.0011826537,0.0017123789,0.0009953172,0.00047945467,0.0010970968,0.0015565938,0.00174897,0.012004895],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0043155765,0.00032397395,0.0072498797,0.002249874,0.0008191192,0.001142717,0.00065647316,0.024102516,0.26172784,0.014858883,0.28660607,0.3959471],"study_design_scores_gemma":[0.0008152536,0.00042107835,0.0056572086,0.00013356324,0.00024099364,0.0019226493,0.00018786639,0.46059552,0.3503755,0.0135553265,0.16582322,0.00027175766],"about_ca_topic_score_codex":0.0014610055,"about_ca_topic_score_gemma":0.002148526,"teacher_disagreement_score":0.022395719,"about_ca_system_score_codex":0.00060444867,"about_ca_system_score_gemma":0.00091030536,"threshold_uncertainty_score":0.07492119},"labels":[],"label_agreement":null},{"id":"W2166815431","doi":"10.1093/bioinformatics/btu284","title":"A combinatorial approach for analyzing intra-tumor heterogeneity from high-throughput sequencing data","year":2014,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Cancer Genomics and Diagnostics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":124,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Burroughs Wellcome Fund; National Human Genome Research Institute; Natural Sciences and Engineering Research Council of Canada; National Cancer Institute; National Institutes of Health; National Science Foundation","keywords":"Cluster analysis; Python (programming language); Tumor heterogeneity; Population; DNA sequencing; Computer science; Biology; Computational biology; Single cell sequencing; Mutation; Algorithm; Genetics; Gene; Machine learning; Cancer; Exome sequencing","score_opus":0.026978806483223386,"score_gpt":0.2588377383390367,"score_spread":0.23185893185581335,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2166815431","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.009040999,0.0001472307,0.98899406,0.00018629445,0.000020905984,0.00009664791,0.00033245067,0.00043965847,0.0007416831],"genre_scores_gemma":[0.13843486,0.0002697904,0.8567298,0.00025335958,0.000116000556,0.00056352647,0.0018217852,0.00040103673,0.0014098195],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9975056,0.0011171988,0.00011260116,0.0005759846,0.0005309621,0.0001576925],"domain_scores_gemma":[0.98677117,0.010376248,0.0010734106,0.0006983937,0.0006398238,0.0004409909],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0041556004,0.0017311489,0.0019513903,0.0037414248,0.0013594228,0.0026291735,0.003047977,0.0018812249,0.003623572],"category_scores_gemma":[0.014492692,0.001369214,0.0023093962,0.0048665833,0.0018895998,0.002336217,0.002047858,0.002488476,0.000773372],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00018585584,0.00018941799,0.0034765403,0.00035565934,0.0002331288,0.00033739352,0.00012785666,0.88864416,0.0039615813,0.02829741,0.0036988663,0.07049211],"study_design_scores_gemma":[0.000026094445,0.000037529335,0.00060463033,0.000017382037,0.000035300804,0.0001138214,0.000026912865,0.9639616,0.0008468368,0.033299837,0.0010138833,0.000016176658],"about_ca_topic_score_codex":0.0037344424,"about_ca_topic_score_gemma":0.004687917,"teacher_disagreement_score":0.0041556004,"about_ca_system_score_codex":0.001999691,"about_ca_system_score_gemma":0.0023080288,"threshold_uncertainty_score":0.021977186},"labels":[],"label_agreement":null},{"id":"W2167050411","doi":"10.1093/bioinformatics/btp532","title":"Predicting homologous signaling pathways using machine learning","year":2009,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Computer science; Homologous chromosome; Artificial intelligence; Computational biology; Signal transduction; Machine learning; Biology; Cell biology; Genetics; Gene","score_opus":0.014601779393406436,"score_gpt":0.223485171631814,"score_spread":0.20888339223840757,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2167050411","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.4841394,0.002880873,0.48749033,0.0014156002,0.000113069786,0.00023899852,0.008444853,0.009260172,0.006016694],"genre_scores_gemma":[0.69840986,0.0009323139,0.28258777,0.0002378177,0.00009148652,0.00015879693,0.016167533,0.00013172276,0.0012826741],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99943227,0.00020558102,0.000044130473,0.00016062419,0.000117071315,0.000040292114],"domain_scores_gemma":[0.99741983,0.0015751678,0.0002940786,0.00022411956,0.00037061903,0.00011623555],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010616796,0.000954468,0.0005497799,0.0032536355,0.0007580753,0.00071117,0.000823057,0.00093623245,0.0023964697],"category_scores_gemma":[0.0030267788,0.00026285963,0.00084410777,0.0018494283,0.00043656715,0.0012815825,0.0005488516,0.00095590315,0.00090132526],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009267103,0.0012668681,0.09757046,0.0012235795,0.0002990188,0.0010518227,0.0001632609,0.3901995,0.06471407,0.0075253583,0.011865817,0.4231935],"study_design_scores_gemma":[0.000050833318,0.00014072024,0.012720498,0.000039698818,0.00005963649,0.0004406145,0.000055655215,0.93868643,0.022533242,0.02111991,0.0041254503,0.000027276195],"about_ca_topic_score_codex":0.0018581182,"about_ca_topic_score_gemma":0.0025155558,"teacher_disagreement_score":0.0032536355,"about_ca_system_score_codex":0.0006280904,"about_ca_system_score_gemma":0.00094934826,"threshold_uncertainty_score":0.008016944},"labels":[],"label_agreement":null},{"id":"W2167226024","doi":"10.1093/bioinformatics/btl292","title":"Analysis of binding site similarity, small-molecule similarity and experimental binding profiles in the human cytosolic sulfotransferase family","year":2007,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Monoclonal and Polyclonal Antibodies Research","field":"Medicine","cited_by":40,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Structural Genomics Consortium; University of Toronto","funders":"Canadian Institutes of Health Research; Wellcome Trust; Ontario Genomics Institute; Genome Canada","keywords":"Small molecule; Structural similarity; Binding site; Similarity (geometry); Sulfotransferase; Molecule; Plasma protein binding; Computational biology; Function (biology); Chemistry; Biology; Biochemistry; Genetics; Sulfation; Computer science","score_opus":0.053375137443588294,"score_gpt":0.34374316574124353,"score_spread":0.29036802829765523,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2167226024","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.989583,0.00042518484,0.008997503,0.000043576973,0.0000017082499,0.00001518802,0.0003963119,0.000096316784,0.00044122152],"genre_scores_gemma":[0.99180037,0.00013817336,0.006608654,0.0000186626,0.0000047120748,0.000019423644,0.0013235705,0.000013567572,0.000072885814],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99918884,0.0002854748,0.00007762536,0.00017946244,0.00022710575,0.000041358755],"domain_scores_gemma":[0.99193364,0.005305527,0.0016025208,0.00054930797,0.00038188373,0.00022707229],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016000018,0.00039165944,0.00071140355,0.0014558696,0.00037166916,0.00055922114,0.000559911,0.00051158573,0.0017705804],"category_scores_gemma":[0.006181089,0.00017427924,0.0005959396,0.0017472534,0.0007092282,0.000702774,0.00040575687,0.00044402914,0.00038456966],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0027346483,0.0009304332,0.3909969,0.0015784752,0.0009006593,0.0005740652,0.00026977577,0.14861359,0.40054953,0.0038057643,0.00052595953,0.048520204],"study_design_scores_gemma":[0.00009772207,0.0012156948,0.27599987,0.000037677084,0.00032558606,0.0019459012,0.0002182766,0.4955636,0.21521285,0.007699991,0.0015954016,0.00008745759],"about_ca_topic_score_codex":0.0010737026,"about_ca_topic_score_gemma":0.0010352107,"teacher_disagreement_score":0.0017705804,"about_ca_system_score_codex":0.00056320895,"about_ca_system_score_gemma":0.0003864228,"threshold_uncertainty_score":0.008461714},"labels":[],"label_agreement":null},{"id":"W2167272772","doi":"10.1093/bioinformatics/btm248","title":"Nucleotide composition string selection in HIV-1 subtyping using whole genomes","year":2007,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":52,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Subtyping; Phylogenetic tree; Genome; String (physics); Computational biology; Biology; Selection (genetic algorithm); Executable; Composition (language); Computer science; Genetics; Programming language; Gene; Artificial intelligence; Mathematics","score_opus":0.0150696711811774,"score_gpt":0.24437248272364961,"score_spread":0.22930281154247223,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2167272772","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.4176865,0.0003973986,0.57507545,0.00016497716,0.00003978764,0.00022901755,0.0007999507,0.0037804407,0.0018265377],"genre_scores_gemma":[0.4457042,0.00026776592,0.5494172,0.00012981067,0.000030642896,0.00021223449,0.002263171,0.00027538452,0.0016995907],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99942017,0.00027208327,0.00004728608,0.00007695202,0.00014857722,0.000034978442],"domain_scores_gemma":[0.99895155,0.0005632928,0.00009775524,0.00010933168,0.00020648832,0.0000715149],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010899277,0.0005512071,0.00078697875,0.001324152,0.0004283736,0.00052155106,0.0006510262,0.00034614914,0.0021507978],"category_scores_gemma":[0.0034716623,0.00024824476,0.0004251654,0.0014088892,0.00030720312,0.00057999056,0.0005979691,0.0004313551,0.00065887574],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001287829,0.00030050124,0.022228513,0.0004078873,0.00015646413,0.0003770162,0.00017318473,0.12713842,0.16169083,0.0040193642,0.0036273547,0.6785927],"study_design_scores_gemma":[0.00010505932,0.00027673505,0.010311045,0.000025133617,0.00007670545,0.00035601936,0.00007149762,0.90730894,0.06986102,0.008541259,0.0030265728,0.000040094328],"about_ca_topic_score_codex":0.0013914091,"about_ca_topic_score_gemma":0.0022969302,"teacher_disagreement_score":0.0021507978,"about_ca_system_score_codex":0.0003056632,"about_ca_system_score_gemma":0.0006063398,"threshold_uncertainty_score":0.0071951747},"labels":[],"label_agreement":null},{"id":"W2167540852","doi":"10.1093/bioinformatics/btq418","title":"MetPA: a web-based metabolomics tool for pathway analysis and visualization","year":2010,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Metabolomics and Mass Spectrometry Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":785,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"National Institute for Nanotechnology; University of Alberta","funders":"","keywords":"Visualization; Computer science; Metabolomics; Context (archaeology); Data visualization; Zoom; Metabolic network; Metabolic pathway; Data mining; Computational biology; Bioinformatics; Biology","score_opus":0.007518783320624244,"score_gpt":0.24923714496547042,"score_spread":0.24171836164484617,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2167540852","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0060025845,0.00066595216,0.42202017,0.00065294193,0.0002112555,0.0002771031,0.0981672,0.4643772,0.0076255845],"genre_scores_gemma":[0.06612741,0.0012230244,0.6994635,0.000530631,0.0002637485,0.0019298855,0.17350201,0.04491902,0.012040816],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9995346,0.00008008111,0.000039854014,0.00008699075,0.00021187445,0.000046700206],"domain_scores_gemma":[0.99860686,0.0006358081,0.00012783438,0.00017660738,0.00030146365,0.0001514404],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010423328,0.0020546038,0.0011635283,0.003203974,0.0005787133,0.0017821556,0.002573448,0.0011711908,0.102701],"category_scores_gemma":[0.0033948133,0.0007512244,0.0011821475,0.0030897963,0.0002563457,0.0020345547,0.0019405456,0.0014880523,0.028799627],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013675462,0.00020900153,0.0028290823,0.0028915827,0.00021571339,0.00083767367,0.0003506871,0.008436637,0.04839533,0.007192343,0.61061937,0.31665504],"study_design_scores_gemma":[0.0010328991,0.00025930232,0.009390996,0.0006557769,0.00015923657,0.0018793914,0.0001495092,0.1888181,0.06420174,0.024927257,0.70812535,0.00040046504],"about_ca_topic_score_codex":0.0015798714,"about_ca_topic_score_gemma":0.0018684816,"teacher_disagreement_score":0.102701,"about_ca_system_score_codex":0.00041011875,"about_ca_system_score_gemma":0.0011514532,"threshold_uncertainty_score":0.3435691},"labels":[],"label_agreement":null},{"id":"W2167566248","doi":"10.1093/bioinformatics/btp595","title":"NAViGaTOR: Network Analysis, Visualization and Graphing Toronto","year":2009,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":273,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"University of Toronto; Ontario Genomics; Ontario Genomics Institute; Genome Canada","keywords":"Visualization; Computer science; Computer graphics (images); Data mining","score_opus":0.00527186983615427,"score_gpt":0.24303482373120608,"score_spread":0.2377629538950518,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2167566248","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0017333281,0.0014106971,0.4789608,0.0010883712,0.00051141257,0.0007363189,0.11547608,0.38472313,0.015359931],"genre_scores_gemma":[0.018207112,0.0021540788,0.7178182,0.00063546683,0.00024981756,0.0042882054,0.16697463,0.07055382,0.019118778],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9973424,0.0005706768,0.0003261202,0.00072053535,0.00085481524,0.00018539588],"domain_scores_gemma":[0.9959196,0.0019418302,0.00029189073,0.00081862975,0.0007633012,0.0002646728],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0043859435,0.0031052805,0.0024504762,0.0062533314,0.0012936131,0.004046976,0.0041220007,0.0011167061,0.12023164],"category_scores_gemma":[0.011831019,0.0021443693,0.0020862978,0.0049148113,0.00089243444,0.0051630023,0.0040631522,0.004671825,0.05210913],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00021915864,0.000080797065,0.0009586278,0.0023997028,0.0003154386,0.00035294006,0.00079328596,0.004160727,0.0065218434,0.025101114,0.8158508,0.14324544],"study_design_scores_gemma":[0.00031209463,0.000081655264,0.0020884168,0.00065797416,0.00013912685,0.0006582516,0.00018601007,0.045364503,0.00894599,0.08593609,0.85542256,0.00020731501],"about_ca_topic_score_codex":0.0036924318,"about_ca_topic_score_gemma":0.004153736,"teacher_disagreement_score":0.12023164,"about_ca_system_score_codex":0.0009560283,"about_ca_system_score_gemma":0.002712589,"threshold_uncertainty_score":0.40221494},"labels":[],"label_agreement":null},{"id":"W2167846577","doi":"10.1093/bioinformatics/btl356","title":"GenePro: a cytoscape plug-in for advanced visualization and analysis of interaction networks","year":2006,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":52,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto; Hospital for Sick Children","funders":"Canadian Institutes of Health Research; Hospital for Sick Children","keywords":"Plug-in; Computer science; Visualization; Parsing; Software; Set (abstract data type); Domain (mathematical analysis); Software engineering; Data science; Data mining; Programming language","score_opus":0.006255042208858774,"score_gpt":0.2517326425909427,"score_spread":0.2454776003820839,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2167846577","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.007329945,0.0017966238,0.39399883,0.0013554231,0.0006094572,0.0006608508,0.049203485,0.5361054,0.008940049],"genre_scores_gemma":[0.06993246,0.0034450726,0.73670286,0.001711046,0.00044644877,0.0044438206,0.10300962,0.060513683,0.019794976],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99926776,0.00012922891,0.00004422364,0.00011973773,0.00038170765,0.000057439996],"domain_scores_gemma":[0.9985037,0.0007927354,0.00011413974,0.00017978223,0.00020574349,0.0002039332],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011148747,0.0017771756,0.0009887476,0.0042375186,0.0010208319,0.0014829491,0.0030500118,0.0013849529,0.025492448],"category_scores_gemma":[0.0040178406,0.0011056182,0.0008928444,0.0024342095,0.00053702015,0.0018060886,0.0018170262,0.0023738605,0.013133299],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005760497,0.00018538459,0.0017916971,0.0027423566,0.00037179937,0.0011949438,0.0003991991,0.007578167,0.046451043,0.010302047,0.72937983,0.19902752],"study_design_scores_gemma":[0.0007479584,0.00013382865,0.0070110103,0.0004091421,0.0002522686,0.0024576858,0.00011545403,0.107283376,0.085269354,0.05244425,0.74344236,0.0004333184],"about_ca_topic_score_codex":0.00503043,"about_ca_topic_score_gemma":0.007199982,"teacher_disagreement_score":0.025492448,"about_ca_system_score_codex":0.00052422495,"about_ca_system_score_gemma":0.0018616227,"threshold_uncertainty_score":0.085280776},"labels":[],"label_agreement":null},{"id":"W2168047871","doi":"10.1093/bioinformatics/btq355","title":"TreesimJ: a flexible, forward time population genetic simulator","year":2010,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic diversity and population structure","field":"Biochemistry, Genetics and Molecular Biology","cited_by":16,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Institute of General Medical Sciences; McGill University; National Science Foundation","keywords":"Computer science; Population; Simulation; Genetic algorithm; Machine learning; Demography","score_opus":0.006045611241174022,"score_gpt":0.22176551904386307,"score_spread":0.21571990780268904,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2168047871","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.033128835,0.00027715485,0.8564024,0.0006634146,0.00034396592,0.00027830116,0.013200943,0.075177334,0.020527577],"genre_scores_gemma":[0.31251764,0.0008653886,0.6232151,0.0008604817,0.00013824808,0.0015680904,0.022035005,0.018965099,0.01983503],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99968624,0.00006720419,0.000023323211,0.000051234074,0.00013222398,0.000039721326],"domain_scores_gemma":[0.99793965,0.0012955632,0.00009578355,0.00020923224,0.00032640173,0.0001333413],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010756365,0.0005792315,0.00058669184,0.0005560322,0.00053170754,0.0011474234,0.0028383594,0.0013194833,0.013864132],"category_scores_gemma":[0.0062033553,0.00057045463,0.0009236465,0.0008000729,0.00045796158,0.0013729519,0.0009616246,0.0016255443,0.0033057875],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00038107258,0.0001980256,0.007900222,0.00058323983,0.00019818847,0.0003669618,0.00047471994,0.74315876,0.007320272,0.045159336,0.102505445,0.09175371],"study_design_scores_gemma":[0.00012454666,0.000027189057,0.00040585588,0.000022217,0.000031499967,0.0000650086,0.000017212556,0.96288943,0.00271296,0.009473781,0.024203034,0.000027273456],"about_ca_topic_score_codex":0.008592096,"about_ca_topic_score_gemma":0.007354026,"teacher_disagreement_score":0.013864132,"about_ca_system_score_codex":0.00064345374,"about_ca_system_score_gemma":0.0017311103,"threshold_uncertainty_score":0.046380162},"labels":[],"label_agreement":null},{"id":"W2168220711","doi":"10.1093/bioinformatics/bts209","title":"MoRFpred, a computational tool for sequence-based prediction and characterization of short disorder-to-order transitioning binding regions in proteins","year":2012,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Protein Structure and Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":369,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"U.S. National Library of Medicine; Natural Sciences and Engineering Research Council of Canada; Killam Trusts; Russian Academy of Sciences; National Institutes of Health; National Science Foundation","keywords":"Sequence (biology); Support vector machine; Computer science; Computational biology; Intrinsically disordered proteins; Artificial intelligence; Chemistry; Biology; Biochemistry","score_opus":0.012483846323756513,"score_gpt":0.2417982339169962,"score_spread":0.2293143875932397,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2168220711","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.30289105,0.0030857988,0.56570965,0.00075596594,0.00017557143,0.0003951465,0.020374304,0.10259269,0.0040198723],"genre_scores_gemma":[0.36979493,0.0009901483,0.5915069,0.00048129144,0.000077068195,0.00066628464,0.031976808,0.002748187,0.0017584254],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99955434,0.00008190772,0.000028256813,0.00016962645,0.00013209728,0.000033711167],"domain_scores_gemma":[0.9989905,0.0005345522,0.00018244072,0.00009481351,0.00011654378,0.000081137594],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011674005,0.0014073839,0.0013137244,0.0016684844,0.00069913425,0.00077883044,0.0015305877,0.0010033165,0.0021899159],"category_scores_gemma":[0.0028452156,0.00059993943,0.0011257506,0.0008350879,0.00041189147,0.001046298,0.0008345124,0.001215583,0.0010001437],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0027007589,0.00091156526,0.03943468,0.0029660338,0.0011952267,0.0014235007,0.00038054818,0.46531796,0.115153134,0.011402996,0.088876836,0.27023682],"study_design_scores_gemma":[0.000096534866,0.00015080899,0.0024704975,0.000045778972,0.000057586185,0.0002997588,0.000039637216,0.9718992,0.015260313,0.003182569,0.006459316,0.00003803885],"about_ca_topic_score_codex":0.0030412655,"about_ca_topic_score_gemma":0.006072129,"teacher_disagreement_score":0.0030412655,"about_ca_system_score_codex":0.0007827974,"about_ca_system_score_gemma":0.0014022834,"threshold_uncertainty_score":0.0073259473},"labels":[],"label_agreement":null},{"id":"W2168361262","doi":"10.1093/bioinformatics/btp525","title":"A report on the 2009 SIG on short read sequencing and algorithms (Short-SIG)","year":2009,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of Toronto","funders":"","keywords":"Computer science; Algorithm; Software; Programming language","score_opus":0.02606938456688283,"score_gpt":0.2555712443093973,"score_spread":0.2295018597425145,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2168361262","genre_codex":"editorial","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0034866044,0.0701374,0.08867732,0.17494713,0.5297322,0.0032803102,0.028066393,0.009244708,0.09242798],"genre_scores_gemma":[0.010349752,0.08647492,0.10806653,0.08777912,0.112986274,0.0025545745,0.07116655,0.0075570135,0.51306534],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.98774904,0.0022041495,0.00086334546,0.0017633558,0.006391847,0.0010282415],"domain_scores_gemma":[0.95321226,0.008772922,0.001716033,0.0025675893,0.024275657,0.009455543],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.019390512,0.0026330824,0.0022435084,0.0031394719,0.002136817,0.008166326,0.0039659603,0.006964802,0.066215],"category_scores_gemma":[0.029042015,0.0009503515,0.0016576113,0.0027646902,0.0015421339,0.0073090266,0.004137395,0.007579773,0.07133269],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00008688349,0.00006169305,0.00016187066,0.00017745746,0.000013994283,0.00004656392,0.000025436935,0.00021763598,0.0008506114,0.0010143426,0.9570476,0.04029594],"study_design_scores_gemma":[0.000013195635,0.0000533442,0.0003261004,0.00014657246,0.000008386695,0.000031945092,0.000028071629,0.00032240848,0.00078401255,0.0009879164,0.9972723,0.000025810998],"about_ca_topic_score_codex":0.0092779575,"about_ca_topic_score_gemma":0.016442832,"teacher_disagreement_score":0.066215,"about_ca_system_score_codex":0.003908072,"about_ca_system_score_gemma":0.0103518525,"threshold_uncertainty_score":0.22151124},"labels":[],"label_agreement":null},{"id":"W2168443777","doi":"10.1093/bioinformatics/btt599","title":"A probabilistic approach to explore human miRNA targetome by integrating miRNA-overexpression data and sequence information","year":2013,"lang":"en","type":"article","venue":"Bioinformatics","topic":"MicroRNA in disease regulation","field":"Biochemistry, Genetics and Molecular Biology","cited_by":39,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Bioconductor; microRNA; Computational biology; Probabilistic logic; Computer science; Data mining; Gene; Biology; Artificial intelligence; Genetics","score_opus":0.03972558507809203,"score_gpt":0.27838884180424195,"score_spread":0.2386632567261499,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2168443777","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.1005057,0.00047354435,0.88902193,0.00020247023,0.000012377777,0.0001233155,0.006522749,0.0019345082,0.0012034366],"genre_scores_gemma":[0.62458456,0.00039328326,0.3594651,0.00016701728,0.000037079128,0.00059243094,0.013274234,0.0002604481,0.0012258376],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9991697,0.00023201873,0.000052212454,0.0002735343,0.00022463246,0.000047930254],"domain_scores_gemma":[0.99897194,0.00061781326,0.00013171483,0.00010013945,0.00014503092,0.000033345023],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016955013,0.00064406765,0.00057501794,0.0023482933,0.00048394065,0.00065971585,0.0006504894,0.00047899305,0.0016770128],"category_scores_gemma":[0.003861613,0.0004746579,0.0012770339,0.0018868274,0.00031011458,0.0006160987,0.00091168506,0.00052948,0.00046507694],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011454866,0.00039030577,0.08773634,0.0011283419,0.0013639753,0.0005027892,0.00030309707,0.43664244,0.10711116,0.016851602,0.0070795952,0.33974484],"study_design_scores_gemma":[0.00003524811,0.00013091006,0.018222637,0.000021867754,0.00014808807,0.00038712053,0.000033760312,0.93912286,0.014207009,0.023306916,0.0043315752,0.000051910883],"about_ca_topic_score_codex":0.0032525545,"about_ca_topic_score_gemma":0.005567807,"teacher_disagreement_score":0.0032525545,"about_ca_system_score_codex":0.00063108193,"about_ca_system_score_gemma":0.0010641665,"threshold_uncertainty_score":0.008966744},"labels":[],"label_agreement":null},{"id":"W2168546919","doi":"10.1093/bioinformatics/btu023","title":"Exploring genome characteristics and sequence quality without a reference","year":2014,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":162,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ontario Institute for Cancer Research","funders":"Government of Ontario; Ontario Institute for Cancer Research; Wellcome Trust","keywords":"Sequence assembly; Reference genome; Software; Genome; Computer science; Sequence (biology); DNA sequencing; Loss of heterozygosity; Computational biology; Key (lock); Data mining; Biology; Genetics; DNA; Gene; Programming language","score_opus":0.1222213845852358,"score_gpt":0.2881844902889477,"score_spread":0.16596310570371192,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2168546919","genre_codex":"empirical","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.87631375,0.0015083741,0.102140896,0.00032733582,0.000040346324,0.00010696122,0.012650365,0.0013331047,0.005578856],"genre_scores_gemma":[0.87791175,0.00047188252,0.093933165,0.0001260962,0.000024463288,0.00008206637,0.025440842,0.0006934138,0.001316352],"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99749494,0.0004957858,0.00023838299,0.000626628,0.0010424799,0.00010179876],"domain_scores_gemma":[0.9852601,0.0064870515,0.0023222517,0.0014753769,0.0040346817,0.00042053033],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0033549215,0.00033917712,0.00049953815,0.0020983417,0.00041294648,0.0015219091,0.00061889825,0.00060950156,0.0033407374],"category_scores_gemma":[0.016830318,0.00019921732,0.00034244423,0.003781931,0.0004528556,0.0010678082,0.00078982156,0.00039485085,0.0017392667],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001382923,0.0002020315,0.6079013,0.0012169565,0.00039245628,0.0006380471,0.00091425376,0.020971805,0.1506979,0.0026059982,0.0056753154,0.20740107],"study_design_scores_gemma":[0.00005657138,0.00079530524,0.72748363,0.00021823995,0.00041666464,0.0026137272,0.00080186623,0.06250018,0.17186046,0.006739743,0.026334548,0.0001792095],"about_ca_topic_score_codex":0.002771104,"about_ca_topic_score_gemma":0.0035899794,"teacher_disagreement_score":0.0033549215,"about_ca_system_score_codex":0.0004735315,"about_ca_system_score_gemma":0.00040904284,"threshold_uncertainty_score":0.017742753},"labels":[],"label_agreement":null},{"id":"W2168859815","doi":"10.1093/bioinformatics/bts340","title":"ChromoHub: a data hub for navigators of chromatin-mediated signalling","year":2012,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":90,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Structural Genomics Consortium; University of Toronto","funders":"Ministero dello Sviluppo Economico; Ontario Ministry of Economic Development and Innovation; Canadian Institutes of Health Research; Genome Canada; Wellcome Trust; Eli Lilly Canada; GlaxoSmithKline; Pfizer; Eli Lilly and Company","keywords":"Chromatin; Histone; Computational biology; Resource (disambiguation); Epigenetics; Biology; Bromodomain; Relevance (law); Chromatin remodeling; Signalling; Nucleosome; Data science; Computer science; Genetics; Gene; Cell biology","score_opus":0.02208753217242906,"score_gpt":0.26162781869681123,"score_spread":0.23954028652438217,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2168859815","genre_codex":"dataset","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00394924,0.0013041723,0.08701888,0.0021249054,0.00046582014,0.00041280163,0.5056223,0.38117898,0.017922929],"genre_scores_gemma":[0.04464365,0.0014478879,0.11024505,0.0017245632,0.00028239534,0.0010401385,0.78312516,0.050040446,0.007450661],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9978175,0.0003381962,0.00027413468,0.0005019786,0.0008422172,0.00022601418],"domain_scores_gemma":[0.98731357,0.0051473677,0.00092480687,0.003021392,0.0017187904,0.0018739554],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0045004943,0.0021517957,0.001906152,0.0066893776,0.0015543049,0.0046326877,0.005110906,0.0024860355,0.076057196],"category_scores_gemma":[0.020475885,0.0011547105,0.001051087,0.010031284,0.0010008719,0.0059177526,0.006997759,0.0030023453,0.04684221],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.002089864,0.00013715404,0.0038899712,0.0031474226,0.00021124771,0.0005237084,0.00045845157,0.0020832845,0.007742622,0.018461777,0.8627411,0.098513365],"study_design_scores_gemma":[0.00070257985,0.00013017633,0.0065516285,0.00064423645,0.00016488267,0.00048209383,0.00021908685,0.022646358,0.022982573,0.03350788,0.9116619,0.0003065658],"about_ca_topic_score_codex":0.0073865633,"about_ca_topic_score_gemma":0.006728829,"teacher_disagreement_score":0.076057196,"about_ca_system_score_codex":0.0023180158,"about_ca_system_score_gemma":0.004243352,"threshold_uncertainty_score":0.25443667},"labels":[],"label_agreement":null},{"id":"W2168872202","doi":"10.1093/bioinformatics/bth196","title":"Identifying multiple alignment regions satisfying simple formulas and patterns","year":2004,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"National Institutes of Health","keywords":"Simple (philosophy); Documentation; Set (abstract data type); Computer science; Software; Data mining; Theoretical computer science; Algorithm; Programming language","score_opus":0.022933498496123354,"score_gpt":0.25007345929627206,"score_spread":0.2271399608001487,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2168872202","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.07665933,0.00035408715,0.894642,0.00042434048,0.000067291185,0.0004032511,0.004639345,0.019113665,0.0036967031],"genre_scores_gemma":[0.10866619,0.00024109411,0.88105476,0.00013618219,0.00003430776,0.0003785316,0.0062786154,0.0021562139,0.0010540514],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9959837,0.0007731724,0.0011055935,0.0009014409,0.0010351205,0.00020098938],"domain_scores_gemma":[0.97768855,0.014176987,0.0040065823,0.0015851274,0.0022463107,0.0002964119],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003950464,0.001562197,0.0011931646,0.0034498496,0.0012132227,0.0029660407,0.0020394488,0.0013859755,0.0065062693],"category_scores_gemma":[0.039764512,0.00085274014,0.0012347861,0.0050242837,0.0011086748,0.004232978,0.0015520657,0.0012151188,0.0023187536],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0016609925,0.0003892886,0.07251358,0.004562925,0.00031399203,0.003504878,0.002317798,0.053160235,0.0793409,0.09837676,0.026938261,0.6569203],"study_design_scores_gemma":[0.00036994237,0.000595554,0.00910692,0.0008751639,0.00028657328,0.004762549,0.0012155453,0.55770457,0.15838625,0.20400184,0.062477216,0.00021798955],"about_ca_topic_score_codex":0.001558208,"about_ca_topic_score_gemma":0.0021422917,"teacher_disagreement_score":0.0065062693,"about_ca_system_score_codex":0.0006879429,"about_ca_system_score_gemma":0.0018482299,"threshold_uncertainty_score":0.02176565},"labels":[],"label_agreement":null},{"id":"W2168950020","doi":"10.1093/bioinformatics/btq184","title":"VARiD: A variation detection framework for color-space and letter-space platforms","year":2010,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Variation (astronomy); Space (punctuation); Computer science; Color space; Computer vision; Artificial intelligence; Computer graphics (images); Image (mathematics); Operating system; Physics; Astronomy","score_opus":0.007877595987041856,"score_gpt":0.2256863102814138,"score_spread":0.21780871429437196,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2168950020","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0009605187,0.000070349044,0.9894749,0.00004646796,0.000020138434,0.000064582884,0.0007953834,0.008364285,0.00020331514],"genre_scores_gemma":[0.025765488,0.000084963365,0.9697674,0.000097915974,0.000034359382,0.00043887226,0.001918761,0.0012535815,0.00063855934],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9934683,0.0024315738,0.000598516,0.0014831543,0.0017267412,0.00029175045],"domain_scores_gemma":[0.9891929,0.0070363255,0.0011012423,0.001276415,0.001055,0.0003380805],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0103653,0.002012698,0.0019396315,0.0051217023,0.0012503125,0.0032269307,0.0057114856,0.002011947,0.0059753205],"category_scores_gemma":[0.024291383,0.001836123,0.0035052253,0.0029233498,0.0017708272,0.0029319336,0.0040358305,0.0027497788,0.0022400708],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0015028081,0.00035106487,0.01712578,0.0016130097,0.0010541774,0.0011805602,0.0007883134,0.13238719,0.022243312,0.13659471,0.041242834,0.64391625],"study_design_scores_gemma":[0.00018772318,0.00012492735,0.0023363286,0.00010688283,0.00012048625,0.0010236683,0.00007667239,0.83424664,0.0152706625,0.10743451,0.038844567,0.00022686188],"about_ca_topic_score_codex":0.0044566337,"about_ca_topic_score_gemma":0.0047070454,"teacher_disagreement_score":0.0103653,"about_ca_system_score_codex":0.0014758783,"about_ca_system_score_gemma":0.0028611093,"threshold_uncertainty_score":0.054817557},"labels":[],"label_agreement":null},{"id":"W2168951863","doi":"10.1093/bioinformatics/btu669","title":"GeneNet Toolbox for MATLAB: a flexible platform for the analysis of gene connectivity in biological networks","year":2014,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":16,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Medical Research Council; European Commission; Natural Sciences and Engineering Research Council of Canada; Wellcome Trust","keywords":"Toolbox; Subnetwork; Computer science; Graphical user interface; MATLAB; Set (abstract data type); Permutation (music); Visualization; Data mining; Gene regulatory network; Biological network; Software; Interface (matter); Bioinformatics; Gene; Programming language; Computer network; Operating system; Biology","score_opus":0.02391273013113176,"score_gpt":0.25529952655803784,"score_spread":0.23138679642690607,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2168951863","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0019735456,0.0006408924,0.6750891,0.0004161277,0.00030249057,0.00015776308,0.02945428,0.28373185,0.008233956],"genre_scores_gemma":[0.036618873,0.0010997627,0.7922863,0.0008583563,0.00022510973,0.0031662835,0.042934153,0.10759884,0.015212373],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99911803,0.00023261642,0.00010488294,0.00021249929,0.00025204226,0.00007986336],"domain_scores_gemma":[0.99722385,0.0015144881,0.00026090944,0.00035012054,0.0005076999,0.00014285605],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0019096695,0.0020077217,0.0015773838,0.0025939632,0.0006023073,0.001763135,0.0031958001,0.001133311,0.13791832],"category_scores_gemma":[0.0064104414,0.0011289776,0.0014089979,0.0017485564,0.00070394325,0.0018446557,0.0021367092,0.002313149,0.064796925],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0004584542,0.00012279833,0.0022524453,0.0029641965,0.0003664999,0.00070955494,0.00050711614,0.033919383,0.016549079,0.034262054,0.7026462,0.2052422],"study_design_scores_gemma":[0.0004588496,0.00016786174,0.0038529963,0.000648131,0.00018167448,0.001449699,0.00013034207,0.21506736,0.026540155,0.100929156,0.65030336,0.0002704293],"about_ca_topic_score_codex":0.0022027588,"about_ca_topic_score_gemma":0.0030831657,"teacher_disagreement_score":0.13791832,"about_ca_system_score_codex":0.0008613622,"about_ca_system_score_gemma":0.0014926048,"threshold_uncertainty_score":0.46138275},"labels":[],"label_agreement":null},{"id":"W2168980979","doi":"10.1093/bioinformatics/btq498","title":"Model-based clustering of microarray expression data via latent Gaussian mixture models","year":2010,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":226,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Guelph","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Mixture model; Bayesian information criterion; Cluster analysis; Covariance; Expectation–maximization algorithm; Computer science; Model selection; Gene chip analysis; Data mining; Gaussian; Statistical model; Bayesian probability; Artificial intelligence; Mathematics; Statistics; DNA microarray; Gene expression; Biology; Gene; Genetics; Maximum likelihood","score_opus":0.03102465879333444,"score_gpt":0.266033582338828,"score_spread":0.23500892354549358,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2168980979","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0035884453,0.00033417635,0.99507046,0.00012645488,0.000012546631,0.000037382742,0.00013274126,0.00046427434,0.00023352185],"genre_scores_gemma":[0.1836413,0.0017001099,0.80887264,0.0002001927,0.00014360015,0.0005765536,0.0026717908,0.00029925673,0.0018946058],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99637705,0.0015599845,0.00017136076,0.0007931546,0.0009599967,0.00013849203],"domain_scores_gemma":[0.9958866,0.0025587946,0.0003943622,0.0005001835,0.0005895766,0.000070432885],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0052217753,0.0013401685,0.0019080016,0.00218399,0.0006772004,0.0015823578,0.0023209916,0.0014998184,0.0013956891],"category_scores_gemma":[0.01220773,0.0008048644,0.0019474567,0.003973768,0.0011982614,0.0019960743,0.00149755,0.00198666,0.00163961],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00019318574,0.00010120061,0.0040486366,0.00053125556,0.00026672118,0.00011945973,0.0003842173,0.73982793,0.008474026,0.038876235,0.0042496487,0.20292743],"study_design_scores_gemma":[0.000013127203,0.000024960804,0.0010726583,0.000029287385,0.00003455353,0.00007854966,0.000019972185,0.96703297,0.0016780049,0.02824108,0.0017387377,0.000035936842],"about_ca_topic_score_codex":0.0053522093,"about_ca_topic_score_gemma":0.0038767292,"teacher_disagreement_score":0.0053522093,"about_ca_system_score_codex":0.0013374627,"about_ca_system_score_gemma":0.0014644517,"threshold_uncertainty_score":0.027615666},"labels":[],"label_agreement":null},{"id":"W2169224664","doi":"10.1093/bioinformatics/18.1.215","title":"Symmetry observations in long nucleotide sequences: a commentary on the Discovery Note of Qi and Cuticchia","year":2002,"lang":"en","type":"letter","venue":"Bioinformatics","topic":"RNA and protein synthesis mechanisms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":22,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Queen's University","funders":"","keywords":"Genome; DNA; Oligonucleotide; Genetics; Biology; Combinatorics; Computational biology; Mathematics; Gene","score_opus":0.023355846145395025,"score_gpt":0.23210959057574332,"score_spread":0.2087537444303483,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2169224664","genre_codex":"commentary","genre_gemma":"commentary","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"commentary","genre_consensus":"commentary","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00011481389,0.071580335,0.00062663184,0.80355394,0.12331488,0.000008943117,0.000103367245,0.000059350434,0.0006377382],"genre_scores_gemma":[0.001297872,0.028594922,0.0007635394,0.8360269,0.1314764,0.00004375471,0.000074313095,0.00014234781,0.0015799063],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9841666,0.003965047,0.0021663145,0.0036346784,0.0054116435,0.00065576844],"domain_scores_gemma":[0.9052092,0.068227604,0.003096984,0.0027957235,0.016695803,0.0039746044],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.02258366,0.0026461605,0.0036757225,0.0036807654,0.0054501044,0.00781681,0.011724506,0.034607176,0.00333711],"category_scores_gemma":[0.067143954,0.001574565,0.0028194136,0.003689509,0.023182256,0.022732573,0.0054691085,0.09085405,0.005156998],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000059619124,0.000021044907,0.00010114636,0.00035064176,0.000032307395,0.00015438444,0.00040454802,0.00007627606,0.00017654999,0.009596935,0.9761378,0.012888704],"study_design_scores_gemma":[0.00004150496,0.00004330469,0.00024297628,0.0007573321,0.000026877264,0.00034965685,0.0002477615,0.00010392725,0.00024939553,0.015123988,0.9827323,0.00008086001],"about_ca_topic_score_codex":0.019161465,"about_ca_topic_score_gemma":0.017363848,"teacher_disagreement_score":0.034607176,"about_ca_system_score_codex":0.0104414,"about_ca_system_score_gemma":0.008210175,"threshold_uncertainty_score":0.11943525},"labels":[],"label_agreement":null},{"id":"W2169558190","doi":"10.1093/bioinformatics/btl326","title":"Difference detection in LC-MS data for protein biomarker discovery","year":2007,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Advanced Proteomics Techniques and Applications","field":"Chemistry","cited_by":91,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of Toronto","funders":"","keywords":"Biomarker discovery; Computer science; Task (project management); SIGNAL (programming language); Biomarker; Class (philosophy); Data mining; Artificial intelligence; Pattern recognition (psychology); Proteomics; Chemistry","score_opus":0.0337800281231293,"score_gpt":0.30226955595067095,"score_spread":0.26848952782754165,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2169558190","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.20468046,0.0017066816,0.7782372,0.0015615872,0.00029404732,0.00039027398,0.0030173124,0.0079124225,0.0021999935],"genre_scores_gemma":[0.52837664,0.00036923884,0.46641836,0.00073223544,0.00012713046,0.00023380127,0.0029028952,0.0002960617,0.0005437956],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99072254,0.0025909217,0.00088146125,0.0018632172,0.0037009541,0.0002409179],"domain_scores_gemma":[0.96166474,0.025005596,0.0035992279,0.0059437836,0.0033387837,0.00044797643],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.010123715,0.0012683475,0.0010134107,0.0038724307,0.00067609706,0.0018806449,0.0018367426,0.0027786514,0.0017632178],"category_scores_gemma":[0.03408317,0.00047371862,0.00085191533,0.003801916,0.0017896276,0.002303929,0.001370388,0.0026506947,0.0017480912],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.003747542,0.0012160809,0.054416314,0.0023331014,0.0006697947,0.0007756435,0.0005381484,0.03746451,0.44815248,0.009173661,0.0076340605,0.43387872],"study_design_scores_gemma":[0.00010858935,0.00093976926,0.038862146,0.0001227799,0.00015882855,0.0021810445,0.00018952196,0.39120218,0.52589977,0.031841103,0.008278646,0.00021565771],"about_ca_topic_score_codex":0.00064083224,"about_ca_topic_score_gemma":0.00056917535,"teacher_disagreement_score":0.010123715,"about_ca_system_score_codex":0.0011417337,"about_ca_system_score_gemma":0.0006854675,"threshold_uncertainty_score":0.05353993},"labels":[],"label_agreement":null},{"id":"W2169802764","doi":"10.1093/bioinformatics/btq373","title":"Improved sequence-based prediction of disordered regions with multilayer fusion of multiple information sources","year":2010,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Protein Structure and Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":201,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Computer science; Fusion; Artificial intelligence; Data mining; Sequence (biology); Dihedral angle; Classifier (UML); Similarity (geometry); Pattern recognition (psychology); Machine learning; Chemistry; Biology; Genetics","score_opus":0.006990120934709545,"score_gpt":0.2063050445354069,"score_spread":0.19931492360069736,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2169802764","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.4061775,0.0015237624,0.58125716,0.00038266045,0.000065617714,0.000110483226,0.002581141,0.0066784285,0.0012232306],"genre_scores_gemma":[0.7666425,0.00033187194,0.22861296,0.00009264632,0.000043455642,0.000069883106,0.0034716025,0.00008770239,0.0006474047],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9996414,0.00007094888,0.00003064591,0.00007932353,0.00012375735,0.000053816297],"domain_scores_gemma":[0.9986889,0.0005750604,0.00021167214,0.000098579585,0.00033318397,0.00009259026],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010873657,0.00090222567,0.00082351797,0.0018131565,0.00028040286,0.0006245569,0.00063021074,0.00073126814,0.0007983506],"category_scores_gemma":[0.0028870406,0.0002773368,0.00068311946,0.0010472628,0.0002689781,0.0011969911,0.0009743608,0.00069169444,0.00041989377],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0012259902,0.00047654432,0.03743978,0.0004126764,0.00029758495,0.00044128663,0.00016404084,0.55988127,0.04862905,0.0014059494,0.0063516283,0.34327415],"study_design_scores_gemma":[0.000010505823,0.000039830964,0.0015176238,0.000008091673,0.0000157901,0.000035940142,0.0000088488905,0.99277663,0.004690036,0.00067728595,0.0002103376,0.00000901108],"about_ca_topic_score_codex":0.0052541937,"about_ca_topic_score_gemma":0.0040909518,"teacher_disagreement_score":0.0052541937,"about_ca_system_score_codex":0.0005265448,"about_ca_system_score_gemma":0.0007687214,"threshold_uncertainty_score":0.010447204},"labels":[],"label_agreement":null},{"id":"W2170080415","doi":"10.1093/bioinformatics/btu292","title":"Probabilistic method for detecting copy number variation in a fetal genome using maternal plasma sequencing","year":2014,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Prenatal Screening and Diagnostics","field":"Medicine","cited_by":19,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"SickKids Foundation; Hospital for Sick Children; University of Toronto","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Copy-number variation; Biology; Genome; Genetics; Single-nucleotide polymorphism; DNA sequencing; Chromosome; Computational biology; SNP; Genomics; SNP array; Genotype; Gene","score_opus":0.034628178887073716,"score_gpt":0.30348910164582593,"score_spread":0.26886092275875223,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2170080415","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.008563017,0.000044421504,0.9901779,0.00006884239,0.000008719204,0.000027134723,0.00013598503,0.0006144341,0.00035953982],"genre_scores_gemma":[0.2966796,0.00015997769,0.7006086,0.00012711085,0.000035366753,0.00030094915,0.00064610416,0.00017888582,0.0012634234],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99890745,0.00042164957,0.00004008921,0.000224251,0.00036481372,0.00004166615],"domain_scores_gemma":[0.99608254,0.0030655246,0.0002739424,0.0002448579,0.0002675626,0.00006555934],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0017177665,0.0004922065,0.00056219654,0.0011297023,0.00043374373,0.00058819575,0.001404183,0.0008342679,0.002549532],"category_scores_gemma":[0.008669811,0.0005207917,0.00075125555,0.000840044,0.00050575344,0.00066855166,0.0007761336,0.0008127979,0.0005846488],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002754138,0.00008720326,0.015328795,0.00016067672,0.00018049283,0.00031881098,0.0001444236,0.8403438,0.01968563,0.022558503,0.0017980502,0.09911826],"study_design_scores_gemma":[0.0000124357775,0.000014069452,0.0008891036,0.000005315747,0.000011497167,0.000104210514,0.0000039302204,0.9902767,0.0022343548,0.005928788,0.00050522503,0.00001424255],"about_ca_topic_score_codex":0.0055557885,"about_ca_topic_score_gemma":0.004689281,"teacher_disagreement_score":0.0055557885,"about_ca_system_score_codex":0.00070407544,"about_ca_system_score_gemma":0.0010765899,"threshold_uncertainty_score":0.011046886},"labels":[],"label_agreement":null},{"id":"W2170123419","doi":"10.1093/bioinformatics/bts491","title":"Intensity quantile estimation and mapping—a novel algorithm for the correction of image non-uniformity bias in HCS data","year":2012,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Cell Image Analysis Techniques","field":"Biochemistry, Genetics and Molecular Biology","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"McGill University and Génome Québec Innovation Centre","funders":"Centre National de la Recherche Scientifique; Fonds Québécois de la Recherche sur la Nature et les Technologies; Infrastructures en Biologie Santé et Agronomie","keywords":"Quantile; Image (mathematics); Pixel; Algorithm; Computer science; MATLAB; Intensity (physics); Function (biology); Image processing; Field (mathematics); Artificial intelligence; Pattern recognition (psychology); Statistics; Mathematics; Physics; Optics","score_opus":0.037872389847886434,"score_gpt":0.29891965728376513,"score_spread":0.2610472674358787,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2170123419","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00268178,0.000071842995,0.9941736,0.000070076625,0.0000239131,0.000036936173,0.00007146416,0.0026836246,0.00018672475],"genre_scores_gemma":[0.02778281,0.00010906656,0.9697262,0.00009980069,0.00003738805,0.00016105056,0.00047203826,0.0006820308,0.00092955615],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9985825,0.00028175776,0.00009929888,0.00035572407,0.00058589363,0.000094835734],"domain_scores_gemma":[0.9961047,0.001591621,0.00048143187,0.0008125531,0.00089543546,0.00011433166],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0034144872,0.0010512116,0.0010310713,0.0014665645,0.00065985764,0.0014547339,0.0024035035,0.0011286971,0.0036302542],"category_scores_gemma":[0.010457553,0.00066969503,0.00094272237,0.0015433942,0.0009280351,0.0014855046,0.0024088556,0.0022723316,0.0026441202],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00033026363,0.00012093746,0.0041394425,0.00025050435,0.00015247203,0.00022489572,0.00030507683,0.04880258,0.08270448,0.011700262,0.014045283,0.8372239],"study_design_scores_gemma":[0.000051652365,0.000079410354,0.0037546284,0.00003089031,0.00003295587,0.00042701935,0.000070792026,0.9112427,0.055960797,0.012103572,0.016177783,0.00006771921],"about_ca_topic_score_codex":0.0023773739,"about_ca_topic_score_gemma":0.0027277754,"teacher_disagreement_score":0.0036302542,"about_ca_system_score_codex":0.00093664014,"about_ca_system_score_gemma":0.0016200221,"threshold_uncertainty_score":0.018057704},"labels":[],"label_agreement":null},{"id":"W2170192148","doi":"10.1093/bioinformatics/18.suppl_1.s111","title":"Binary tree-structured vector quantization approach toclustering and visualizing microarray data","year":2002,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":73,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ontario Institute for Cancer Research","funders":"National Cancer Institute","keywords":"Cluster analysis; Computer science; Vector quantization; Data mining; Binary tree; Binary data; Tree (set theory); k-d tree; Artificial intelligence; Pattern recognition (psychology); Binary number; Mathematics; Algorithm; Tree traversal","score_opus":0.05006105434218325,"score_gpt":0.27732884328221424,"score_spread":0.227267788940031,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2170192148","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.009620927,0.0005196299,0.98643386,0.00027851795,0.000054525728,0.00010211978,0.00045332694,0.0016105453,0.00092648354],"genre_scores_gemma":[0.12322999,0.0006687272,0.87227184,0.000118318676,0.00005108808,0.0003923589,0.0015815902,0.0003061493,0.0013798665],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9985886,0.0005440829,0.0001263736,0.00019968217,0.00047143287,0.00006974097],"domain_scores_gemma":[0.99735814,0.0010170863,0.00024716032,0.0002794925,0.0010057249,0.000092396884],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0020271374,0.00086017046,0.00081519625,0.0018765382,0.0005563579,0.0014902572,0.0014455422,0.0007626714,0.004828113],"category_scores_gemma":[0.0055519356,0.00032613528,0.00062366406,0.0029384282,0.00086223835,0.0015096342,0.0011890528,0.001010581,0.001450635],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0004124758,0.00018681854,0.0027273672,0.0017427165,0.00014905569,0.00022925375,0.0010176689,0.29215515,0.05657519,0.074904844,0.017264491,0.552635],"study_design_scores_gemma":[0.000028774666,0.00005963882,0.0009373881,0.00006753531,0.0000148045865,0.000087129294,0.00012845708,0.9427189,0.01179218,0.039385702,0.0047380775,0.00004146232],"about_ca_topic_score_codex":0.0026168395,"about_ca_topic_score_gemma":0.0020451983,"teacher_disagreement_score":0.004828113,"about_ca_system_score_codex":0.00084477756,"about_ca_system_score_gemma":0.0011593523,"threshold_uncertainty_score":0.016151667},"labels":[],"label_agreement":null},{"id":"W2170330565","doi":"10.1093/bioinformatics/btr353","title":"A tree-based approach for motif discovery and sequence classification","year":2011,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Princess Margaret Cancer Centre; University Health Network; University of Toronto; Ontario Institute for Cancer Research","funders":"Canadian Institutes of Health Research","keywords":"Computer science; Multiple sequence alignment; Artificial intelligence; Pattern recognition (psychology); Parallelizable manifold; Sequence (biology); Tree (set theory); Data mining; Sequence alignment; Algorithm; Mathematics; Biology; Peptide sequence; Genetics","score_opus":0.038376186216780225,"score_gpt":0.23357254132165156,"score_spread":0.19519635510487132,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2170330565","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.002402987,0.00025279386,0.9930074,0.00015981871,0.000043684922,0.0002290315,0.0008250337,0.002652474,0.0004266163],"genre_scores_gemma":[0.018595992,0.00017375931,0.97661746,0.00013193887,0.000045831046,0.00045457808,0.0030385512,0.00021687253,0.0007250641],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99537057,0.0010601728,0.000488913,0.0013836058,0.0014789837,0.0002178709],"domain_scores_gemma":[0.9927884,0.0039473423,0.00053990004,0.0010439062,0.0014196814,0.0002607397],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0035265002,0.001864311,0.0019148226,0.0056650634,0.0013764185,0.002785565,0.0045979233,0.00264324,0.005059199],"category_scores_gemma":[0.014423822,0.0009338061,0.0030761317,0.008272955,0.0011583803,0.003347519,0.002154122,0.0035214988,0.0038730886],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00035384487,0.00036782306,0.0049718595,0.0007925049,0.00040396806,0.0002652761,0.00027294355,0.09279051,0.012981336,0.024746587,0.024902953,0.8371504],"study_design_scores_gemma":[0.000053177333,0.000096320095,0.0005979054,0.000051509942,0.0000476649,0.00029283678,0.00004663558,0.93028706,0.003664678,0.054073993,0.010760635,0.000027547445],"about_ca_topic_score_codex":0.0044018636,"about_ca_topic_score_gemma":0.0050135083,"teacher_disagreement_score":0.0056650634,"about_ca_system_score_codex":0.0015809323,"about_ca_system_score_gemma":0.0025133432,"threshold_uncertainty_score":0.018650115},"labels":[],"label_agreement":null},{"id":"W2170645625","doi":"10.1093/bioinformatics/btl346","title":"Detecting potential labeling errors in microarrays by data perturbation","year":2006,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":80,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Computer science; Classifier (UML); Computation; Data mining; Support vector machine; Pattern recognition (psychology); Artificial intelligence; Algorithm; Machine learning","score_opus":0.014430927581674356,"score_gpt":0.24735452887072357,"score_spread":0.2329236012890492,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2170645625","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"methods","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":"methods","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.11876527,0.00029765567,0.87851375,0.00043047007,0.00006507945,0.00012981237,0.00008604041,0.0014494548,0.0002624959],"genre_scores_gemma":[0.6009535,0.00013579604,0.3971134,0.00033333202,0.00008215704,0.00021054065,0.00041620378,0.00020748904,0.0005474507],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9905103,0.0044450527,0.000576804,0.0018969908,0.0023577374,0.00021306526],"domain_scores_gemma":[0.943909,0.040841166,0.0062306067,0.0050845235,0.0035039394,0.00043073238],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.010979452,0.0009311581,0.00091917434,0.0015107753,0.0009971538,0.0010355243,0.0011303944,0.0013546551,0.0004188793],"category_scores_gemma":[0.053855747,0.0004636883,0.0006185557,0.0015317335,0.0025273918,0.0014940398,0.0012566318,0.0011991189,0.00033656805],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0024030132,0.0004686906,0.07446575,0.00075371756,0.00037054636,0.0009887476,0.0020867665,0.3332418,0.19403282,0.014860303,0.0032338481,0.373094],"study_design_scores_gemma":[0.00004662457,0.00033334125,0.015742462,0.000048071113,0.00007138435,0.00093452877,0.00017904432,0.79129606,0.16403437,0.02556238,0.0016679559,0.00008377828],"about_ca_topic_score_codex":0.0009477576,"about_ca_topic_score_gemma":0.00068119547,"teacher_disagreement_score":0.9890205,"about_ca_system_score_codex":0.001288326,"about_ca_system_score_gemma":0.0009019868,"threshold_uncertainty_score":0.058065593},"labels":[],"label_agreement":null},{"id":"W2170652857","doi":"10.1093/bioinformatics/16.5.465","title":"BIND—a data specification for storing and describing biomolecular interactions, molecular complexes and pathways","year":2000,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":181,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Lunenfeld-Tanenbaum Research Institute","funders":"","keywords":"Computer science; Computational biology; Protein–protein interaction; Function (biology); Object (grammar); Sequence (biology); Database; Theoretical computer science; Data mining; Biology; Genetics; Artificial intelligence","score_opus":0.056821403276481976,"score_gpt":0.2643492590564262,"score_spread":0.20752785577994423,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2170652857","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0061067287,0.0009926762,0.5463828,0.0024927503,0.0012226201,0.0014128978,0.3365308,0.07508987,0.029768847],"genre_scores_gemma":[0.02879161,0.001820029,0.29178274,0.0027288632,0.00031745696,0.0023953428,0.6326846,0.015749607,0.023729764],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9927148,0.0014517739,0.0022367383,0.0005859785,0.0025799742,0.00043070287],"domain_scores_gemma":[0.9722468,0.007979347,0.0017564972,0.008472819,0.008017972,0.0015265195],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.013193525,0.0024607084,0.0024836643,0.0047488445,0.0025840583,0.008903163,0.005714216,0.0034578554,0.037334204],"category_scores_gemma":[0.022636257,0.0017514161,0.0021331355,0.007306364,0.002107308,0.009925242,0.004617142,0.003908213,0.0404774],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0019453387,0.00034926363,0.006577131,0.0023474663,0.00020113391,0.00075644895,0.0012043639,0.014981756,0.009145343,0.22328639,0.62133646,0.11786885],"study_design_scores_gemma":[0.00018257581,0.00009550003,0.0010142806,0.00037946503,0.000047024852,0.00034074424,0.00026536215,0.007794806,0.007853169,0.03232537,0.94955885,0.00014292366],"about_ca_topic_score_codex":0.024499742,"about_ca_topic_score_gemma":0.0113074565,"teacher_disagreement_score":0.037334204,"about_ca_system_score_codex":0.004031582,"about_ca_system_score_gemma":0.008449797,"threshold_uncertainty_score":0.124895334},"labels":[],"label_agreement":null},{"id":"W2170726882","doi":"10.1093/bioinformatics/btp205","title":"Model-based clustering of array CGH data","year":2009,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomic variations and chromosomal abnormalities","field":"Biochemistry, Genetics and Molecular Biology","cited_by":24,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia; BC Cancer Agency","funders":"","keywords":"Cluster analysis; Comparative genomic hybridization; Computer science; Population; Computational biology; Software; Cluster (spacecraft); Data mining; Pattern recognition (psychology); Artificial intelligence; Biology; Medicine; Genetics; Chromosome","score_opus":0.028346948070858743,"score_gpt":0.25154376086899205,"score_spread":0.2231968127981333,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2170726882","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.041648295,0.00025311756,0.9531879,0.00039751377,0.000035658806,0.00018148992,0.001881368,0.0019770416,0.0004376934],"genre_scores_gemma":[0.45392528,0.00031686496,0.53126657,0.00027671526,0.00010340903,0.00063321186,0.011509102,0.0004186217,0.0015501522],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9974241,0.0009771889,0.00017018887,0.00082856265,0.0004764301,0.0001235051],"domain_scores_gemma":[0.9936225,0.004004695,0.00052686175,0.0009331187,0.00076463347,0.0001483243],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0036521917,0.0011052665,0.0011721819,0.0022283217,0.0007427347,0.0013271536,0.002390067,0.0013232537,0.0012534286],"category_scores_gemma":[0.014872715,0.00055223407,0.0012629349,0.0023997142,0.0009032497,0.0010636506,0.0012685945,0.0013106838,0.0009040892],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00020233766,0.00010944473,0.008601541,0.0002290976,0.00021998804,0.00014917209,0.00018336918,0.9193314,0.005213472,0.0054071452,0.0040632854,0.056289747],"study_design_scores_gemma":[0.000019472362,0.000020662103,0.0013840406,0.000008748101,0.000016945978,0.00006206851,0.000018676306,0.98452026,0.0015293094,0.011662918,0.0007399914,0.00001683705],"about_ca_topic_score_codex":0.007919367,"about_ca_topic_score_gemma":0.009724245,"teacher_disagreement_score":0.007919367,"about_ca_system_score_codex":0.0013785851,"about_ca_system_score_gemma":0.001556562,"threshold_uncertainty_score":0.019314826},"labels":[],"label_agreement":null},{"id":"W2171154262","doi":"10.1093/bioinformatics/btv232","title":"Cypiripi: exact genotyping of <i>CYP2D6</i> using high-throughput sequencing data","year":2015,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Pharmacogenetics and Drug Metabolism","field":"Pharmacology, Toxicology and Pharmaceutics","cited_by":49,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"National Institute of General Medical Sciences; National Human Genome Research Institute","keywords":"Genotyping; Context (archaeology); Pseudogene; Biology; Computational biology; Copy-number variation; Genotype; Computer science; Genetics; Data mining; Algorithm; Gene; Genome","score_opus":0.4461480385382918,"score_gpt":0.47036580555392304,"score_spread":0.02421776701563122,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2171154262","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.04969396,0.0012242716,0.86318326,0.0015916936,0.0002517038,0.00045232283,0.019453138,0.058910083,0.0052396487],"genre_scores_gemma":[0.25478026,0.0007879964,0.6928635,0.001008182,0.00017966368,0.00066132564,0.043977797,0.0027388437,0.0030024443],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99884087,0.0003829362,0.000060437844,0.00036833022,0.00027668942,0.00007067807],"domain_scores_gemma":[0.99779665,0.0011922948,0.00023379856,0.0003358336,0.0003172712,0.00012413194],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00252684,0.0010092272,0.0008960082,0.00087825296,0.00069736456,0.0013662219,0.0016396709,0.0011018992,0.0060775853],"category_scores_gemma":[0.007951858,0.0005320875,0.0007430309,0.0011096172,0.00043007202,0.00082569214,0.000995855,0.0014339467,0.0051318025],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0025796841,0.0004732058,0.042780574,0.0014731224,0.00072249357,0.0010973018,0.00028657005,0.2101962,0.038821924,0.008474849,0.13057117,0.5625228],"study_design_scores_gemma":[0.0002924614,0.00020815301,0.007012959,0.00007101637,0.00009068106,0.0007813138,0.00004714622,0.925543,0.024522085,0.013725713,0.027631283,0.00007421302],"about_ca_topic_score_codex":0.0027478782,"about_ca_topic_score_gemma":0.0034691505,"teacher_disagreement_score":0.0060775853,"about_ca_system_score_codex":0.00065805553,"about_ca_system_score_gemma":0.0016230326,"threshold_uncertainty_score":0.020331562},"labels":[],"label_agreement":null},{"id":"W2171164011","doi":"10.1093/bioinformatics/btg406","title":"JDotter: a Java interface to multiple dotplots generated by dotter","year":2004,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Poxvirus research and outbreaks","field":"Immunology and Microbiology","cited_by":129,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Victoria","funders":"National Institute of Allergy and Infectious Diseases","keywords":"Java; Interface (matter); Computer science; Software; Programming language; Computer graphics (images); Operating system","score_opus":0.015113549004505537,"score_gpt":0.2588652430169957,"score_spread":0.24375169401249017,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2171164011","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0019311713,0.00032226948,0.27678496,0.00023413394,0.0004560814,0.00034811362,0.037668787,0.675542,0.0067125517],"genre_scores_gemma":[0.02284044,0.0006407444,0.48993143,0.0007413609,0.00026477242,0.002002274,0.09409694,0.3583444,0.03113759],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99856037,0.0001879969,0.00019870978,0.00042021752,0.00044130362,0.00019151915],"domain_scores_gemma":[0.996179,0.0016179042,0.00031204912,0.00064274296,0.00087682717,0.0003714048],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0023492991,0.0029313834,0.002614159,0.0032099735,0.0008214571,0.0023249038,0.0050958763,0.001616035,0.28163803],"category_scores_gemma":[0.007295391,0.002067982,0.0021550776,0.0021389113,0.0004993249,0.0030893027,0.0026626696,0.0030106653,0.14511663],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0016972745,0.00033823584,0.002083457,0.0026655612,0.00029449788,0.0006341015,0.00048638144,0.0016559165,0.030493729,0.0034281097,0.7290079,0.22721481],"study_design_scores_gemma":[0.001154517,0.00028467117,0.009578515,0.0007142625,0.0001880376,0.0014954538,0.00017867489,0.045251925,0.08662546,0.0134341745,0.8404504,0.000643937],"about_ca_topic_score_codex":0.0013637982,"about_ca_topic_score_gemma":0.0018577371,"teacher_disagreement_score":0.28163803,"about_ca_system_score_codex":0.0008276847,"about_ca_system_score_gemma":0.0010937607,"threshold_uncertainty_score":0.94217306},"labels":[],"label_agreement":null},{"id":"W2171259901","doi":"10.1093/bioinformatics/btu556","title":"Prediction of Indel flanking regions in protein sequences using a variable-order Markov model","year":2014,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Machine Learning in Bioinformatics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Concordia University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Indel; Flanking maneuver; Markov chain; Variable (mathematics); Markov model; Computer science; Markov chain Monte Carlo; Order (exchange); Genetics; Mathematics; Statistics; Artificial intelligence; Biology; Gene; Bayesian probability; Geography","score_opus":0.018877510514529117,"score_gpt":0.24661550447183547,"score_spread":0.22773799395730634,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2171259901","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.3180895,0.00075209857,0.67719746,0.0002567934,0.00003318892,0.000064239444,0.00070640084,0.002151911,0.0007485036],"genre_scores_gemma":[0.8758524,0.00045861668,0.12069743,0.00011071916,0.000042273543,0.00008481994,0.0016909862,0.00007710447,0.0009855871],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99967563,0.00009454984,0.00002443347,0.00009402427,0.000075440534,0.000035969],"domain_scores_gemma":[0.99784017,0.0015308036,0.0002640567,0.00010382863,0.00018108032,0.00008004911],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00093034527,0.00041502173,0.00064051774,0.00086288847,0.00032327647,0.0004370983,0.0006650866,0.00074560364,0.0009838088],"category_scores_gemma":[0.0028001014,0.00026746243,0.00063490646,0.0005662773,0.00028381057,0.000777059,0.00038190908,0.0007769127,0.0005706167],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010693822,0.0002755809,0.042745948,0.00027826114,0.00017357446,0.0005870599,0.00013030139,0.7769694,0.034337763,0.005182698,0.0021228895,0.13612711],"study_design_scores_gemma":[0.0000061492633,0.000044166696,0.0010044646,0.0000048547727,0.0000130318795,0.00005525246,0.0000044928343,0.9959169,0.0017149984,0.0010856399,0.00014423276,0.000005791791],"about_ca_topic_score_codex":0.0028059047,"about_ca_topic_score_gemma":0.0037290493,"teacher_disagreement_score":0.0028059047,"about_ca_system_score_codex":0.00033861864,"about_ca_system_score_gemma":0.00086269167,"threshold_uncertainty_score":0.005579114},"labels":[],"label_agreement":null},{"id":"W2171475724","doi":"10.1093/bioinformatics/btl629","title":"Assembling millions of short DNA sequences using SSAKE","year":2006,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":506,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"BC Cancer Agency","funders":"BC Cancer Agency; Michael Smith Health Research BC","keywords":"DNA sequencing; Sanger sequencing; Computational biology; Hybrid genome assembly; Biology; Sequence assembly; Genome; Software; Sequence (biology); k-mer; Leverage (statistics); Deep sequencing; Genetics; DNA; Computer science; Reference genome; Gene; Transcriptome; Artificial intelligence","score_opus":0.02626783783145639,"score_gpt":0.26256828783320746,"score_spread":0.23630045000175107,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2171475724","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.10262821,0.0014334306,0.8670198,0.00028760484,0.00024468277,0.0002675462,0.003226757,0.02059858,0.0042933584],"genre_scores_gemma":[0.12824106,0.00064889644,0.8556283,0.00018094847,0.00005429872,0.00046937278,0.009549356,0.0014105773,0.0038171748],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9993561,0.00009544373,0.00009421589,0.0002348501,0.0001732849,0.000046055036],"domain_scores_gemma":[0.9982609,0.0007779068,0.0002367761,0.00035582748,0.00025259575,0.00011590302],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00080827955,0.00094738073,0.001092639,0.0011020268,0.00075597956,0.0010483479,0.00083352893,0.00072956836,0.006112167],"category_scores_gemma":[0.0040326887,0.0006746189,0.001116904,0.0011879541,0.0004490108,0.0014122912,0.0012681518,0.001330711,0.0092563],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014065984,0.00015965402,0.005075798,0.0017217208,0.0002431811,0.0006176047,0.00080108765,0.0069079218,0.7017155,0.0048539424,0.008126274,0.2683707],"study_design_scores_gemma":[0.0001366108,0.0004972334,0.0044806995,0.00017904876,0.00019701112,0.00143079,0.0002715543,0.07781446,0.7972476,0.009730401,0.10788245,0.00013219602],"about_ca_topic_score_codex":0.00024711573,"about_ca_topic_score_gemma":0.00081735506,"teacher_disagreement_score":0.006112167,"about_ca_system_score_codex":0.00024615822,"about_ca_system_score_gemma":0.00041252488,"threshold_uncertainty_score":0.020447195},"labels":[],"label_agreement":null},{"id":"W2171675548","doi":"10.1093/bioinformatics/btt729","title":"Site-heterogeneous mutation-selection models within the PhyloBayes-MPI package","year":2013,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":89,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"","keywords":"Computer science; Software; Focus (optics); Scalability; Selection (genetic algorithm); Interface (matter); Model selection; Data mining; Mutation; Variety (cybernetics); Dirichlet process; Bayesian probability; Theoretical computer science; Machine learning; Artificial intelligence; Programming language; Biology; Genetics; Parallel computing; Database; Gene","score_opus":0.010939629712216885,"score_gpt":0.21276376770997538,"score_spread":0.2018241379977585,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2171675548","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0107302135,0.00039895647,0.8887529,0.0009927166,0.00023515748,0.00018848227,0.010520731,0.080722846,0.0074581066],"genre_scores_gemma":[0.09737002,0.00084079686,0.83852965,0.0005292427,0.00024218157,0.002388828,0.019844884,0.03429701,0.005957357],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9981583,0.00072615535,0.00012798356,0.00030119752,0.00049194874,0.00019443437],"domain_scores_gemma":[0.99578923,0.0021334211,0.00029452893,0.0008568228,0.00063551724,0.00029051182],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0051761023,0.0026812982,0.0027357265,0.0016638476,0.0018507359,0.002531422,0.0088604605,0.0021859768,0.02148998],"category_scores_gemma":[0.014809223,0.0019659214,0.002344517,0.0029534341,0.0010969526,0.0031434218,0.0036003205,0.005390263,0.0173432],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013740706,0.0004522436,0.010534055,0.002185227,0.00090834516,0.0009404322,0.0014154284,0.4670433,0.014331171,0.16542456,0.21703213,0.11835898],"study_design_scores_gemma":[0.0002960725,0.0000660709,0.00135516,0.0001588907,0.00009482578,0.00020178342,0.0000639875,0.8348324,0.0050860546,0.09210357,0.06558255,0.00015856978],"about_ca_topic_score_codex":0.009031018,"about_ca_topic_score_gemma":0.005393037,"teacher_disagreement_score":0.02148998,"about_ca_system_score_codex":0.0014297209,"about_ca_system_score_gemma":0.0033041334,"threshold_uncertainty_score":0.07189113},"labels":[],"label_agreement":null},{"id":"W2171821554","doi":"10.1093/bioinformatics/btn073","title":"SubSeqer: a graph-based approach for the detection and identification of repetitive elements in low-complexity sequences","year":2008,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"SickKids Foundation; Hospital for Sick Children; University of Toronto","funders":"Hospital for Sick Children; Heart and Stroke Foundation of Canada","keywords":"Computer science; Visualization; Identification (biology); Palette (painting); Graph; Computational biology; Matching (statistics); Computational complexity theory; Interspersed repeat; Theoretical computer science; Biology; Algorithm; Data mining; Genome; Mathematics; Genetics","score_opus":0.030244628776910823,"score_gpt":0.24561592944355418,"score_spread":0.21537130066664334,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2171821554","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.021103712,0.0003581177,0.9123911,0.00024991942,0.00009764014,0.0003086823,0.009200426,0.054529045,0.0017612837],"genre_scores_gemma":[0.0406544,0.0004077068,0.93842626,0.00014396783,0.000033835215,0.0004107497,0.01421689,0.003520038,0.0021861957],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99909246,0.00023625328,0.000084353574,0.00024950298,0.00028358318,0.000053920223],"domain_scores_gemma":[0.9976549,0.0013157638,0.00028970581,0.00033615628,0.00026702517,0.00013645222],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012873092,0.0019218618,0.0013145368,0.005539395,0.0010927395,0.001437513,0.0020176251,0.0009863987,0.009043248],"category_scores_gemma":[0.0044649113,0.00076021836,0.001502963,0.0032270562,0.00058571546,0.0017564328,0.0015065868,0.0014841371,0.0034284226],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001325452,0.0004296175,0.0069122375,0.0033474744,0.0008605295,0.0013478871,0.0015168374,0.026896792,0.31013533,0.026539458,0.056943707,0.5637447],"study_design_scores_gemma":[0.00036593032,0.0005555326,0.011378391,0.00025123244,0.000389145,0.002835304,0.00055763556,0.5398493,0.171317,0.07283966,0.19920671,0.00045420707],"about_ca_topic_score_codex":0.0021504858,"about_ca_topic_score_gemma":0.003376014,"teacher_disagreement_score":0.009043248,"about_ca_system_score_codex":0.0005978175,"about_ca_system_score_gemma":0.0010299118,"threshold_uncertainty_score":0.030252695},"labels":[],"label_agreement":null},{"id":"W2171832486","doi":"10.1093/bioinformatics/btl017","title":"Polynomial model approach for resynchronization analysis of cell-cycle gene expression data","year":2006,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene Regulatory Network Analysis","field":"Biochemistry, Genetics and Molecular Biology","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Computational biology; Microarray analysis techniques; Gene; Identification (biology); Cell cycle; Computer science; Expression (computer science); Gene expression; Biology; Functional genomics; Microarray; Gene expression profiling; Genomics; Algorithm; Genetics; Genome","score_opus":0.015229784318238205,"score_gpt":0.23799432060265668,"score_spread":0.22276453628441847,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2171832486","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.010999482,0.000059662212,0.98812324,0.00006133055,0.000006804323,0.000017775437,0.00012955985,0.0004038782,0.00019822305],"genre_scores_gemma":[0.50469977,0.0005391889,0.48831913,0.000102809485,0.00006241053,0.00030892316,0.0018380696,0.00034839823,0.0037813492],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99958915,0.00011123371,0.000017672926,0.000116329604,0.000121732184,0.000043849923],"domain_scores_gemma":[0.9990994,0.00051926856,0.00011949945,0.00011287597,0.0001204599,0.000028629365],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007851435,0.0006689366,0.00064013276,0.0008901376,0.00030151993,0.00043763738,0.00093953736,0.0005259905,0.0017788986],"category_scores_gemma":[0.002818517,0.000257401,0.0007694484,0.0012099394,0.0003149576,0.00068898685,0.00046118096,0.0010225509,0.00078429404],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002116856,0.000069046815,0.0016518213,0.00012762447,0.00006855477,0.0001506942,0.00010809759,0.81819534,0.027551956,0.013307916,0.001315665,0.13724169],"study_design_scores_gemma":[0.0000027396025,0.0000102829545,0.0002213359,8.5971266e-7,0.0000031201578,0.000011460748,0.0000039885435,0.99701154,0.001050698,0.0013941288,0.00028610916,0.0000036473364],"about_ca_topic_score_codex":0.006029769,"about_ca_topic_score_gemma":0.006088125,"teacher_disagreement_score":0.006029769,"about_ca_system_score_codex":0.0007571569,"about_ca_system_score_gemma":0.0008216642,"threshold_uncertainty_score":0.0119892955},"labels":[],"label_agreement":null},{"id":"W2171977579","doi":"10.1093/bioinformatics/btt485","title":"iBMQ: a R/Bioconductor package for integrated Bayesian modeling of eQTL data","year":2013,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Montreal Clinical Research Institute","funders":"National Human Genome Research Institute; National Institutes of Health","keywords":"Bioconductor; Expression quantitative trait loci; R package; Computer science; Bayesian probability; Markov chain Monte Carlo; Univariate; Data mining; Computational biology; Biology; Machine learning; Artificial intelligence; Single-nucleotide polymorphism; Gene; Genetics; Computational science","score_opus":0.04416219096831347,"score_gpt":0.28701331923114004,"score_spread":0.24285112826282657,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2171977579","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0021050153,0.002221184,0.5942948,0.001524455,0.0006503772,0.00047551983,0.20348375,0.1885285,0.0067163566],"genre_scores_gemma":[0.027818494,0.0019128553,0.66175467,0.0020717713,0.0004919269,0.0059641832,0.16198337,0.12934515,0.008657526],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9939361,0.0021450669,0.00047131698,0.0016702428,0.0014546161,0.00032262053],"domain_scores_gemma":[0.9838187,0.010317982,0.0013488203,0.0022045402,0.0017715988,0.00053825916],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0130601665,0.004876665,0.005155007,0.0054004467,0.0015277831,0.0046760333,0.007681073,0.0027482219,0.115293264],"category_scores_gemma":[0.040201847,0.0029334482,0.004630274,0.006921233,0.0016257641,0.0027608948,0.005167689,0.005388952,0.07277671],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007008143,0.00010185529,0.006333827,0.0048781694,0.0023791469,0.0004881278,0.00045367313,0.02013929,0.0033293061,0.029590888,0.84937805,0.08222684],"study_design_scores_gemma":[0.0011353625,0.00019460234,0.009201236,0.0012001075,0.0014507155,0.00097371166,0.00014262635,0.16370094,0.004825979,0.2075852,0.6091529,0.0004366658],"about_ca_topic_score_codex":0.008971779,"about_ca_topic_score_gemma":0.010690131,"teacher_disagreement_score":0.115293264,"about_ca_system_score_codex":0.001484451,"about_ca_system_score_gemma":0.006022063,"threshold_uncertainty_score":0.38569438},"labels":[],"label_agreement":null},{"id":"W2172174827","doi":"10.1093/bioinformatics/btv356","title":"EXIMS: an improved data analysis pipeline based on a new peak picking method for EXploring Imaging Mass Spectrometry data","year":2015,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Mass Spectrometry Techniques and Applications","field":"Chemistry","cited_by":35,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"IONICS Mass Spectrometry (Canada)","funders":"State Government of Victoria; University of Melbourne; Vanderbilt University","keywords":"Normalization (sociology); Computer science; Cluster analysis; Mass spectrometry imaging; Data mining; Pattern recognition (psychology); Pipeline (software); Spatial analysis; Segmentation; Artificial intelligence; Mass spectrometry; Mathematics; Chemistry","score_opus":0.1579918128006178,"score_gpt":0.36703765716931597,"score_spread":0.20904584436869816,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2172174827","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0064451126,0.00015017415,0.96505684,0.0001138459,0.00004574579,0.00013339175,0.0013290712,0.026389655,0.00033621365],"genre_scores_gemma":[0.018084768,0.0001227022,0.97600013,0.00008701535,0.0000221738,0.00016424332,0.0034068213,0.00094529474,0.0011668486],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99899036,0.00009617645,0.00011017897,0.00032208455,0.00041898797,0.00006224299],"domain_scores_gemma":[0.99881625,0.00025305944,0.00013396618,0.00023933197,0.0004857182,0.000071732866],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014352506,0.002175348,0.0009648725,0.0031966434,0.00074429414,0.0016286776,0.0021532294,0.00087164884,0.0047340747],"category_scores_gemma":[0.002137253,0.00076844776,0.0014647773,0.0020711392,0.00062015717,0.0021530872,0.0018735039,0.0014755467,0.0035467967],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008851196,0.00027708794,0.0046050157,0.0005537498,0.00028294616,0.0005952006,0.00035747132,0.01821031,0.2351417,0.0035969843,0.023124263,0.7123701],"study_design_scores_gemma":[0.00014555625,0.00027368256,0.007978251,0.000063641535,0.00009333224,0.0012494851,0.00019234692,0.6757824,0.2613592,0.0066019013,0.045993987,0.00026621958],"about_ca_topic_score_codex":0.002780978,"about_ca_topic_score_gemma":0.0038638772,"teacher_disagreement_score":0.0047340747,"about_ca_system_score_codex":0.00058665633,"about_ca_system_score_gemma":0.0013505934,"threshold_uncertainty_score":0.015837014},"labels":[],"label_agreement":null},{"id":"W2172269857","doi":"10.1093/bioinformatics/btl188","title":"FCP: functional coverage of the proteome by structures","year":2006,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Protein Structure and Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":16,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Instituto de Salud Carlos III; International Union of Biochemistry and Molecular Biology; Universitat Pompeu Fabra","keywords":"Proteome; Computer science; Population; Computational biology; Bioinformatics; Biology","score_opus":0.0032564763635078114,"score_gpt":0.1856639181624849,"score_spread":0.18240744179897708,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2172269857","genre_codex":"dataset","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.21029837,0.008585105,0.1984187,0.002492892,0.00020690792,0.0003982424,0.44585118,0.10545361,0.028294941],"genre_scores_gemma":[0.44659775,0.0046130978,0.14739697,0.00053991395,0.00035573306,0.0010235936,0.38791615,0.007359872,0.004197036],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99811685,0.00024581162,0.00009102701,0.00048301916,0.0008751011,0.0001881989],"domain_scores_gemma":[0.99388677,0.002846566,0.00091667497,0.0010650452,0.0008171228,0.00046777536],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0030448067,0.002665031,0.0016219444,0.008578304,0.0011280423,0.002150288,0.0024622174,0.00096377695,0.012223886],"category_scores_gemma":[0.011946286,0.00042736332,0.0009770436,0.009636682,0.00062288914,0.002628622,0.0019438834,0.0011963061,0.0056803157],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0029886742,0.0003059095,0.08933625,0.006041129,0.00093932374,0.0008998732,0.0008317662,0.020532548,0.052502472,0.023742732,0.36380342,0.43807587],"study_design_scores_gemma":[0.00035101673,0.00064193545,0.24666308,0.0008994353,0.0008867906,0.008489766,0.00054046646,0.22752233,0.120116636,0.0892659,0.3041573,0.00046541204],"about_ca_topic_score_codex":0.0045225094,"about_ca_topic_score_gemma":0.0015093323,"teacher_disagreement_score":0.012223886,"about_ca_system_score_codex":0.00097203214,"about_ca_system_score_gemma":0.0010863479,"threshold_uncertainty_score":0.04089302},"labels":[],"label_agreement":null},{"id":"W2172358623","doi":"10.1093/bioinformatics/btv693","title":"Genefu: an R/Bioconductor package for computation of gene expression-based signatures in breast cancer","year":2015,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":410,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Princess Margaret Cancer Centre; University of Toronto; University Health Network","funders":"National Cancer Institute; Cancer Research Society; Canadian Institutes of Health Research; Instituto de Salud Carlos III; Banco Bilbao Vizcaya Argentaria; Fondation Brain Canada","keywords":"Bioconductor; Compendium; Subtyping; Computer science; R package; Breast cancer; Source code; Computational biology; Data mining; Bioinformatics; Cancer; Gene; Biology; Programming language; Genetics","score_opus":0.031093044095763682,"score_gpt":0.30028275011701877,"score_spread":0.2691897060212551,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2172358623","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.004503896,0.0047249473,0.32923323,0.0021771153,0.0010991893,0.00056806934,0.2724838,0.3721277,0.013082006],"genre_scores_gemma":[0.04705314,0.0033013255,0.501153,0.002353496,0.00039347797,0.0061891917,0.30279732,0.118740916,0.018018216],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99680364,0.0009417438,0.00028902447,0.00088479946,0.0008084525,0.00027228397],"domain_scores_gemma":[0.9933709,0.003148016,0.00056303135,0.0010844059,0.0014880892,0.00034552353],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0049215974,0.0034091093,0.0038880815,0.004686331,0.0011970755,0.0030688755,0.006291312,0.0017232044,0.07262591],"category_scores_gemma":[0.01833904,0.0016212936,0.0028615734,0.006439942,0.0009837315,0.0021714463,0.0035170391,0.0033883145,0.08145295],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007485022,0.00005990236,0.0039643073,0.0040004016,0.0010564874,0.00027473064,0.00031926928,0.0064232894,0.0034869441,0.009514165,0.89673626,0.07341562],"study_design_scores_gemma":[0.000613947,0.00019203957,0.011637441,0.0010836426,0.0009098543,0.0008947784,0.00016274562,0.09261557,0.011489379,0.05323693,0.8268584,0.00030523608],"about_ca_topic_score_codex":0.0070724934,"about_ca_topic_score_gemma":0.007580501,"teacher_disagreement_score":0.07262591,"about_ca_system_score_codex":0.0013614958,"about_ca_system_score_gemma":0.0050687944,"threshold_uncertainty_score":0.24295783},"labels":[],"label_agreement":null},{"id":"W2176522635","doi":"10.1093/bioinformatics/btv616","title":"IsoMIF Finder: online detection of binding site molecular interaction field similarities","year":2015,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Computational Drug Discovery Methods","field":"Computer Science","cited_by":26,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Sherbrooke","funders":"Natural Sciences and Engineering Research Council of Canada; Québec Consortium for Drug Discovery","keywords":"Plug-in; Identification (biology); Computer science; Java; Session (web analytics); Benchmark (surveying); Interface (matter); Function (biology); Field (mathematics); Protein Data Bank (RCSB PDB); Computational biology; Data mining; World Wide Web; Operating system; Biology; Genetics; Cartography","score_opus":0.040824633121552156,"score_gpt":0.3105782690473786,"score_spread":0.26975363592582646,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2176522635","genre_codex":"software","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.102681845,0.004235292,0.2544824,0.00115611,0.000605641,0.0012295388,0.20923232,0.39978245,0.026594397],"genre_scores_gemma":[0.20177452,0.001440953,0.47941664,0.00045694795,0.0002069642,0.00262503,0.27940273,0.023097754,0.011578474],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99873847,0.00011863885,0.00010004017,0.00032081894,0.0005502225,0.00017196094],"domain_scores_gemma":[0.9984332,0.00058755977,0.00029619233,0.00023188573,0.00023487411,0.00021624229],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0021962454,0.0022864682,0.002685987,0.008586963,0.0015953705,0.002638287,0.0032736112,0.0016794078,0.06736419],"category_scores_gemma":[0.0063892193,0.0010192989,0.0016978469,0.004134119,0.0003868109,0.0026265774,0.002817189,0.0016064597,0.019743439],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0029589992,0.00046168163,0.016769916,0.0029818516,0.00059870386,0.0012515875,0.0005457409,0.0036792376,0.042610932,0.011368432,0.6535724,0.2632005],"study_design_scores_gemma":[0.0018763714,0.0007319469,0.04965153,0.0008369051,0.00065076683,0.0063356115,0.0006958497,0.21276417,0.16389784,0.038757894,0.52307725,0.0007238936],"about_ca_topic_score_codex":0.0015774093,"about_ca_topic_score_gemma":0.0026335309,"teacher_disagreement_score":0.06736419,"about_ca_system_score_codex":0.0011767839,"about_ca_system_score_gemma":0.002070608,"threshold_uncertainty_score":0.22535563},"labels":[],"label_agreement":null},{"id":"W2177550115","doi":"10.1093/bioinformatics/btv597","title":"PDID: database of molecular-level putative protein–drug interactions in the structural human proteome","year":2015,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":48,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"National Institutes of Health; U.S. National Library of Medicine; Natural Sciences and Engineering Research Council of Canada; National Natural Science Foundation of China","keywords":"Proteome; Human proteome project; Computational biology; Drug; Computer science; Database; Bioinformatics; Biology; Proteomics; Genetics; Pharmacology; Gene","score_opus":0.03435201254339759,"score_gpt":0.2947702654347224,"score_spread":0.2604182528913248,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2177550115","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.015921798,0.0038649458,0.017441843,0.00051844714,0.00008135905,0.00022103201,0.93683916,0.018658081,0.0064532836],"genre_scores_gemma":[0.021412263,0.0014551203,0.017335765,0.00015189446,0.000028669047,0.00022845724,0.95827717,0.00045721504,0.0006533631],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9993242,0.00011050732,0.000109333254,0.00020780659,0.0001931013,0.000055049983],"domain_scores_gemma":[0.998563,0.00051973824,0.0003339719,0.00022717551,0.00016596851,0.00019010973],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011018257,0.0022267185,0.002259128,0.0044343127,0.0011397765,0.0021210893,0.0023599884,0.001536694,0.027176954],"category_scores_gemma":[0.0038188957,0.0009233957,0.0009810275,0.006736485,0.00033481664,0.0019059824,0.0018656045,0.0012805559,0.016486328],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0041833394,0.00035080925,0.020932162,0.011660684,0.00059481873,0.0017598956,0.0002575046,0.010869922,0.024594424,0.013586852,0.8268862,0.084323555],"study_design_scores_gemma":[0.0014582484,0.0005825621,0.035310883,0.00094078213,0.0007851692,0.0042580143,0.00027476842,0.034622278,0.023464577,0.020412402,0.8776845,0.00020596266],"about_ca_topic_score_codex":0.0015957261,"about_ca_topic_score_gemma":0.0020612315,"teacher_disagreement_score":0.027176954,"about_ca_system_score_codex":0.0011703131,"about_ca_system_score_gemma":0.003098019,"threshold_uncertainty_score":0.09091592},"labels":[],"label_agreement":null},{"id":"W2177764460","doi":"10.1093/bioinformatics/btv619","title":"ancGWAS: a post genome-wide association study method for interaction, pathway and ancestry analysis in homogeneous and admixed populations","year":2015,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":34,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Common Fund; National Eye Institute; National Human Genome Research Institute; African Institute for Mathematical Sciences; Division of Mathematical Sciences; Government of Canada; National Institutes of Health; University of Cape Town; International Development Research Centre","keywords":"Genome-wide association study; Linkage disequilibrium; Genetic association; Interactome; Biology; Computational biology; Breast cancer; Genetics; Single-nucleotide polymorphism; Gene; Cancer; Genotype","score_opus":0.044067027943376816,"score_gpt":0.33288749626912667,"score_spread":0.28882046832574987,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2177764460","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.008162709,0.00018791766,0.9713183,0.00022673552,0.00015945733,0.00020049607,0.0036718166,0.015263544,0.0008090374],"genre_scores_gemma":[0.096697226,0.0002612949,0.8846405,0.00029359746,0.00010959244,0.001664752,0.008620307,0.0048783063,0.002834465],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9987525,0.0007255293,0.00005664988,0.0002308494,0.00016958559,0.000064925865],"domain_scores_gemma":[0.9956457,0.003311149,0.0001746426,0.0004547566,0.0002854054,0.00012827561],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004168779,0.0011764847,0.0011082777,0.0020055359,0.0009840422,0.0012527335,0.0023140647,0.0011436881,0.023593087],"category_scores_gemma":[0.015127241,0.0008462641,0.0023898669,0.0017394894,0.000698438,0.0010241503,0.0018630745,0.0019055696,0.0040214714],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0017656578,0.00040519176,0.038570147,0.0018613278,0.0046447627,0.0029167994,0.001497103,0.20808381,0.014576211,0.09706091,0.14853975,0.4800783],"study_design_scores_gemma":[0.0006070836,0.00015174905,0.0059847673,0.00011502939,0.00044873983,0.0007615384,0.00013223824,0.83789253,0.00338856,0.08930816,0.061101105,0.000108480766],"about_ca_topic_score_codex":0.0057616695,"about_ca_topic_score_gemma":0.009134071,"teacher_disagreement_score":0.023593087,"about_ca_system_score_codex":0.00043811154,"about_ca_system_score_gemma":0.001925794,"threshold_uncertainty_score":0.0789268},"labels":[],"label_agreement":null},{"id":"W2179127114","doi":"10.1093/bioinformatics/btv557","title":"Cytoscape.js: a graph theory library for visualisation and analysis","year":2015,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":721,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"National Institute of General Medical Sciences; National Human Genome Research Institute; National Institutes of Health; National Institute for Health and Care Research","keywords":"Computer science; JavaScript; Visualization; Documentation; Source code; Graph; Programming language; Software; Web application; Graph drawing; World Wide Web; Information retrieval; Data mining; Theoretical computer science","score_opus":0.014564893441759852,"score_gpt":0.24308882647624352,"score_spread":0.22852393303448368,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2179127114","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.002608378,0.002314099,0.4342207,0.0017696866,0.0011963444,0.0004686712,0.16589586,0.37649205,0.015034283],"genre_scores_gemma":[0.030387813,0.0036730897,0.5199459,0.0021815875,0.0008733785,0.002309956,0.30470255,0.11603828,0.019887492],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99817383,0.00031065734,0.0001450608,0.00042342168,0.00082433247,0.00012260544],"domain_scores_gemma":[0.99568474,0.0018507027,0.00035944345,0.0007326837,0.0009055817,0.00046699497],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002157151,0.0022704438,0.0024758773,0.0079660565,0.0018539786,0.0032483772,0.0058343024,0.00214844,0.1166176],"category_scores_gemma":[0.01175748,0.0019310842,0.0022085558,0.0052284026,0.0007042358,0.0048592477,0.003313399,0.005108802,0.07486907],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00019450263,0.00009058722,0.0009369985,0.0043902295,0.00046179022,0.00047674193,0.0004071753,0.004351578,0.023693295,0.016438842,0.8378126,0.110745534],"study_design_scores_gemma":[0.00021168958,0.000048373604,0.0016209295,0.00053790986,0.00020778926,0.001049849,0.00011691981,0.033005197,0.031245437,0.07617369,0.8554478,0.0003343252],"about_ca_topic_score_codex":0.004599222,"about_ca_topic_score_gemma":0.0065105055,"teacher_disagreement_score":0.1166176,"about_ca_system_score_codex":0.00088414294,"about_ca_system_score_gemma":0.0029753,"threshold_uncertainty_score":0.39012468},"labels":[],"label_agreement":null},{"id":"W2180318265","doi":"10.1093/bioinformatics/btv487","title":"EPGA2: memory-efficient <i>de novo</i> assembler","year":2015,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":25,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Saskatchewan","funders":"Program for New Century Excellent Talents in University; National Natural Science Foundation of China","keywords":"Computer science; Sequence assembly; Pipeline (software); Contig; De Bruijn sequence; De Bruijn graph; Construct (python library); Parallels; Genome; Reference genome; Graph; Theoretical computer science; Parallel computing; Programming language; Biology; Genetics; Gene","score_opus":0.022655545222036964,"score_gpt":0.24651608930198826,"score_spread":0.2238605440799513,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2180318265","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.017351456,0.00048581557,0.69811803,0.00030639803,0.0002966728,0.0003696372,0.0034573583,0.2736438,0.0059707947],"genre_scores_gemma":[0.06431088,0.00042880495,0.8688828,0.0006740133,0.00011961089,0.0010035181,0.02190332,0.032240473,0.010436684],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9990602,0.00015978474,0.00008039371,0.00035100253,0.00023875911,0.0001097908],"domain_scores_gemma":[0.99844605,0.00044897597,0.00012323495,0.00053940713,0.0003445895,0.00009782381],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011690414,0.0028561186,0.0010778552,0.0013970642,0.0013158044,0.0019834605,0.0032680442,0.0010729148,0.011786404],"category_scores_gemma":[0.004522267,0.001504609,0.0014883036,0.0015683703,0.0006676195,0.0020688067,0.0023472803,0.0024097865,0.012609135],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00206538,0.00040121173,0.0069149546,0.0014136409,0.0004054373,0.00091509806,0.00081054337,0.029178511,0.13837288,0.011558768,0.2620817,0.5458819],"study_design_scores_gemma":[0.00080784876,0.0004694299,0.006603714,0.000178435,0.00029713922,0.0016091998,0.00023849182,0.4588234,0.23969533,0.016013218,0.27484092,0.00042283622],"about_ca_topic_score_codex":0.0032525903,"about_ca_topic_score_gemma":0.0035824806,"teacher_disagreement_score":0.011786404,"about_ca_system_score_codex":0.00067090016,"about_ca_system_score_gemma":0.0012466664,"threshold_uncertainty_score":0.039429426},"labels":[],"label_agreement":null},{"id":"W2180921306","doi":"10.1093/bioinformatics/btv587","title":"CSSSCL: a python package that uses combined sequence similarity scores for accurate taxonomic classification of long and short sequence reads","year":2015,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ontario Institute for Cancer Research","funders":"Government of Ontario; Ontario Institute for Cancer Research","keywords":"Python (programming language); Sequence (biology); Computer science; Similarity (geometry); R package; Computational biology; Artificial intelligence; Biology; Programming language; Genetics","score_opus":0.14316428405831452,"score_gpt":0.3136729286223924,"score_spread":0.17050864456407785,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2180921306","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.009795403,0.00014479525,0.33603996,0.00035814857,0.00026574178,0.00089719845,0.074204996,0.57388055,0.0044132653],"genre_scores_gemma":[0.06092689,0.00026550447,0.6846606,0.001048921,0.00017846243,0.0039968337,0.15616421,0.08368832,0.009070247],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99804485,0.00027493737,0.00020742559,0.00054483156,0.0007328902,0.00019512014],"domain_scores_gemma":[0.99739385,0.0010102411,0.00037303253,0.00036710245,0.0005890987,0.0002666312],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0033597688,0.0026413426,0.0015077846,0.0034059952,0.0014596928,0.0020324006,0.0040018763,0.00096137135,0.030658554],"category_scores_gemma":[0.009163749,0.0012901083,0.0025258064,0.0023411557,0.0011768703,0.0025950873,0.003953039,0.002826783,0.024557708],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011620986,0.0004274443,0.013557482,0.0027488214,0.0007916481,0.0004870574,0.00077327713,0.014791217,0.033108193,0.005508411,0.6692913,0.25735304],"study_design_scores_gemma":[0.0007044999,0.00036827306,0.0196427,0.000404627,0.00022319652,0.00072677305,0.00026166785,0.6044565,0.06979806,0.026576217,0.2762658,0.0005717671],"about_ca_topic_score_codex":0.0057185525,"about_ca_topic_score_gemma":0.0063580945,"teacher_disagreement_score":0.030658554,"about_ca_system_score_codex":0.0011121681,"about_ca_system_score_gemma":0.003638769,"threshold_uncertainty_score":0.10256302},"labels":[],"label_agreement":null},{"id":"W2191136822","doi":"10.1093/bioinformatics/btv662","title":"rHAT: fast alignment of noisy long reads with regional hashing","year":2015,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":48,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National High-tech Research and Development Program; University of Toronto; National Natural Science Foundation of China","keywords":"Computer science; Hash function; Bottleneck; Hash table; Source code; Theoretical computer science; Data mining; Computational biology; Algorithm; Biology; Programming language","score_opus":0.024469740640056834,"score_gpt":0.23372364095691464,"score_spread":0.2092539003168578,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2191136822","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.025299948,0.0018861208,0.87571234,0.0002979205,0.00039590665,0.00037271398,0.0072441744,0.08652196,0.0022689411],"genre_scores_gemma":[0.111611165,0.00047211524,0.8611929,0.0003112728,0.00018236352,0.0005980585,0.018262276,0.0047331285,0.0026367642],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99741983,0.0004751911,0.0002457379,0.0009512286,0.0007658353,0.00014214624],"domain_scores_gemma":[0.99721503,0.0010147195,0.00044034023,0.0006385836,0.0005261218,0.00016518429],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0021933133,0.0015363436,0.0014001696,0.0013593108,0.0009766876,0.0012849842,0.002662681,0.0012337868,0.008229319],"category_scores_gemma":[0.0084433835,0.00092562556,0.0013405897,0.0021307576,0.00069453614,0.0019868275,0.0021756669,0.0018184591,0.009619068],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0025830343,0.0003105251,0.008054066,0.0033692578,0.0006428519,0.0012259402,0.000898547,0.04728563,0.21903735,0.012546167,0.08980556,0.6142412],"study_design_scores_gemma":[0.0006318201,0.0011844708,0.006710199,0.00028739648,0.00028357658,0.0022988385,0.00030261424,0.6440665,0.2307183,0.025611144,0.08742611,0.00047911587],"about_ca_topic_score_codex":0.0012788312,"about_ca_topic_score_gemma":0.0017273542,"teacher_disagreement_score":0.008229319,"about_ca_system_score_codex":0.0004504182,"about_ca_system_score_gemma":0.0014614125,"threshold_uncertainty_score":0.027529776},"labels":[],"label_agreement":null},{"id":"W2206213513","doi":"10.1093/bioinformatics/btx215","title":"A Bayesian group sparse multi-task regression model for imaging genetics","year":2017,"lang":"en","type":"preprint","venue":"Bioinformatics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University; University of Victoria","funders":"National Institute on Aging; National Institutes of Health","keywords":"Imaging genetics; Computer science; Bayesian probability; Neuroimaging; Estimator; Inference; Bayesian inference; Artificial intelligence; Bayesian hierarchical modeling; Point estimation; Machine learning; Pattern recognition (psychology); Statistics; Mathematics; Biology","score_opus":0.04123079165782697,"score_gpt":0.31617125678560093,"score_spread":0.274940465127774,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2206213513","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0048301495,0.00038886286,0.98983127,0.001415184,0.00006820881,0.0001294756,0.0010425315,0.0004153923,0.0018789105],"genre_scores_gemma":[0.30465636,0.0018679505,0.65442365,0.0016299275,0.00070006313,0.002549238,0.005440662,0.0006066624,0.028125558],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9968845,0.001629896,0.00010077603,0.0007162715,0.00044747995,0.00022111004],"domain_scores_gemma":[0.9935947,0.004588802,0.000599141,0.00032902058,0.00068620394,0.00020220326],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008105721,0.0016670234,0.0019694967,0.0014947135,0.00057029043,0.0018320829,0.004869649,0.0039027208,0.01137101],"category_scores_gemma":[0.016364826,0.0011077784,0.0018585878,0.0023470314,0.0019298668,0.0020006686,0.0020311938,0.0037655158,0.0039168936],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002782779,0.00013810837,0.0033686084,0.00031623075,0.00021465354,0.0003236789,0.0003152643,0.66221166,0.0015479054,0.26193973,0.011959685,0.057386175],"study_design_scores_gemma":[0.00005412566,0.00003434398,0.0005484074,0.000045462813,0.000037250607,0.00007091293,0.00001647374,0.9207034,0.00014344013,0.0749097,0.0034033712,0.00003303893],"about_ca_topic_score_codex":0.014085556,"about_ca_topic_score_gemma":0.011065026,"teacher_disagreement_score":0.014085556,"about_ca_system_score_codex":0.0018084025,"about_ca_system_score_gemma":0.0024178547,"threshold_uncertainty_score":0.04286766},"labels":[],"label_agreement":null},{"id":"W2215244420","doi":"10.1093/bioinformatics/btv430","title":"Statistically identifying tumor suppressors and oncogenes from pan-cancer genome-sequencing data","year":2015,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Ferroptosis and cancer prognosis","field":"Medicine","cited_by":67,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Cancer Institute; National Institutes of Health; Canadian Institutes of Health Research; Alvin J. Siteman Cancer Center","keywords":"Gene; Cancer; Biology; Genome; Suppressor; Genetics; Computational biology; Function (biology); Mutation","score_opus":0.15881024992191645,"score_gpt":0.3492056189539231,"score_spread":0.19039536903200666,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2215244420","genre_codex":"empirical","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7200384,0.005798137,0.16146627,0.0012680653,0.00014299064,0.00044782014,0.09710706,0.01012439,0.003606922],"genre_scores_gemma":[0.807282,0.00083040295,0.09989352,0.0004797429,0.00012719246,0.0002573434,0.089381404,0.00070604857,0.0010423876],"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9979526,0.0005004271,0.00017935052,0.0008638354,0.00039153663,0.0001122272],"domain_scores_gemma":[0.98809004,0.008317881,0.0012185471,0.0010662952,0.0008352929,0.0004719234],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005978162,0.0009254435,0.0007997802,0.0030453657,0.00046713432,0.0010231381,0.0009757371,0.00069904,0.0030996806],"category_scores_gemma":[0.015513455,0.00022596444,0.0011376494,0.002585718,0.00048264448,0.0006577433,0.0008563523,0.0009427152,0.0011032573],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0022999388,0.00031447815,0.62154824,0.0033091272,0.0024001927,0.0007013996,0.00024306735,0.06546256,0.05391459,0.00225038,0.03897566,0.2085803],"study_design_scores_gemma":[0.0005619469,0.0011022256,0.44301203,0.00035466327,0.0014023888,0.0029590447,0.0003733811,0.38287964,0.098169066,0.024544055,0.044429205,0.00021233587],"about_ca_topic_score_codex":0.0027332106,"about_ca_topic_score_gemma":0.004848664,"teacher_disagreement_score":0.005978162,"about_ca_system_score_codex":0.0005444579,"about_ca_system_score_gemma":0.00093182863,"threshold_uncertainty_score":0.031615973},"labels":[],"label_agreement":null},{"id":"W2227918652","doi":"10.1093/bioinformatics/btv724","title":"Sequence database versioning for command line and Galaxy bioinformatics servers","year":2015,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Public Health Agency of Canada; University of British Columbia","funders":"Genome Canada","keywords":"Computer science; Server; Sequence (biology); Sequence database; Database; Line (geometry); Software versioning; Software; Computational biology; Programming language; Operating system; Biology; Genetics; Gene; Mathematics","score_opus":0.052223927870846086,"score_gpt":0.27682404881970574,"score_spread":0.22460012094885964,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2227918652","genre_codex":"software","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0067061028,0.00048610702,0.15352665,0.0009829937,0.00079665956,0.00073315034,0.13086839,0.69012606,0.01577401],"genre_scores_gemma":[0.03334949,0.0005904196,0.27446014,0.0020652602,0.00035773104,0.0024232978,0.52002114,0.14794978,0.018782714],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9941222,0.000888207,0.001314666,0.0015833253,0.001501431,0.0005902743],"domain_scores_gemma":[0.9851472,0.0016311054,0.001000624,0.007838099,0.0033548384,0.0010281305],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0077122143,0.0018481701,0.0013110951,0.003016654,0.0015769257,0.004127211,0.004584793,0.0018151868,0.07782043],"category_scores_gemma":[0.024586793,0.0020664593,0.0013361293,0.0036737637,0.0008627291,0.005369427,0.0037681926,0.0042712735,0.10940365],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0015297658,0.00011619761,0.004031103,0.0012385007,0.00015430487,0.00031329674,0.0008394745,0.0008714739,0.023730516,0.008247743,0.84754586,0.111381724],"study_design_scores_gemma":[0.00026364773,0.0001784522,0.006557046,0.0002943967,0.00010776078,0.00063751696,0.00016087132,0.005372277,0.03764079,0.009954466,0.93856734,0.00026538884],"about_ca_topic_score_codex":0.00442739,"about_ca_topic_score_gemma":0.0037386047,"teacher_disagreement_score":0.07782043,"about_ca_system_score_codex":0.0019827539,"about_ca_system_score_gemma":0.0029529808,"threshold_uncertainty_score":0.26033527},"labels":[],"label_agreement":null},{"id":"W2232721574","doi":"10.1093/bioinformatics/btv504","title":"JBASE: Joint Bayesian Analysis of Subphenotypes and Epistasis","year":2015,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Hospital for Sick Children; University of Toronto","funders":"Canadian Institutes of Health Research; Sickkids Research Institute; Hospital for Sick Children; Natural Sciences and Engineering Research Council of Canada; Instituto Mexicano del Seguro Social; Instituto de Seguriidad y Servicios Sociales de los Trabadores del Estado; Consejo Nacional de Ciencia y Tecnología","keywords":"Epistasis; Heritability; Genome-wide association study; Genotyping; Bayesian probability; Biology; Sample size determination; Genetic association; Computational biology; Genetics; Genotype; Missing heritability problem; Penetrance; Evolutionary biology; Phenotype; Genetic variants; Computer science; Statistics; Gene; Artificial intelligence; Mathematics; Single-nucleotide polymorphism","score_opus":0.024990506030581467,"score_gpt":0.26428429347428606,"score_spread":0.23929378744370458,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2232721574","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00921279,0.0003158947,0.9870036,0.00048674855,0.000032376625,0.00011695185,0.0010302069,0.0012319119,0.0005694125],"genre_scores_gemma":[0.22410351,0.00074874423,0.7615925,0.00093315094,0.00019187301,0.00088993117,0.006989395,0.0011035161,0.0034473517],"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9944402,0.0035958197,0.00020460977,0.0008474945,0.00066309876,0.00024875737],"domain_scores_gemma":[0.98077077,0.01492765,0.0009877123,0.0017349957,0.001100729,0.00047807806],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.011195971,0.0015602123,0.00228401,0.0021206865,0.0011429797,0.002666266,0.0040339697,0.001609073,0.006965098],"category_scores_gemma":[0.03699561,0.001374838,0.0035270366,0.002497955,0.0017526937,0.002140129,0.004181968,0.0040465607,0.0019432931],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0020526669,0.00041611536,0.059239574,0.0011604592,0.004170892,0.00084750634,0.0010455275,0.41726708,0.009368975,0.10708259,0.030637616,0.36671102],"study_design_scores_gemma":[0.00020419447,0.00009359061,0.006220828,0.000072966825,0.00034373844,0.00022854807,0.00007708141,0.8851802,0.0011951725,0.09980938,0.0064970646,0.000077267316],"about_ca_topic_score_codex":0.018664392,"about_ca_topic_score_gemma":0.018902153,"teacher_disagreement_score":0.018664392,"about_ca_system_score_codex":0.0010443203,"about_ca_system_score_gemma":0.0030839252,"threshold_uncertainty_score":0.059210658},"labels":[],"label_agreement":null},{"id":"W2235818404","doi":"10.1093/bioinformatics/btv535","title":"Group and sparse group partial least square approaches applied in genomics context","year":2015,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":57,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Context (archaeology); Multivariate statistics; Computational biology; Relevance (law); Computer science; Genomics; SNP; Multivariate analysis; Omics; Data mining; Biology; Bioinformatics; Machine learning; Gene; Genome; Genetics; Single-nucleotide polymorphism; Genotype","score_opus":0.05526268856140494,"score_gpt":0.2367565548618991,"score_spread":0.18149386630049413,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2235818404","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0048139705,0.00032808224,0.99361295,0.00025771518,0.000038361366,0.00003921236,0.00015805851,0.00036583818,0.00038589793],"genre_scores_gemma":[0.13239826,0.00072697387,0.8620348,0.00036849536,0.00027056207,0.00041622604,0.001188938,0.00034816793,0.002247564],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99794704,0.0010355588,0.00006631884,0.000493731,0.00037661652,0.000080762264],"domain_scores_gemma":[0.9960461,0.0027043284,0.0003151286,0.0004123762,0.00040883513,0.00011318358],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0028400186,0.0014819623,0.0014121574,0.0013328753,0.0005889379,0.0009832053,0.0013181888,0.001287606,0.0038845523],"category_scores_gemma":[0.0077042454,0.0005847638,0.0015892733,0.0023688786,0.0012398876,0.0012367412,0.0018528699,0.002083875,0.0013748365],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00031299813,0.00016274974,0.005712465,0.000777056,0.0006103811,0.00030699925,0.00037731315,0.50981486,0.010633816,0.048625845,0.009244121,0.41342136],"study_design_scores_gemma":[0.000033886223,0.000106174,0.0011993785,0.000029059012,0.00004898214,0.000116971976,0.000054685654,0.92085797,0.0017799718,0.070618786,0.005128756,0.000025392488],"about_ca_topic_score_codex":0.002413896,"about_ca_topic_score_gemma":0.0022450266,"teacher_disagreement_score":0.0038845523,"about_ca_system_score_codex":0.00048371742,"about_ca_system_score_gemma":0.0012858044,"threshold_uncertainty_score":0.015019596},"labels":[],"label_agreement":null},{"id":"W2256500242","doi":"10.1093/bioinformatics/btv763","title":"Sampletrees and Rsampletrees: sampling gene genealogies conditional on SNP genotype data","year":2016,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University; University of Ottawa","funders":"Natural Sciences and Engineering Research Council of Canada; Canadian Institutes of Health Research; Michael Smith Health Research BC","keywords":"R package; Computer science; Documentation; Markov chain Monte Carlo; Source code; Code (set theory); SNP; Genotype; Markov chain; Data mining; Monte Carlo method; Programming language; Statistics; Biology; Genetics; Gene; Single-nucleotide polymorphism; Mathematics","score_opus":0.09098278476994492,"score_gpt":0.3176090273950153,"score_spread":0.22662624262507036,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2256500242","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.012779107,0.00045863114,0.9083276,0.0006160754,0.00032359394,0.00045736745,0.044433407,0.028368376,0.0042358725],"genre_scores_gemma":[0.086874224,0.0005561138,0.8138753,0.0006351473,0.00027105949,0.0025119581,0.067594126,0.02014478,0.007537315],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99734604,0.0015388813,0.00013283589,0.00054635946,0.00032793716,0.00010798535],"domain_scores_gemma":[0.9864943,0.009902602,0.0005917842,0.0019609565,0.00069965154,0.00035074895],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0069085676,0.0011171475,0.0013017816,0.0026114562,0.0011261458,0.0020859712,0.0031073315,0.0012756421,0.06429059],"category_scores_gemma":[0.032446314,0.0013919146,0.0017871428,0.003254418,0.0010375438,0.0020123713,0.001787694,0.0029976547,0.014798236],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013421176,0.00027418716,0.023342278,0.0014820964,0.0010188014,0.0006010773,0.0009454484,0.09751559,0.003956322,0.07942317,0.45371187,0.33638698],"study_design_scores_gemma":[0.0009367931,0.00018185278,0.006241075,0.000373259,0.00026164707,0.00082709483,0.00016165376,0.56699353,0.0079111345,0.19555357,0.22036923,0.00018910797],"about_ca_topic_score_codex":0.004150932,"about_ca_topic_score_gemma":0.0073010046,"teacher_disagreement_score":0.06429059,"about_ca_system_score_codex":0.0006858184,"about_ca_system_score_gemma":0.0022565778,"threshold_uncertainty_score":0.2150734},"labels":[],"label_agreement":null},{"id":"W2256795974","doi":"10.1093/bioinformatics/btw045","title":"Predicting physiologically relevant SH3 domain mediated protein–protein interactions in yeast","year":2016,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"National Institute of General Medical Sciences; Canadian Institutes of Health Research","keywords":"Computational biology; Protein–protein interaction; Protein sequencing; Computer science; Peptide library; SH3 domain; Biology; Peptide sequence; Genetics; Signal transduction; Gene","score_opus":0.009898811290404376,"score_gpt":0.22567551340616274,"score_spread":0.21577670211575836,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2256795974","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9363616,0.00066633214,0.048467085,0.00010025717,0.00001257144,0.000037903817,0.009128351,0.0042022457,0.0010235959],"genre_scores_gemma":[0.89379865,0.0003736578,0.08455669,0.000064957596,0.0000076290953,0.00005573163,0.020272406,0.0002703105,0.000599916],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99981123,0.00003707329,0.000012929751,0.000062356834,0.00005307015,0.000023481492],"domain_scores_gemma":[0.99949515,0.00031373263,0.00006424946,0.000025966385,0.00005361812,0.000047341568],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007044036,0.0010069971,0.00050802744,0.0011027534,0.00032878606,0.0006952927,0.00049629394,0.00065971806,0.0011019417],"category_scores_gemma":[0.0011938586,0.0002622448,0.0007798716,0.000764182,0.00015713932,0.0003761126,0.0005324858,0.00044188366,0.0004651268],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0029622817,0.0004149593,0.1705428,0.0017911742,0.0008804881,0.0018288175,0.00023203313,0.5314269,0.20205782,0.001751225,0.007938727,0.0781728],"study_design_scores_gemma":[0.0000727143,0.00016044089,0.043598987,0.00003561287,0.00009364925,0.00037519404,0.00016231339,0.8939843,0.05637961,0.0021869133,0.002899712,0.000050523548],"about_ca_topic_score_codex":0.0049396777,"about_ca_topic_score_gemma":0.005776976,"teacher_disagreement_score":0.0049396777,"about_ca_system_score_codex":0.00044416738,"about_ca_system_score_gemma":0.0005617044,"threshold_uncertainty_score":0.009821832},"labels":[],"label_agreement":null},{"id":"W2261245081","doi":"10.1093/bioinformatics/btv755","title":"simDEF: definition-based semantic similarity measure of gene ontology terms for functional similarity analysis of genes","year":2015,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":31,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Ottawa; Dalhousie University","funders":"Natural Sciences and Engineering Research Council of Canada; Canada Research Chairs","keywords":"Semantic similarity; Computer science; Similarity (geometry); Similarity measure; Cosine similarity; Correlation; Measure (data warehouse); Data mining; Artificial intelligence; Natural language processing; Information retrieval; Theoretical computer science; Mathematics; Pattern recognition (psychology); Image (mathematics)","score_opus":0.04728626949081537,"score_gpt":0.2517290138241195,"score_spread":0.20444274433330412,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2261245081","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.24453516,0.0024450787,0.69693357,0.0003390642,0.00020739231,0.0007178667,0.037726674,0.008609592,0.008485672],"genre_scores_gemma":[0.45751804,0.00058672245,0.49163404,0.0001530443,0.000087332846,0.0010453903,0.0468244,0.0007391933,0.0014118807],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99737954,0.0004423909,0.00042633712,0.0005557447,0.0010634067,0.00013252793],"domain_scores_gemma":[0.99657047,0.0013325218,0.00064562575,0.00044664842,0.00084564276,0.00015905112],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0017110703,0.0008738143,0.00095098274,0.008163026,0.0005936026,0.0013984425,0.0010520453,0.0006615694,0.004402825],"category_scores_gemma":[0.009453135,0.00021668119,0.0012210156,0.0063994233,0.0006810292,0.0021609983,0.0017316072,0.00083965034,0.0012920111],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0022377225,0.0006023779,0.076214425,0.0046755043,0.00094455434,0.00061004964,0.0013737117,0.024413789,0.11475611,0.04921908,0.030983707,0.69396895],"study_design_scores_gemma":[0.00046008508,0.0020997673,0.15891138,0.0007005892,0.0005540828,0.004760185,0.0022301122,0.4503315,0.11470793,0.15726368,0.10746111,0.00051946077],"about_ca_topic_score_codex":0.0011978522,"about_ca_topic_score_gemma":0.0015879301,"teacher_disagreement_score":0.008163026,"about_ca_system_score_codex":0.0009798729,"about_ca_system_score_gemma":0.0010160151,"threshold_uncertainty_score":0.014728904},"labels":[],"label_agreement":null},{"id":"W2261944919","doi":"10.1093/bioinformatics/btv723","title":"PharmacoGx: an R package for analysis of large pharmacogenomic datasets","year":2015,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":299,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Hospital for Sick Children; Montreal Clinical Research Institute; Institute of Cancer Research; Ontario Institute for Cancer Research; University Health Network; University of Toronto; Princess Margaret Cancer Centre","funders":"Ontario Institute for Cancer Research; Fondation Brain Canada; Canadian Institutes of Health Research; Cancer Research Society","keywords":"Pharmacogenomics; Computer science; R package; Drug response; Source code; Cancer cell lines; Data science; Data mining; Open source; Realization (probability); Precision medicine; Software; Drug; Bioinformatics; Cancer; Medicine; Biology; Pharmacology; Programming language","score_opus":0.04707972317749436,"score_gpt":0.36029830260763024,"score_spread":0.31321857943013587,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2261944919","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.004132795,0.0010831829,0.48866603,0.002097548,0.0005407597,0.00086449593,0.14755218,0.34997094,0.0050920453],"genre_scores_gemma":[0.05816478,0.00144027,0.65755975,0.002083457,0.00044278565,0.007700899,0.12788104,0.13809507,0.006632041],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9954066,0.002155835,0.0004233401,0.00084262487,0.0008956262,0.0002759507],"domain_scores_gemma":[0.97442293,0.01933285,0.0019266515,0.0023582517,0.0012222214,0.00073708966],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.009313316,0.003403332,0.002838389,0.0041906247,0.000745329,0.0024802242,0.0043667853,0.0013185123,0.0701431],"category_scores_gemma":[0.042072833,0.0021584046,0.0037505731,0.0038288515,0.0015398358,0.002610058,0.004656725,0.0045417505,0.032178394],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001269814,0.00011028006,0.010481593,0.004288276,0.0031677142,0.00067404046,0.00049123965,0.017846853,0.0032796834,0.014711703,0.84376657,0.09991225],"study_design_scores_gemma":[0.0023817206,0.00039290296,0.016684765,0.0010072066,0.0015995055,0.0015817583,0.00019007908,0.16278303,0.009509526,0.10632627,0.69699836,0.00054477894],"about_ca_topic_score_codex":0.0035942926,"about_ca_topic_score_gemma":0.0044889534,"teacher_disagreement_score":0.0701431,"about_ca_system_score_codex":0.0009964121,"about_ca_system_score_gemma":0.005176151,"threshold_uncertainty_score":0.23465204},"labels":[],"label_agreement":null},{"id":"W2269649163","doi":"10.1093/bioinformatics/btw252","title":"Classifying and segmenting microscopy images with deep multiple instance learning","year":2016,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Cell Image Analysis Techniques","field":"Biochemistry, Genetics and Molecular Biology","cited_by":449,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"Natural Sciences and Engineering Research Council of Canada; Nvidia","keywords":"Pooling; Artificial intelligence; Convolutional neural network; Computer science; Pattern recognition (psychology); Segmentation; Deep learning; Feature (linguistics); Microscopy; Pixel; Contextual image classification; Machine learning; Computer vision; Image (mathematics)","score_opus":0.005466362266401763,"score_gpt":0.23233109526310197,"score_spread":0.2268647329967002,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2269649163","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.06715748,0.00065500435,0.9133796,0.0006891648,0.000060131657,0.00014933627,0.0017720438,0.013586161,0.0025510688],"genre_scores_gemma":[0.39990547,0.0005197762,0.5898533,0.0002692629,0.000053971013,0.00016695914,0.0058512953,0.00040685936,0.0029731812],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9995301,0.000056323322,0.000030584983,0.00020203002,0.000121097546,0.00005993437],"domain_scores_gemma":[0.9992092,0.00020001656,0.00014861133,0.00022206402,0.0001641569,0.000055918033],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008694059,0.0011739597,0.00057639554,0.00089896575,0.00033377163,0.0011293541,0.002074155,0.0010407914,0.0028056337],"category_scores_gemma":[0.0022502763,0.0005613337,0.00079664413,0.0010415288,0.00066598697,0.0018783663,0.0011183593,0.0016046269,0.0016078623],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00046509548,0.00039080757,0.007303204,0.00046869257,0.00018456306,0.0003233078,0.00012331353,0.26867354,0.15311968,0.008284786,0.014764489,0.5458985],"study_design_scores_gemma":[0.000009698023,0.000045080367,0.0014908371,0.000018256686,0.000023121,0.000087262815,0.000028312692,0.9165812,0.074042544,0.005044286,0.0026116967,0.000017638007],"about_ca_topic_score_codex":0.0047206404,"about_ca_topic_score_gemma":0.0056424434,"teacher_disagreement_score":0.0047206404,"about_ca_system_score_codex":0.0012250084,"about_ca_system_score_gemma":0.00083917874,"threshold_uncertainty_score":0.009386301},"labels":[],"label_agreement":null},{"id":"W2287141830","doi":"10.1093/bioinformatics/btw105","title":"ecceTERA: comprehensive gene tree-species tree reconciliation using parsimony","year":2016,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":115,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Tree (set theory); Tree rearrangement; Phylogenetic tree; Biology; Maximum parsimony; Binary tree; Gene; Computer science; Computational biology; Phylogenetics; Genetic algorithm; Evolutionary biology; Algorithm; Genetics; Mathematics; Combinatorics; Clade","score_opus":0.0399052319189952,"score_gpt":0.24245713169802435,"score_spread":0.20255189977902915,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2287141830","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.030944832,0.0013838315,0.83505297,0.0011232106,0.000835551,0.0005928596,0.02583589,0.09649574,0.007735105],"genre_scores_gemma":[0.117523745,0.0005785725,0.78208894,0.0004506011,0.00030312428,0.001998863,0.052699674,0.040446416,0.003910017],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9974595,0.0010944494,0.00014957965,0.0007951719,0.00035908105,0.00014229634],"domain_scores_gemma":[0.9941748,0.003151211,0.00048461466,0.0014644237,0.00050857506,0.00021644885],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008143756,0.002728616,0.0029026167,0.004866875,0.0026634308,0.0030454525,0.0059824167,0.0026766467,0.037059683],"category_scores_gemma":[0.022312643,0.002141291,0.00366565,0.0043698484,0.0011851845,0.0046159127,0.0053337333,0.0039561605,0.013219327],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0030467082,0.00051442214,0.013834086,0.004504665,0.0037360229,0.0011767662,0.0018723427,0.09751028,0.021732394,0.053687446,0.33357868,0.46480623],"study_design_scores_gemma":[0.0009988111,0.0003454301,0.00828297,0.00073479477,0.0009306987,0.0014291223,0.00023976152,0.65008175,0.017368117,0.10281514,0.21639596,0.00037741655],"about_ca_topic_score_codex":0.0009789519,"about_ca_topic_score_gemma":0.0016766654,"teacher_disagreement_score":0.037059683,"about_ca_system_score_codex":0.0009960026,"about_ca_system_score_gemma":0.0017471752,"threshold_uncertainty_score":0.123977005},"labels":[],"label_agreement":null},{"id":"W2288603533","doi":"10.1093/bioinformatics/btv746","title":"Robust quantitative scratch assay","year":2015,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Cell Image Analysis Techniques","field":"Biochemistry, Genetics and Molecular Biology","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Lunenfeld-Tanenbaum Research Institute; Princess Margaret Cancer Centre; University Health Network; University of Toronto","funders":"","keywords":"Artificial intelligence; Outlier; Computer science; Machine learning; Mathematics","score_opus":0.037231881019762096,"score_gpt":0.2895079666158675,"score_spread":0.25227608559610537,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2288603533","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.06278374,0.008844337,0.800177,0.002103862,0.003614315,0.0027685987,0.04733973,0.035907157,0.036461346],"genre_scores_gemma":[0.24248113,0.0060823373,0.59057665,0.002859796,0.0008149509,0.006034705,0.09634163,0.0058144573,0.048994325],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99142075,0.0007399323,0.0006197593,0.0024825267,0.0042071147,0.00053002057],"domain_scores_gemma":[0.99325424,0.0013929338,0.0007680577,0.0014162572,0.0029156494,0.0002529001],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0036585547,0.0026078655,0.002191688,0.0020312588,0.00088311825,0.0026295872,0.0034274207,0.0026542724,0.0201161],"category_scores_gemma":[0.0077947765,0.0012963156,0.00220963,0.0013476844,0.0010565796,0.0012890183,0.002344487,0.0024410088,0.027780771],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010408964,0.00042612836,0.005281767,0.0036678219,0.00024289418,0.00033303397,0.00019731042,0.007847249,0.7364079,0.0034239055,0.063541114,0.17759],"study_design_scores_gemma":[0.00012962733,0.0012161043,0.009975113,0.0004614458,0.00032311675,0.0011483086,0.00018325557,0.04936244,0.7658824,0.002723316,0.16824158,0.000353281],"about_ca_topic_score_codex":0.0020601833,"about_ca_topic_score_gemma":0.0029670745,"teacher_disagreement_score":0.0201161,"about_ca_system_score_codex":0.0011452574,"about_ca_system_score_gemma":0.001569963,"threshold_uncertainty_score":0.067295015},"labels":[],"label_agreement":null},{"id":"W2307896283","doi":"10.1093/bioinformatics/btw031","title":"ELASPIC web-server: proteome-wide structure-based prediction of mutation effects on protein stability and binding affinity","year":2016,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":82,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"UniProt; Web server; Computer science; Proteome; Computational biology; Protein domain; Domain (mathematical analysis); Homology modeling; Bioinformatics; Biology; Genetics; The Internet; World Wide Web; Mathematics","score_opus":0.007831043950518006,"score_gpt":0.1995485839587797,"score_spread":0.1917175400082617,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2307896283","genre_codex":"dataset","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.118048966,0.0034899514,0.12669243,0.0015166237,0.0006451545,0.00065510644,0.50710386,0.22918771,0.01266022],"genre_scores_gemma":[0.19086911,0.00091588986,0.0907369,0.00057723804,0.0002675275,0.00068132253,0.7028372,0.008380623,0.004734237],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99913317,0.00010600678,0.00006227652,0.00032656683,0.0003003865,0.00007156557],"domain_scores_gemma":[0.9988423,0.00025431346,0.00017776288,0.00018553964,0.00038609386,0.00015396716],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014890076,0.0029489892,0.001661295,0.0017791417,0.00065396575,0.0012960562,0.0032087897,0.0012939718,0.021881886],"category_scores_gemma":[0.0034211383,0.00052487344,0.0010754695,0.001627281,0.00031088185,0.0011075421,0.0010661971,0.0012949957,0.019804325],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0031788405,0.0005573877,0.026903028,0.002708615,0.00065077224,0.00052828674,0.000096118616,0.025824683,0.0481976,0.002576429,0.7595855,0.12919272],"study_design_scores_gemma":[0.0013928056,0.00070462417,0.059795294,0.00038911172,0.00048043334,0.001821314,0.0001249625,0.65457416,0.12207328,0.016962208,0.14127353,0.00040826222],"about_ca_topic_score_codex":0.0021164245,"about_ca_topic_score_gemma":0.002459641,"teacher_disagreement_score":0.021881886,"about_ca_system_score_codex":0.0007425337,"about_ca_system_score_gemma":0.0011610924,"threshold_uncertainty_score":0.07320219},"labels":[],"label_agreement":null},{"id":"W2318488401","doi":"10.1093/bioinformatics/btw155","title":"MOLGENIS/connect: a system for semi-automatic integration of heterogeneous phenotype data with applications in biobanks","year":2016,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Biomedical Text Mining and Ontologies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":18,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; McGill University Health Centre","funders":"Terveyden ja hyvinvoinnin laitos; Nederlandse Organisatie voor Wetenschappelijk Onderzoek; European Commission","keywords":"Computer science; Biobank; Documentation; Source code; Data integration; Data mining; Terminology; Open source; Information retrieval; Ontology; Categorical variable; Task (project management); Data source; Matching (statistics); Software; Programming language; Machine learning; Bioinformatics","score_opus":0.024251684551665875,"score_gpt":0.2718240070330186,"score_spread":0.24757232248135275,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2318488401","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.02013474,0.00066386664,0.35870367,0.0010192548,0.00025335982,0.0011304755,0.04997556,0.5613034,0.0068157907],"genre_scores_gemma":[0.111026205,0.00060412847,0.68876797,0.0019024645,0.00022529275,0.0035245495,0.14354391,0.044225667,0.0061797304],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9968316,0.0007436209,0.00042234952,0.0011336361,0.00074261177,0.0001262927],"domain_scores_gemma":[0.99176604,0.004494541,0.00093248446,0.001666647,0.0007109348,0.00042943546],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.006365995,0.002132892,0.0012450407,0.005397011,0.0009174209,0.0029119859,0.0028511048,0.0012589978,0.025294712],"category_scores_gemma":[0.018640399,0.0014801438,0.0016954239,0.003537375,0.0007676438,0.003071574,0.005656826,0.0012843044,0.010033202],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.004415947,0.0004507673,0.02808722,0.0033245673,0.0011785389,0.0022661649,0.0018579023,0.012007986,0.031893376,0.014013835,0.5380815,0.3624222],"study_design_scores_gemma":[0.0015912291,0.0004652497,0.033184957,0.0011015804,0.00052035874,0.003709823,0.0009076775,0.27273223,0.10026887,0.052475147,0.53250045,0.0005424732],"about_ca_topic_score_codex":0.002413324,"about_ca_topic_score_gemma":0.0027342986,"teacher_disagreement_score":0.025294712,"about_ca_system_score_codex":0.001352452,"about_ca_system_score_gemma":0.0022770737,"threshold_uncertainty_score":0.08461928},"labels":[],"label_agreement":null},{"id":"W2319614057","doi":"10.1093/bioinformatics/btw096","title":"Reaction Decoder Tool (RDT): extracting features from chemical reactions","year":2016,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Microbial Metabolic Engineering and Bioproduction","field":"Biochemistry, Genetics and Molecular Biology","cited_by":108,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canada's Michael Smith Genome Sciences Centre; University of British Columbia","funders":"","keywords":"Computer science; Software; Chemical reaction; Java; Task (project management); Chemistry; Programming language; Biochemistry","score_opus":0.0072915514519623185,"score_gpt":0.221542326067819,"score_spread":0.2142507746158567,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2319614057","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.012329791,0.0006395708,0.53313744,0.00029412936,0.00020536475,0.0002870559,0.063015535,0.38446295,0.0056282156],"genre_scores_gemma":[0.080821924,0.0007425974,0.74813604,0.00020697,0.00008367026,0.0007288374,0.13111268,0.029478027,0.008689208],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9994185,0.00005766058,0.00006387621,0.00020722613,0.00021181347,0.000040900344],"domain_scores_gemma":[0.99861157,0.0007131335,0.00017827265,0.00019884111,0.00023550827,0.000062625026],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010094512,0.0024403587,0.0010571148,0.002742886,0.0005692567,0.0016970694,0.0021682521,0.0011628757,0.034406126],"category_scores_gemma":[0.004512568,0.00082539476,0.0021095444,0.0016650066,0.00042635883,0.00183875,0.0012029164,0.0016139984,0.01587088],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011484716,0.000290191,0.008133922,0.006125334,0.0003629474,0.0013515286,0.0005629146,0.02716027,0.08514769,0.014076792,0.28179064,0.57384944],"study_design_scores_gemma":[0.0003331873,0.0002800142,0.006118007,0.00030685958,0.00019550083,0.0017622111,0.00024496496,0.33340684,0.31550252,0.024924507,0.31664646,0.00027896726],"about_ca_topic_score_codex":0.0020527488,"about_ca_topic_score_gemma":0.002470043,"teacher_disagreement_score":0.034406126,"about_ca_system_score_codex":0.0006492238,"about_ca_system_score_gemma":0.0013882528,"threshold_uncertainty_score":0.11509997},"labels":[],"label_agreement":null},{"id":"W2323614149","doi":"10.1093/bioinformatics/btw129","title":"mDCC_tools: characterizing multi-modal atomic motions in molecular dynamics trajectories","year":2016,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Protein Structure and Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Institute of Genetics; Japan Society for the Promotion of Science; Research Organization of Information and Systems","keywords":"Modal; Molecular dynamics; Computer science; Dynamics (music); Biological system; Physics; Statistical physics; Classical mechanics; Chemistry; Biology; Acoustics; Quantum mechanics","score_opus":0.007779964906881822,"score_gpt":0.23137124298156164,"score_spread":0.22359127807467982,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2323614149","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.04924817,0.0005573524,0.7294459,0.00046088695,0.00027129374,0.0004020573,0.028169474,0.18635663,0.0050882525],"genre_scores_gemma":[0.26610917,0.0005101747,0.66993165,0.00023566437,0.00011864909,0.0015021662,0.03986952,0.018511381,0.003211666],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99943894,0.00008615355,0.000055600307,0.00015282701,0.00021739934,0.000048993596],"domain_scores_gemma":[0.99814284,0.0007745456,0.0002435429,0.00027531493,0.00041250716,0.00015123322],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010849006,0.0012752078,0.000832956,0.0017406505,0.00090349163,0.001343266,0.0025383774,0.0011806963,0.016482633],"category_scores_gemma":[0.005909528,0.0005151188,0.00080682326,0.0013334772,0.0004920747,0.0012176225,0.0014798653,0.0013403012,0.0050129285],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011975875,0.0005221121,0.02808991,0.0032739239,0.0005158837,0.00081972766,0.0012091993,0.22811164,0.0622434,0.036877405,0.26127365,0.37586558],"study_design_scores_gemma":[0.000061919134,0.00004157474,0.0030362394,0.0000661134,0.000031014668,0.00012109844,0.00004386054,0.93702126,0.024874352,0.005164881,0.02943926,0.00009827672],"about_ca_topic_score_codex":0.004023919,"about_ca_topic_score_gemma":0.0038252377,"teacher_disagreement_score":0.016482633,"about_ca_system_score_codex":0.00073033734,"about_ca_system_score_gemma":0.0016116911,"threshold_uncertainty_score":0.05513984},"labels":[],"label_agreement":null},{"id":"W2333196156","doi":"10.1093/bioinformatics/btw178","title":"Cell-free DNA fragment-size distribution analysis for non-invasive prenatal CNV prediction","year":2016,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Prenatal Screening and Diagnostics","field":"Medicine","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Hospital for Sick Children; University of Toronto","funders":"","keywords":"Copy-number variation; Single-nucleotide polymorphism; SNP; Cell-free fetal DNA; Genome; Biology; Genetics; SNP array; Human genome; Computational biology; Prenatal diagnosis; Fetus; Gene; Genotype; Pregnancy","score_opus":0.01048325492405987,"score_gpt":0.2294190667248588,"score_spread":0.2189358118007989,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2333196156","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.10044003,0.0031390248,0.86486685,0.00052618753,0.00013414313,0.0002793022,0.0074406927,0.020716028,0.0024577836],"genre_scores_gemma":[0.3185236,0.0008694338,0.6717782,0.00027688907,0.00008177382,0.00034304772,0.005443601,0.0008759862,0.0018074696],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99910223,0.00018881279,0.000045475987,0.00026788912,0.00034844727,0.000047061247],"domain_scores_gemma":[0.9966543,0.0022114795,0.00036394084,0.00021107499,0.00044721068,0.000112072645],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014006203,0.00079001806,0.00069722725,0.0026082243,0.0002942451,0.0006503831,0.0010974581,0.0010289311,0.0047873794],"category_scores_gemma":[0.0065484345,0.00039642115,0.0006108938,0.001051404,0.00033792853,0.0005316847,0.00072938274,0.00066211034,0.0022265532],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014214114,0.00027477508,0.10886392,0.0012915769,0.0004680255,0.0008645569,0.00036310148,0.05246529,0.16161264,0.0030752537,0.016447,0.6528524],"study_design_scores_gemma":[0.00010762405,0.00017838951,0.041209072,0.00014594843,0.0001479502,0.0020377415,0.000084852465,0.85696787,0.08087616,0.0064347773,0.011651659,0.00015794308],"about_ca_topic_score_codex":0.004071081,"about_ca_topic_score_gemma":0.005230459,"teacher_disagreement_score":0.0047873794,"about_ca_system_score_codex":0.00057011057,"about_ca_system_score_gemma":0.0007084382,"threshold_uncertainty_score":0.01601541},"labels":[],"label_agreement":null},{"id":"W2334167823","doi":"10.1093/bioinformatics/btw134","title":"GSP: a web-based platform for designing genome-specific primers in polyploids","year":2016,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":41,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Agriculture and Agri-Food Canada","funders":"","keywords":"Genome; Upload; Polyploid; Primer (cosmetics); Biology; Computational biology; Sequence (biology); Web server; Computer science; Source code; World Wide Web; Genetics; The Internet; Gene; Programming language","score_opus":0.022221481360277615,"score_gpt":0.2309668739902974,"score_spread":0.2087453926300198,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2334167823","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.014076468,0.0012272601,0.7110225,0.0006832819,0.00040213377,0.0010517484,0.034008745,0.2292591,0.0082688],"genre_scores_gemma":[0.035782438,0.0007052623,0.8842814,0.0007574663,0.000093898314,0.0023149177,0.04404801,0.022926679,0.009089882],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99839526,0.0002905264,0.00015895595,0.00054491265,0.0004817986,0.00012849769],"domain_scores_gemma":[0.9976319,0.0011605347,0.00041068235,0.00030906088,0.00029162926,0.0001961882],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0032095066,0.0030050299,0.0013532103,0.0024446307,0.0009597323,0.0016088071,0.002730319,0.0020479066,0.04225553],"category_scores_gemma":[0.0050024744,0.0024895025,0.0019498987,0.0017930731,0.00092739466,0.0022260733,0.002087237,0.0027888732,0.031298093],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.003156706,0.0006458913,0.003591772,0.0038886291,0.00039332517,0.001441702,0.0008069038,0.013726865,0.51365,0.010118327,0.1869014,0.26167846],"study_design_scores_gemma":[0.00083766965,0.00078834227,0.0035388325,0.00040084263,0.00026239036,0.0014406437,0.00017998465,0.048190698,0.5659172,0.016046897,0.3620552,0.00034132748],"about_ca_topic_score_codex":0.0010920807,"about_ca_topic_score_gemma":0.0012173454,"teacher_disagreement_score":0.04225553,"about_ca_system_score_codex":0.0010850767,"about_ca_system_score_gemma":0.0014125118,"threshold_uncertainty_score":0.1413588},"labels":[],"label_agreement":null},{"id":"W2336321619","doi":"10.1093/bioinformatics/btw053","title":"Sprites: detection of deletions from sequencing data by re-aligning split reads","year":2016,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":18,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Saskatchewan","funders":"National Natural Science Foundation of China","keywords":"Computer science; Reference genome; DNA sequencing; Computational biology; Biology; Genetics; Gene","score_opus":0.031238478506569654,"score_gpt":0.2468003291338725,"score_spread":0.21556185062730285,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2336321619","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.21511106,0.0031876508,0.6537031,0.00060275407,0.00042365325,0.0006571142,0.020512655,0.10044713,0.0053548357],"genre_scores_gemma":[0.22861099,0.0005444905,0.71926934,0.00040889753,0.00013818814,0.0005045194,0.04066776,0.0052995775,0.0045562442],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99725753,0.00054133945,0.00024346409,0.0009008408,0.00091929187,0.00013755406],"domain_scores_gemma":[0.9936386,0.0032158273,0.0011283073,0.0008304796,0.0009223627,0.00026436092],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0031571607,0.0018100119,0.0013814396,0.0031043137,0.0007950444,0.0011917889,0.0016855756,0.0011370595,0.004485207],"category_scores_gemma":[0.01437393,0.0006827507,0.0012941838,0.0020299174,0.000591919,0.0014674488,0.0016176183,0.0014338619,0.0051014083],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.004202769,0.00054178236,0.059618764,0.0032700624,0.0017094248,0.0019070611,0.0015863089,0.04281273,0.19112241,0.0039904644,0.06705381,0.6221844],"study_design_scores_gemma":[0.0005280816,0.0011671139,0.03562107,0.0003349828,0.0004508772,0.004120925,0.0007827117,0.66347206,0.22663334,0.009094712,0.05747162,0.00032259928],"about_ca_topic_score_codex":0.0025175295,"about_ca_topic_score_gemma":0.0052672615,"teacher_disagreement_score":0.004485207,"about_ca_system_score_codex":0.00039938954,"about_ca_system_score_gemma":0.0010919699,"threshold_uncertainty_score":0.01669681},"labels":[],"label_agreement":null},{"id":"W2336477031","doi":"10.1093/bioinformatics/btw227","title":"densityCut: an efficient and versatile topological approach for automatic clustering of biological data","year":2016,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Single-cell and spatial transcriptomics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":39,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Cluster analysis; Computer science; Topological data analysis; Data mining; Topology (electrical circuits); Artificial intelligence; Algorithm; Mathematics","score_opus":0.06716936385813714,"score_gpt":0.27231245685480077,"score_spread":0.2051430929966636,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2336477031","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0022983283,0.0004395301,0.9865291,0.00022047762,0.000073869,0.00013596207,0.0011739798,0.008400342,0.0007283048],"genre_scores_gemma":[0.03997392,0.00063873787,0.9457998,0.0002624697,0.000115494106,0.0006076771,0.008179685,0.003184459,0.0012376802],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99541086,0.0010192724,0.00031157606,0.0009175988,0.0021431542,0.0001975403],"domain_scores_gemma":[0.9947635,0.002246456,0.0005095882,0.0008410868,0.0013595526,0.00027975775],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0039298893,0.0027831686,0.0023104346,0.009299907,0.0021003524,0.0041864384,0.005005753,0.002421447,0.007362462],"category_scores_gemma":[0.017857768,0.0021927112,0.0033424746,0.0068573877,0.0019404208,0.0047488175,0.0049345046,0.0035764484,0.0036236837],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00033803968,0.00016748291,0.0052927746,0.0021278348,0.0006629392,0.0006320273,0.0010390524,0.20817147,0.019430885,0.09187891,0.09301518,0.5772434],"study_design_scores_gemma":[0.000087341374,0.00008333235,0.001315219,0.00018131845,0.000071161856,0.0006967378,0.00019395429,0.7948279,0.011136982,0.14377768,0.047501747,0.00012663731],"about_ca_topic_score_codex":0.005779493,"about_ca_topic_score_gemma":0.008322738,"teacher_disagreement_score":0.009299907,"about_ca_system_score_codex":0.0021995865,"about_ca_system_score_gemma":0.0031722258,"threshold_uncertainty_score":0.02462989},"labels":[],"label_agreement":null},{"id":"W2337408142","doi":"10.1093/bioinformatics/btw211","title":"Partitioning and correlating subgroup characteristics from Aligned Pattern Clusters","year":2016,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Machine Learning in Bioinformatics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University; University of Waterloo","funders":"","keywords":"A priori and a posteriori; Sequence alignment; Multiple sequence alignment; Biology; Class (philosophy); Similarity (geometry); Function (biology); Computational biology; Conserved sequence; Divergence (linguistics); Sequence (biology); In silico; Protein superfamily; Structural Classification of Proteins database; Rank (graph theory); Cluster analysis; Phylogenetic tree; Computer science; Artificial intelligence; Genetics; Protein structure; Peptide sequence; Mathematics; Gene; Combinatorics","score_opus":0.006556936634770645,"score_gpt":0.2177660405102531,"score_spread":0.21120910387548247,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2337408142","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.31189218,0.0002682788,0.6786723,0.00019035212,0.000039477414,0.000471728,0.002297067,0.0040036184,0.002165026],"genre_scores_gemma":[0.5083349,0.00011093071,0.4832621,0.000045419452,0.000039171722,0.0003138204,0.0060889246,0.00033729864,0.0014674835],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9988128,0.00015597088,0.000119875804,0.00040349152,0.00039055856,0.00011734952],"domain_scores_gemma":[0.99634945,0.0012693716,0.00060117897,0.00057661685,0.0010350612,0.00016817283],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00074462796,0.0008754006,0.00071001303,0.003913442,0.0005696065,0.0012081291,0.0010394774,0.000844784,0.0017497953],"category_scores_gemma":[0.005954119,0.0002784623,0.0006631738,0.0028770426,0.00045751085,0.0011062805,0.0009751795,0.0005289173,0.001185836],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011294518,0.0005154764,0.07756667,0.00060011854,0.0002404556,0.00046689168,0.00072765816,0.0839405,0.07621155,0.0053296247,0.007191804,0.74607986],"study_design_scores_gemma":[0.000056254186,0.00014732801,0.01497938,0.000033680022,0.000062165185,0.00041889786,0.0003567799,0.9313679,0.036808267,0.011890509,0.0038432465,0.000035561083],"about_ca_topic_score_codex":0.0030217725,"about_ca_topic_score_gemma":0.0031257963,"teacher_disagreement_score":0.003913442,"about_ca_system_score_codex":0.00053968,"about_ca_system_score_gemma":0.0010864115,"threshold_uncertainty_score":0.0060083866},"labels":[],"label_agreement":null},{"id":"W2346950316","doi":"10.1093/bioinformatics/btw228","title":"Drug repositioning based on comprehensive similarity measures and Bi-Random walk algorithm","year":2016,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Computational Drug Discovery Methods","field":"Computer Science","cited_by":470,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Saskatchewan","funders":"Program for New Century Excellent Talents in University; U.S. Food and Drug Administration; National Natural Science Foundation of China; Stowers Institute for Medical Research","keywords":"Similarity (geometry); Drug repositioning; Drug; Computer science; Disease; Random walk; Machine learning; Drug development; Data mining; Artificial intelligence; Mathematics; Medicine; Statistics; Pharmacology","score_opus":0.023641085473261738,"score_gpt":0.2691134950196628,"score_spread":0.24547240954640107,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2346950316","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.041103803,0.0006829105,0.9551173,0.00022352983,0.00004837935,0.00017158005,0.00014043396,0.0008871612,0.0016249793],"genre_scores_gemma":[0.5372532,0.00059362286,0.45784584,0.0002109184,0.00007406106,0.00045043536,0.00087310723,0.000113910704,0.00258492],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99917114,0.0001876641,0.00009366151,0.00021972318,0.0002504481,0.00007739447],"domain_scores_gemma":[0.99886,0.00057762535,0.00017788746,0.0000915309,0.00019878514,0.000094173425],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00093826174,0.0010601771,0.0019721189,0.002943567,0.0005333252,0.00094807823,0.0017325537,0.0013707993,0.0016437905],"category_scores_gemma":[0.0028885235,0.00047830504,0.0012332052,0.0018887385,0.00056286145,0.0015581064,0.0009462071,0.0008045538,0.0003537487],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00011035244,0.00016809753,0.00320752,0.00015830777,0.00014572372,0.00016283685,0.00005396845,0.83500034,0.002484581,0.0075465264,0.0017795423,0.14918214],"study_design_scores_gemma":[0.0000094081415,0.00002994204,0.00014438012,0.000003473412,0.000012396135,0.000034297304,0.000004214147,0.9976914,0.00028316493,0.0016015208,0.00018078227,0.00000498867],"about_ca_topic_score_codex":0.006409603,"about_ca_topic_score_gemma":0.0053672045,"teacher_disagreement_score":0.006409603,"about_ca_system_score_codex":0.0007669459,"about_ca_system_score_gemma":0.0015266162,"threshold_uncertainty_score":0.012744606},"labels":[],"label_agreement":null},{"id":"W2395270406","doi":"10.4018/978-1-4666-3604-0.ch029","title":"Biclustering of DNA Microarray Data","year":2013,"lang":"en","type":"book-chapter","venue":"Bioinformatics","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"National Research Council Canada","funders":"","keywords":"Biclustering; Computer science; Identification (biology); Microarray analysis techniques; Data mining; Computational biology; Microarray databases; Cluster analysis; Microarray; Biological data; Gene chip analysis; Bioinformatics; Artificial intelligence; Gene; Biology; Genetics","score_opus":0.038118952466101724,"score_gpt":0.2569467814462027,"score_spread":0.21882782898010097,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2395270406","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0037503415,0.0040941034,0.9871666,0.00036516052,0.00026583683,0.00009690002,0.0009615529,0.0013719519,0.0019276321],"genre_scores_gemma":[0.055076085,0.0075708316,0.92142963,0.00081634475,0.00037140882,0.0008192288,0.0068909517,0.00078177993,0.006243742],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99813867,0.0006886774,0.00013806835,0.00038665233,0.00057734887,0.00007064691],"domain_scores_gemma":[0.9976921,0.0014863007,0.00012080376,0.0002461341,0.00040568848,0.000048976242],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0017996989,0.0011493022,0.0015056417,0.0024041645,0.00047408696,0.0012707405,0.0010250274,0.00059663353,0.0035154133],"category_scores_gemma":[0.006009515,0.00068380986,0.0012542366,0.0026975204,0.0006961464,0.001000817,0.0013752464,0.0015904745,0.003000174],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00031954786,0.00007506028,0.0022255909,0.0020513628,0.00045135317,0.00038777923,0.0005946214,0.11555305,0.033949718,0.04693886,0.06317145,0.73428166],"study_design_scores_gemma":[0.00004059357,0.00016048303,0.0029567513,0.0003605552,0.00011011602,0.0011186936,0.00030075823,0.75874823,0.020971887,0.13021746,0.08486672,0.00014774327],"about_ca_topic_score_codex":0.0007823671,"about_ca_topic_score_gemma":0.0011250945,"teacher_disagreement_score":0.0035154133,"about_ca_system_score_codex":0.00048746317,"about_ca_system_score_gemma":0.00049192744,"threshold_uncertainty_score":0.011760235},"labels":[],"label_agreement":null},{"id":"W2418153904","doi":"10.1093/bioinformatics/btw337","title":"CAGEd-oPOSSUM: motif enrichment analysis from CAGE-derived TSSs","year":2016,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":12,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Opossum; Perl; Transcription factor; DNA binding site; Computational biology; Biology; Gene; Promoter; Binding site; Motif (music); Computer science; Genetics; World Wide Web; Gene expression","score_opus":0.005895834575798081,"score_gpt":0.2108232657155318,"score_spread":0.20492743113973372,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2418153904","genre_codex":"empirical","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.5293564,0.0027033046,0.21305549,0.00057219545,0.00041052548,0.00022593161,0.17076235,0.072186194,0.010727587],"genre_scores_gemma":[0.47005293,0.0008696994,0.24699298,0.00043790264,0.000084955485,0.0003916713,0.26376247,0.010972443,0.0064349966],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99976414,0.000018602099,0.000014072112,0.000094606476,0.000079045756,0.000029543779],"domain_scores_gemma":[0.9995875,0.00014491077,0.00007783034,0.00007522425,0.000058632948,0.00005593877],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004313475,0.0006390315,0.00054705935,0.00086347904,0.00035972206,0.00057013414,0.00095808733,0.00043424123,0.015129832],"category_scores_gemma":[0.0011609226,0.00027515658,0.00089444773,0.00097078865,0.0002615317,0.00043852816,0.0009236465,0.0006979117,0.0047721704],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0030404287,0.00014914935,0.04131024,0.002540405,0.00047071485,0.00089959713,0.00047019936,0.008241547,0.70216864,0.0036189447,0.07365245,0.16343768],"study_design_scores_gemma":[0.00042992475,0.0009123428,0.11002552,0.00025953533,0.0005066619,0.0022881501,0.00040682007,0.12107779,0.5904683,0.0068351496,0.1665284,0.0002613992],"about_ca_topic_score_codex":0.002015229,"about_ca_topic_score_gemma":0.0052036187,"teacher_disagreement_score":0.015129832,"about_ca_system_score_codex":0.0002532626,"about_ca_system_score_gemma":0.00040575437,"threshold_uncertainty_score":0.050614357},"labels":[],"label_agreement":null},{"id":"W2420922808","doi":"10.1093/bioinformatics/btw280","title":"DFLpred: High-throughput prediction of disordered flexible linker regions in protein sequences","year":2016,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Protein Structure and Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":92,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Proteome; Linker; Computer science; Human proteome project; Sequence (biology); Intrinsically disordered proteins; Throughput; Domain (mathematical analysis); Computational biology; Protein domain; Biological system; Chemistry; Biology; Bioinformatics; Proteomics; Mathematics; Biochemistry","score_opus":0.010321845923509628,"score_gpt":0.22764567802067875,"score_spread":0.21732383209716913,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2420922808","genre_codex":"empirical","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.72972727,0.0049978374,0.22128807,0.00063430844,0.00014569753,0.0002652574,0.014163531,0.026091937,0.0026861143],"genre_scores_gemma":[0.791435,0.0011021384,0.17896557,0.00027241366,0.00007582457,0.0002711462,0.025611212,0.0005026129,0.0017641289],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99930906,0.00014488146,0.000041544678,0.00019987501,0.00024297279,0.000061565515],"domain_scores_gemma":[0.9985892,0.0005815798,0.0002879662,0.00012767155,0.0002893586,0.0001241132],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0019097225,0.0012226147,0.00072995777,0.0011980382,0.00040587893,0.0007068325,0.0008934874,0.0007180739,0.0010070762],"category_scores_gemma":[0.0032451276,0.0003863178,0.0008068486,0.0007110175,0.00034025006,0.00084442185,0.0006131869,0.00088266225,0.0013191792],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0027448167,0.00088662974,0.13560216,0.002646296,0.0006884932,0.0011744951,0.0002781528,0.2776944,0.29181945,0.0032125553,0.061430708,0.22182193],"study_design_scores_gemma":[0.000103328806,0.00045151543,0.020814968,0.00005750895,0.00007922623,0.00062329095,0.000057952642,0.86852175,0.100501455,0.001623169,0.007081572,0.00008428025],"about_ca_topic_score_codex":0.0022724506,"about_ca_topic_score_gemma":0.0029163186,"teacher_disagreement_score":0.0022724506,"about_ca_system_score_codex":0.0007090924,"about_ca_system_score_gemma":0.00085054914,"threshold_uncertainty_score":0.010099649},"labels":[],"label_agreement":null},{"id":"W2463091895","doi":"10.1093/bioinformatics/btw397","title":"ntHash: recursive nucleotide hashing","year":2016,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Algorithms and Data Compression","field":"Computer Science","cited_by":98,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canada's Michael Smith Genome Sciences Centre","funders":"National Human Genome Research Institute; National Institutes of Health","keywords":"Hash function; Computer science; Expediting; Software; Data mining; Sequence (biology); Universal hashing; Theoretical computer science; Hash table; Biology; Genetics; Programming language","score_opus":0.017284746271392315,"score_gpt":0.23445768803110562,"score_spread":0.2171729417597133,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2463091895","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.027334634,0.0015148252,0.93281287,0.0003716009,0.0004521915,0.0003314487,0.0026428127,0.025621118,0.00891849],"genre_scores_gemma":[0.25663736,0.00067356805,0.71539766,0.00042379048,0.00024335671,0.0005057827,0.009912255,0.0020704654,0.014135736],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9986986,0.0002309792,0.00012711108,0.00023058326,0.00060782535,0.00010494505],"domain_scores_gemma":[0.9981304,0.0005490819,0.000143654,0.000607425,0.00046153204,0.00010793246],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012249296,0.0006355576,0.00072500296,0.0009525071,0.00066093827,0.0010783852,0.002021729,0.0006327538,0.011136613],"category_scores_gemma":[0.0059810868,0.00040599718,0.00058967003,0.0014617229,0.0008782407,0.0017259783,0.0024001934,0.0009290772,0.009234875],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001615801,0.00015828894,0.0051684803,0.0009871981,0.000112095295,0.00025985282,0.000460339,0.04329789,0.045042094,0.04136975,0.07508264,0.7864456],"study_design_scores_gemma":[0.00036887793,0.0009581667,0.003337289,0.00016895735,0.00009384479,0.0015799772,0.0002537536,0.5846991,0.15500417,0.09393184,0.1593722,0.00023190223],"about_ca_topic_score_codex":0.0015493548,"about_ca_topic_score_gemma":0.0018730557,"teacher_disagreement_score":0.011136613,"about_ca_system_score_codex":0.00058308116,"about_ca_system_score_gemma":0.0014061568,"threshold_uncertainty_score":0.037255704},"labels":[],"label_agreement":null},{"id":"W2467728588","doi":"10.1093/bioinformatics/btw382","title":"ChAsE: chromatin analysis and exploration tool","year":2016,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":47,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"BC Cancer Agency; Canada's Michael Smith Genome Sciences Centre; Simon Fraser University; University of British Columbia","funders":"Canadian Institutes of Health Research","keywords":"Chromatin; Computer science; Computational biology; Biology; Genetics; DNA","score_opus":0.007374422358469707,"score_gpt":0.21430018004961965,"score_spread":0.20692575769114993,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2467728588","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0042276,0.0008020242,0.3914509,0.00038756587,0.00026232898,0.00045917789,0.0437235,0.5543835,0.0043033767],"genre_scores_gemma":[0.052473467,0.0014814544,0.6185928,0.0018672721,0.0003176237,0.0045600072,0.15207237,0.14851263,0.020122347],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9992855,0.00010547311,0.00005851213,0.00016920813,0.00029655124,0.00008479021],"domain_scores_gemma":[0.998741,0.00061405654,0.00007938574,0.00019049043,0.00020652177,0.00016850352],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0021231207,0.0020435357,0.0013863906,0.0023621102,0.0009780956,0.0017526611,0.0030253485,0.0013231277,0.066856675],"category_scores_gemma":[0.0034768297,0.0014164392,0.0012432584,0.0014347018,0.00042166476,0.0018124715,0.003482625,0.0026247632,0.03275677],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0012180493,0.00023530373,0.0042700567,0.002382071,0.00036104707,0.0009164581,0.0006093834,0.0048312903,0.059682224,0.0055997763,0.74567205,0.1742223],"study_design_scores_gemma":[0.0011446591,0.0002611249,0.0094047915,0.0005287338,0.00016553278,0.001998357,0.00021750158,0.08694817,0.10645606,0.01849798,0.773843,0.0005340524],"about_ca_topic_score_codex":0.0018998954,"about_ca_topic_score_gemma":0.0031084379,"teacher_disagreement_score":0.066856675,"about_ca_system_score_codex":0.00041324954,"about_ca_system_score_gemma":0.0012002782,"threshold_uncertainty_score":0.2236579},"labels":[],"label_agreement":null},{"id":"W2474973645","doi":"10.1093/bioinformatics/btw371","title":"deBGA: read alignment with de Bruijn graph-based seed and extension","year":2016,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":105,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Hospital for Sick Children; University of Toronto","funders":"","keywords":"De Bruijn graph; De Bruijn sequence; Computer science; Graph; Limiting; Theoretical computer science; Mathematics","score_opus":0.00835405773695894,"score_gpt":0.20520258990491902,"score_spread":0.19684853216796008,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2474973645","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.013505981,0.00096634065,0.9122814,0.00033956507,0.0003742083,0.00035736588,0.0034600378,0.06546545,0.003249675],"genre_scores_gemma":[0.06112774,0.0003104651,0.91973805,0.00036578838,0.00008714981,0.00041488782,0.009555109,0.004770259,0.0036305205],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99833065,0.0004658368,0.00014726448,0.00063932367,0.00031711537,0.00009976547],"domain_scores_gemma":[0.9977895,0.001006112,0.00022545565,0.00053714187,0.00035171004,0.000090063004],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013028671,0.0019785997,0.0011359683,0.002328644,0.0011602793,0.0014180385,0.0020166498,0.0014648772,0.0074140243],"category_scores_gemma":[0.006662197,0.0010311061,0.0012209138,0.002021961,0.0009326409,0.0018993455,0.0017762013,0.0022305155,0.007199198],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014697958,0.00043984008,0.004325117,0.0024681347,0.00048636994,0.0010764399,0.0009210633,0.10772115,0.14169604,0.03338362,0.10108257,0.6049299],"study_design_scores_gemma":[0.00047666323,0.0004903316,0.0017400243,0.00019382886,0.00015513659,0.0009794384,0.00023801895,0.7099889,0.106439084,0.064056635,0.11498884,0.00025317445],"about_ca_topic_score_codex":0.0027019961,"about_ca_topic_score_gemma":0.0055410527,"teacher_disagreement_score":0.0074140243,"about_ca_system_score_codex":0.0007714759,"about_ca_system_score_gemma":0.0013783894,"threshold_uncertainty_score":0.024802327},"labels":[],"label_agreement":null},{"id":"W2479952736","doi":"10.1093/bioinformatics/btw489","title":"Motif comparison based on similarity of binding affinity profiles","year":2016,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":28,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"McGill University; Canadian Institute for Advanced Research; McGill University and Génome Québec Innovation Centre; University of Toronto","funders":"Natural Sciences and Engineering Research Council of Canada; Canadian Institutes of Health Research","keywords":"Motif (music); Sequence motif; Computational biology; Similarity (geometry); Web server; Computer science; Transcription factor; Biology; Artificial intelligence; Genetics; Gene; The Internet; World Wide Web","score_opus":0.01486414306976094,"score_gpt":0.2435220269961427,"score_spread":0.22865788392638176,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2479952736","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7437085,0.0013093712,0.22872834,0.000092205526,0.000084795596,0.00032856132,0.013597671,0.0070405477,0.0051100804],"genre_scores_gemma":[0.81728554,0.00031713085,0.16014837,0.000080811355,0.000028761406,0.00039342244,0.019152004,0.00077941414,0.0018145978],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9980597,0.00029170088,0.00022420603,0.0006016693,0.0006439671,0.00017888991],"domain_scores_gemma":[0.99671555,0.0013457903,0.00053544604,0.00042451575,0.0006829324,0.00029585173],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014014676,0.00074053206,0.0012632103,0.0062732305,0.0005711226,0.0009951063,0.0008859133,0.00068009435,0.0076477514],"category_scores_gemma":[0.0070228996,0.00036081442,0.001179552,0.0036274455,0.00032153033,0.001170641,0.0012684605,0.0008531445,0.002953239],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0043472857,0.00061406294,0.15930448,0.0019554505,0.0015316717,0.00057679106,0.0006139274,0.02024147,0.59320426,0.00577973,0.0081109665,0.20371985],"study_design_scores_gemma":[0.00024677758,0.0014875776,0.2866782,0.00015935604,0.00082841376,0.004141688,0.0007039794,0.38523078,0.27278247,0.018950015,0.028403847,0.0003869463],"about_ca_topic_score_codex":0.0013565335,"about_ca_topic_score_gemma":0.0025248048,"teacher_disagreement_score":0.0076477514,"about_ca_system_score_codex":0.00052452134,"about_ca_system_score_gemma":0.0004697469,"threshold_uncertainty_score":0.02558428},"labels":[],"label_agreement":null},{"id":"W2490276334","doi":"10.4018/978-1-4666-3604-0.ch045","title":"Incorporating Correlations among Gene Ontology Terms into Predicting Protein Functions","year":2013,"lang":"en","type":"book-chapter","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"York University; University of Toronto","funders":"","keywords":"Gene ontology; Computational biology; Computer science; Function (biology); Artificial intelligence; Biology; Gene; Evolutionary biology; Genetics; Gene expression","score_opus":0.009257427965719768,"score_gpt":0.19877013691578357,"score_spread":0.1895127089500638,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2490276334","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.03867572,0.0049625905,0.93003875,0.0012483121,0.0005393557,0.000085154395,0.001793465,0.0035754442,0.019081296],"genre_scores_gemma":[0.22203451,0.007892907,0.74673975,0.00060430006,0.0003362827,0.00016980186,0.0043169283,0.00072987244,0.01717567],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.999655,0.00005135007,0.000021225502,0.00007085403,0.00018200389,0.000019597794],"domain_scores_gemma":[0.99907696,0.00061519083,0.00005743648,0.000081235834,0.0001459918,0.000023223745],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008489103,0.0011963983,0.0005235917,0.0017663193,0.00034710532,0.0010092475,0.000613445,0.00046301782,0.0023628192],"category_scores_gemma":[0.0029827184,0.00038329326,0.0010540487,0.0022220947,0.00027756297,0.001588461,0.0007726994,0.0012036064,0.002036299],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001036543,0.00018247099,0.014909532,0.0004245423,0.00028361205,0.00027047488,0.00013252429,0.041496098,0.022529788,0.024583722,0.019834226,0.87524927],"study_design_scores_gemma":[0.000038257735,0.0001510833,0.016649492,0.00026174128,0.0003450204,0.00078777975,0.000109311746,0.75021696,0.03658964,0.12960817,0.06508905,0.00015345943],"about_ca_topic_score_codex":0.00395549,"about_ca_topic_score_gemma":0.010156304,"teacher_disagreement_score":0.00395549,"about_ca_system_score_codex":0.00039633954,"about_ca_system_score_gemma":0.0007028851,"threshold_uncertainty_score":0.00790441},"labels":[],"label_agreement":null},{"id":"W2491260480","doi":"10.4018/978-1-4666-3604-0.ch048","title":"Computational Sequence Design Techniques for DNA Microarray Technologies","year":2013,"lang":"en","type":"book-chapter","venue":"Bioinformatics","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"National Research Council Canada","funders":"","keywords":"DNA microarray; Gene chip analysis; Computational biology; Microarray; Computer science; Selection (genetic algorithm); Oligonucleotide; Biology; Data mining; DNA; Gene; Genetics; Gene expression; Artificial intelligence","score_opus":0.03884519221147376,"score_gpt":0.26852966897528324,"score_spread":0.2296844767638095,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2491260480","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00028912615,0.0009243327,0.9947495,0.00014323993,0.000097242715,0.000068627,0.00015122315,0.0008090414,0.002767684],"genre_scores_gemma":[0.0062629012,0.0015421553,0.98689055,0.0002251753,0.00009467328,0.0007129225,0.0007107735,0.0003782697,0.0031824752],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9979498,0.0007744851,0.00015615224,0.00032293383,0.00072322256,0.00007348363],"domain_scores_gemma":[0.9976872,0.0016679999,0.00008912507,0.00020872259,0.00031333393,0.000033596447],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0024255835,0.0019293453,0.001315054,0.001154677,0.0006410769,0.0015557981,0.0023813318,0.0013882674,0.014107094],"category_scores_gemma":[0.0056101796,0.0010872652,0.0019997281,0.0021811612,0.0009606736,0.0014076139,0.0014780436,0.0030466528,0.0073214755],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00012970288,0.00010787431,0.00034823065,0.0013408688,0.00016287774,0.00013458852,0.00013324295,0.22738984,0.0093485,0.28839153,0.030032527,0.44248027],"study_design_scores_gemma":[0.00007361286,0.00007297789,0.000115576375,0.00016470416,0.00004165406,0.00013523242,0.000031219355,0.5440793,0.006196053,0.32333508,0.12571876,0.000035824312],"about_ca_topic_score_codex":0.00090894685,"about_ca_topic_score_gemma":0.001247852,"teacher_disagreement_score":0.014107094,"about_ca_system_score_codex":0.0010020245,"about_ca_system_score_gemma":0.001302266,"threshold_uncertainty_score":0.04719299},"labels":[],"label_agreement":null},{"id":"W2491870111","doi":"10.4018/978-1-4666-3604-0.ch093","title":"Use of SciDBMaker as Tool for the Design of Specialized Biological Databases","year":2013,"lang":"en","type":"book-chapter","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université Laval","funders":"","keywords":"Biological database; Database; Computer science; Data science; Biology; Bioinformatics","score_opus":0.11169445205417258,"score_gpt":0.27260752855210474,"score_spread":0.16091307649793216,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2491870111","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0029027266,0.008564931,0.7998881,0.0026429957,0.0012680348,0.00045472913,0.0068052714,0.058854852,0.118618414],"genre_scores_gemma":[0.007771195,0.009917656,0.873216,0.001695459,0.0004233523,0.0005727661,0.019322187,0.012944292,0.074137054],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.998069,0.00027472622,0.00021618526,0.00033335306,0.0010415784,0.00006526447],"domain_scores_gemma":[0.9966792,0.0016666903,0.000132506,0.00050836155,0.00080055586,0.00021277717],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0037312026,0.0013639599,0.0010405539,0.006270969,0.0013344523,0.006879118,0.0033268994,0.0012105653,0.025159512],"category_scores_gemma":[0.0068289437,0.001451294,0.0014936073,0.008312162,0.0009752338,0.007762902,0.0029924333,0.0038943288,0.025148744],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00013612377,0.00009420632,0.00051034807,0.0015907146,0.00011358716,0.00066195347,0.00069378177,0.0028940954,0.0097405305,0.108832575,0.33136538,0.5433667],"study_design_scores_gemma":[0.00002928252,0.000021692565,0.0002731734,0.00028530185,0.000038170157,0.0008423768,0.000061843646,0.00624326,0.0064152866,0.034521155,0.9512154,0.00005312148],"about_ca_topic_score_codex":0.0011600986,"about_ca_topic_score_gemma":0.0019140462,"teacher_disagreement_score":0.025159512,"about_ca_system_score_codex":0.0016618116,"about_ca_system_score_gemma":0.0020647706,"threshold_uncertainty_score":0.084166944},"labels":[],"label_agreement":null},{"id":"W2493472693","doi":"10.1093/bioinformatics/btw487","title":"<i>genipe</i>: an automated genome-wide imputation pipeline with automatic reporting and statistical tools","year":2016,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":32,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal; Montreal Heart Institute","funders":"","keywords":"Computer science; Imputation (statistics); Data mining; Python (programming language); Software; Suite; Documentation; Missing data; Machine learning; Programming language","score_opus":0.017022826343548837,"score_gpt":0.28385688207671955,"score_spread":0.2668340557331707,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2493472693","genre_codex":"dataset","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0018832259,0.00040102366,0.27815247,0.0011035238,0.00036964103,0.0006505955,0.43893874,0.2718486,0.0066522225],"genre_scores_gemma":[0.013326137,0.00041186542,0.38109517,0.001437067,0.00026390588,0.0035574671,0.49506652,0.09799344,0.006848466],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9955776,0.0015134448,0.00048803817,0.0011676195,0.0008856356,0.0003677729],"domain_scores_gemma":[0.98648006,0.0058277575,0.0016470412,0.0033958678,0.00198263,0.0006666564],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0119221695,0.0026404834,0.002603516,0.004454195,0.0014321553,0.0036150203,0.006115992,0.0016811731,0.15084994],"category_scores_gemma":[0.032802794,0.0025691187,0.003215,0.0067094415,0.0009263905,0.0022428252,0.006315307,0.0035271656,0.10364855],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003755914,0.000041000632,0.0031230676,0.0011197015,0.00046802693,0.00017887089,0.00021554169,0.0019348719,0.0025304279,0.0031770095,0.93874854,0.048087344],"study_design_scores_gemma":[0.0008714505,0.00012706876,0.011717584,0.00050233933,0.0003950274,0.0008233377,0.00008400102,0.027145237,0.015284803,0.029619219,0.9129921,0.00043771445],"about_ca_topic_score_codex":0.004578328,"about_ca_topic_score_gemma":0.007803057,"teacher_disagreement_score":0.15084994,"about_ca_system_score_codex":0.001118388,"about_ca_system_score_gemma":0.004151415,"threshold_uncertainty_score":0.5046433},"labels":[],"label_agreement":null},{"id":"W2507770958","doi":"10.1093/bioinformatics/btw400","title":"LCA*: an entropy-based measure for taxonomic assignment within assembled metagenomes","year":2016,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":34,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Computer science; Taxonomic rank; Taxonomy (biology); Voting; Phylogenetic tree; Data mining; Information retrieval; Machine learning; Biology; Ecology; Genetics","score_opus":0.025684718457889524,"score_gpt":0.24164813046747244,"score_spread":0.2159634120095829,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2507770958","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.11340335,0.00096076995,0.87380767,0.0003997735,0.00013855004,0.00020908006,0.003325577,0.0041277553,0.0036274053],"genre_scores_gemma":[0.6193474,0.0002625613,0.37294102,0.0002551337,0.00023173101,0.00052963936,0.004714665,0.00080680836,0.00091099634],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9963612,0.001091605,0.00034259562,0.00060116453,0.0014154739,0.00018807851],"domain_scores_gemma":[0.9802105,0.012734555,0.0027084472,0.0019491623,0.0017976214,0.0005997471],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0052479743,0.00068229216,0.0008345163,0.006587846,0.000996811,0.0019966357,0.0016245014,0.0011333288,0.002437237],"category_scores_gemma":[0.02620943,0.000359436,0.0010841215,0.0033468793,0.0019696653,0.0023318268,0.002345491,0.0016522525,0.0007586017],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010895302,0.00031561975,0.121031865,0.0019504695,0.0007409095,0.00040548647,0.001155069,0.19565769,0.04552978,0.08392691,0.018425992,0.5297706],"study_design_scores_gemma":[0.00005237631,0.00032235813,0.029693225,0.00018933431,0.0000908858,0.00044600735,0.00025072222,0.8207389,0.025714349,0.111355215,0.011007833,0.00013879857],"about_ca_topic_score_codex":0.00094969745,"about_ca_topic_score_gemma":0.00094106264,"teacher_disagreement_score":0.006587846,"about_ca_system_score_codex":0.0011628751,"about_ca_system_score_gemma":0.0013514142,"threshold_uncertainty_score":0.027754307},"labels":[],"label_agreement":null},{"id":"W2508423234","doi":"10.1093/bioinformatics/btw467","title":"A program for verification of phylogenetic network models","year":2016,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genome Rearrangement Algorithms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":16,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Toronto Metropolitan University","funders":"","keywords":"Phylogenetic tree; Computer science; Phylogenetic network; Biology; Genetics; Gene","score_opus":0.018765223702526555,"score_gpt":0.2517337533348999,"score_spread":0.23296852963237333,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2508423234","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0041525136,0.000054030195,0.98073155,0.0003126266,0.00004084633,0.00017061265,0.00085391954,0.011639423,0.0020445012],"genre_scores_gemma":[0.11013038,0.00014056978,0.8797709,0.00032612964,0.00008508217,0.00075415586,0.0034544733,0.0023619977,0.0029764515],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99652475,0.0011023902,0.00033371575,0.0009201148,0.0007468882,0.00037214698],"domain_scores_gemma":[0.9820765,0.0139966365,0.0006160955,0.0013023507,0.0016932167,0.00031515973],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003846649,0.0018955093,0.0011462425,0.0014820534,0.0011242047,0.002313073,0.0031765155,0.0019633444,0.01954983],"category_scores_gemma":[0.02055389,0.00096345413,0.003050618,0.0012735607,0.0018449594,0.0039258525,0.004157973,0.0030812274,0.0037116355],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0016399255,0.0006609123,0.006343098,0.0022031965,0.00030493623,0.0014294928,0.00090041844,0.32970324,0.017827705,0.28388518,0.042799324,0.3123026],"study_design_scores_gemma":[0.00018023583,0.000105524494,0.00027071236,0.00009888512,0.00005236805,0.00021909016,0.00006610882,0.824439,0.0071280575,0.15425703,0.013151786,0.00003128238],"about_ca_topic_score_codex":0.0026646953,"about_ca_topic_score_gemma":0.0025758974,"teacher_disagreement_score":0.01954983,"about_ca_system_score_codex":0.0018150356,"about_ca_system_score_gemma":0.0027013868,"threshold_uncertainty_score":0.06540066},"labels":[],"label_agreement":null},{"id":"W2512452276","doi":"10.1093/bioinformatics/btw538","title":"BatchQC: interactive software for evaluating sample and batch effects in genomic data","year":2016,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":67,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Princess Margaret Cancer Centre; Institute of Cancer Research; University of Toronto; University Health Network","funders":"National Institutes of Health","keywords":"Bioconductor; Computer science; Pipeline (software); Software; Data mining; A priori and a posteriori; Sample (material); Batch processing; Data science; Operating system","score_opus":0.04013611400870473,"score_gpt":0.3309031692002055,"score_spread":0.2907670551915008,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2512452276","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0039563784,0.00028218783,0.36815348,0.00058456016,0.00036348123,0.000465685,0.055261884,0.56839824,0.002534172],"genre_scores_gemma":[0.03530321,0.0003547457,0.6717827,0.0010559651,0.00022141643,0.0035827446,0.08443858,0.19859532,0.0046654036],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9874679,0.0039070435,0.001538715,0.002946658,0.0035173302,0.00062243803],"domain_scores_gemma":[0.9306348,0.04590515,0.0034561874,0.009299004,0.009601382,0.001103443],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.03502371,0.0038931596,0.0028335992,0.0063327975,0.0028893722,0.0067060646,0.007193155,0.0021310416,0.07643316],"category_scores_gemma":[0.08470316,0.003078159,0.0043170936,0.0084669,0.001891781,0.0055897557,0.005493491,0.0050994386,0.022846049],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0017908659,0.000200766,0.013031305,0.0028742212,0.0014214037,0.00032006556,0.0014420798,0.020577243,0.012405752,0.010956503,0.7988153,0.13616446],"study_design_scores_gemma":[0.0015367572,0.00030391815,0.027815508,0.00065478816,0.0008350106,0.00070431316,0.00058406044,0.41315514,0.07033662,0.07800348,0.4047602,0.0013102246],"about_ca_topic_score_codex":0.02045248,"about_ca_topic_score_gemma":0.027758665,"teacher_disagreement_score":0.07643316,"about_ca_system_score_codex":0.0032286043,"about_ca_system_score_gemma":0.009368698,"threshold_uncertainty_score":0.2556944},"labels":[],"label_agreement":null},{"id":"W2514454343","doi":"10.1093/bioinformatics/btw569","title":"Nanocall: an open source basecaller for Oxford Nanopore sequencing data","year":2016,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":143,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto; Ontario Institute for Cancer Research","funders":"Canadian Institutes of Health Research; University of Nottingham; Oxford Nanopore Technologies","keywords":"Minion; Nanopore sequencing; Computer science; DNA sequencer; MIT License; DNA sequencing; Nanopore; Open source; Computational biology; Hybrid genome assembly; DNA; Biology; Reference genome; Genetics; Operating system; Software; Engineering","score_opus":0.06328263123334449,"score_gpt":0.29263038869075014,"score_spread":0.22934775745740565,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2514454343","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.007590911,0.0012472902,0.2030279,0.0007750762,0.00057462987,0.0007090875,0.07150655,0.70437,0.010198534],"genre_scores_gemma":[0.054011427,0.00071506033,0.5137478,0.0027279458,0.00019256446,0.0028849007,0.18943483,0.2171194,0.019166013],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99444795,0.0005651751,0.000586272,0.0016035024,0.002397961,0.0003991193],"domain_scores_gemma":[0.99344987,0.002153608,0.00063284993,0.001625983,0.0015357946,0.0006018883],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005171368,0.002770403,0.0023946716,0.0021333455,0.001523991,0.0028612579,0.0062890793,0.0024917936,0.06083064],"category_scores_gemma":[0.017384231,0.0022582642,0.0014778002,0.0015487898,0.0009070734,0.0037922394,0.0038126917,0.0030408953,0.04524883],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00350195,0.0001501065,0.0052223383,0.0036292651,0.00060178345,0.0008902591,0.0007808301,0.0050005377,0.10027018,0.007804129,0.7179381,0.15421054],"study_design_scores_gemma":[0.00094237074,0.00033406072,0.006435585,0.0006612333,0.0002430093,0.0015789153,0.00027602795,0.07280499,0.29030114,0.01620756,0.60915655,0.0010585511],"about_ca_topic_score_codex":0.007929385,"about_ca_topic_score_gemma":0.012523373,"teacher_disagreement_score":0.06083064,"about_ca_system_score_codex":0.0020890068,"about_ca_system_score_gemma":0.0032368137,"threshold_uncertainty_score":0.20349878},"labels":[],"label_agreement":null},{"id":"W2515651902","doi":"10.1093/bioinformatics/btw525","title":"Gene Slider: sequence logo interactive data-visualization for education and research","year":2016,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":20,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Computer science; Source code; JavaScript; Visualization; Computational biology; Biology; World Wide Web; Programming language; Artificial intelligence","score_opus":0.053187148363871645,"score_gpt":0.3748135805252692,"score_spread":0.3216264321613975,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2515651902","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0032719336,0.0006113065,0.17044874,0.001291373,0.0006187224,0.0005384204,0.15660296,0.65007037,0.016546277],"genre_scores_gemma":[0.037432954,0.001566567,0.4500036,0.0018534301,0.00038288216,0.0038470472,0.24730358,0.20594107,0.051668853],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99913764,0.0001491032,0.00007402701,0.00019279552,0.00031997776,0.00012637461],"domain_scores_gemma":[0.9972994,0.0010255076,0.00014114399,0.00052814867,0.00065894664,0.00034688835],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0022733049,0.002417229,0.0017244279,0.0030960196,0.000849013,0.002578069,0.003192357,0.0016116343,0.2531469],"category_scores_gemma":[0.0060713897,0.0013169314,0.0013313657,0.0026254822,0.0005948138,0.0041665174,0.003601696,0.0027821367,0.10090004],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0004935081,0.000101942795,0.0009007899,0.0008237236,0.00005948449,0.00019904856,0.0002869237,0.00044169428,0.012694358,0.0027260787,0.9151142,0.06615822],"study_design_scores_gemma":[0.0006140381,0.00015925561,0.006077276,0.00067144394,0.00008377683,0.00067659013,0.00023997168,0.022687053,0.042661365,0.019068291,0.90674,0.000320996],"about_ca_topic_score_codex":0.0023869595,"about_ca_topic_score_gemma":0.0036762531,"teacher_disagreement_score":0.2531469,"about_ca_system_score_codex":0.00085689913,"about_ca_system_score_gemma":0.0014130156,"threshold_uncertainty_score":0.8468607},"labels":[],"label_agreement":null},{"id":"W2515980744","doi":"10.1093/bioinformatics/btw425","title":"Pathway-based approach using hierarchical components of collapsed rare variants","year":2016,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":50,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"National Institutes of Health; Korea National Institute of Health; National Institute of Diabetes and Digestive and Kidney Diseases; Korea Health Industry Development Institute; National Science Foundation, United Arab Emirates; National Research Foundation","keywords":"KEGG; Computer science; Computational biology; Software; Exome; Data mining; Biology; Phenotype; Gene; Exome sequencing; Genetics; Transcriptome; Gene expression","score_opus":0.030695636914016593,"score_gpt":0.25354921627007143,"score_spread":0.22285357935605485,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2515980744","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.011129518,0.0003204873,0.9861059,0.00015135702,0.000034974,0.00010870007,0.0006285281,0.0010646209,0.0004558589],"genre_scores_gemma":[0.3372507,0.0007528617,0.6534777,0.00028676528,0.00016003738,0.0005416701,0.004115946,0.0005733518,0.0028408887],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99782926,0.000835002,0.00012100983,0.0006406104,0.00041485552,0.00015918993],"domain_scores_gemma":[0.9952224,0.002662652,0.00037023323,0.000684923,0.0007717338,0.00028802294],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002999849,0.0010098562,0.0012304262,0.0036686726,0.0009136844,0.0014743768,0.002231046,0.0009773094,0.0053498386],"category_scores_gemma":[0.010722116,0.0005982698,0.0024819437,0.0039010898,0.00094900775,0.0016636694,0.0021119276,0.0019293351,0.0010698951],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009457629,0.00026180953,0.038376134,0.00092069886,0.0015352111,0.0012108195,0.00071129616,0.47025427,0.012601009,0.073339686,0.008123299,0.39172003],"study_design_scores_gemma":[0.000057643367,0.000058952963,0.003599256,0.0000396411,0.00018828695,0.00028473928,0.000060059665,0.9310728,0.0017318825,0.059604656,0.003245642,0.000056486304],"about_ca_topic_score_codex":0.014568901,"about_ca_topic_score_gemma":0.013521505,"teacher_disagreement_score":0.014568901,"about_ca_system_score_codex":0.0008824731,"about_ca_system_score_gemma":0.0033793221,"threshold_uncertainty_score":0.028968215},"labels":[],"label_agreement":null},{"id":"W2517179206","doi":"10.1093/bioinformatics/btw536","title":"SiNVICT: ultra-sensitive detection of single nucleotide variants and indels in circulating tumour DNA","year":2016,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Cancer Genomics and Diagnostics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":87,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia; Simon Fraser University","funders":"","keywords":"Indel; Computational biology; DNA sequencing; Liquid biopsy; Biology; Deep sequencing; False positive paradox; DNA; Genome; Cancer; Single-nucleotide polymorphism; Genetics; Computer science; Gene; Genotype; Artificial intelligence","score_opus":0.00948069767126719,"score_gpt":0.2085639938520451,"score_spread":0.1990832961807779,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2517179206","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.08217669,0.0020015854,0.8899105,0.00045909602,0.0005018085,0.0002761865,0.00244457,0.016347298,0.0058823414],"genre_scores_gemma":[0.25570172,0.0006758485,0.734622,0.00045344842,0.000110015615,0.0004690455,0.0037102858,0.0010462977,0.0032112785],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99882144,0.000256989,0.00005265717,0.00028311124,0.0005309657,0.00005493064],"domain_scores_gemma":[0.9971385,0.0017756867,0.00036328295,0.0002041913,0.0004021381,0.00011628303],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001933305,0.00081315316,0.0008819338,0.001131697,0.00056336407,0.0010870125,0.0015039451,0.0011525007,0.0019766958],"category_scores_gemma":[0.004269988,0.00050374476,0.00081694167,0.00092774956,0.00062110595,0.00066894933,0.000896791,0.0010275842,0.0007257934],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013737141,0.0004012093,0.03790772,0.002775328,0.00073240086,0.0016109154,0.00049273914,0.28561822,0.1791756,0.020133287,0.026848594,0.44293025],"study_design_scores_gemma":[0.00007638204,0.00019167022,0.0031045764,0.000054543496,0.000062797626,0.00084912404,0.00003772566,0.93060774,0.05110828,0.0049114623,0.0088930605,0.00010264368],"about_ca_topic_score_codex":0.0029658054,"about_ca_topic_score_gemma":0.0056752227,"teacher_disagreement_score":0.0029658054,"about_ca_system_score_codex":0.00063430343,"about_ca_system_score_gemma":0.0015658888,"threshold_uncertainty_score":0.010224402},"labels":[],"label_agreement":null},{"id":"W2518304058","doi":"10.1093/bioinformatics/btw472","title":"On cross-conditional and fluctuation correlations in competitive RNA networks","year":2016,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Single-cell and spatial transcriptomics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Ottawa; Ottawa Hospital","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Competing endogenous RNA; Messenger RNA; microRNA; Biology; RNA; Statistical physics; Physics; Genetics; Gene","score_opus":0.009638407625203002,"score_gpt":0.2349723969062814,"score_spread":0.2253339892810784,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2518304058","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.36569676,0.0023901241,0.615683,0.0027649854,0.00010636585,0.000112218804,0.0005030272,0.00044215866,0.01230133],"genre_scores_gemma":[0.9864943,0.0010760055,0.009643661,0.0002834001,0.000115473274,0.00013286863,0.00023250532,0.000090625465,0.0019311054],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99839044,0.00067399733,0.00006325067,0.0003964331,0.00023711778,0.00023888751],"domain_scores_gemma":[0.97744715,0.016610244,0.0031408253,0.0006999402,0.0010205446,0.0010813847],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0033371479,0.0012554636,0.00140117,0.0019501713,0.0011235494,0.001954996,0.0015092245,0.0020543037,0.0032077148],"category_scores_gemma":[0.02238234,0.00083672826,0.0013769672,0.0010584117,0.0033221676,0.0042894785,0.0020769413,0.0018806601,0.00028759948],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000096045216,0.000063178944,0.003892142,0.00016161581,0.00011074259,0.00036521346,0.00022179556,0.7838057,0.0034510659,0.20236434,0.0010089676,0.00445924],"study_design_scores_gemma":[0.0000074620484,0.000034265213,0.0009758189,0.00001865828,0.000020071971,0.00007489036,0.000033421537,0.9167181,0.0003306586,0.081453525,0.00030742935,0.00002571606],"about_ca_topic_score_codex":0.003938251,"about_ca_topic_score_gemma":0.0029118434,"teacher_disagreement_score":0.003938251,"about_ca_system_score_codex":0.0023625505,"about_ca_system_score_gemma":0.0008529201,"threshold_uncertainty_score":0.017648697},"labels":[],"label_agreement":null},{"id":"W2518374209","doi":"10.1093/bioinformatics/btw463","title":"CoLoRMap: Correcting Long Reads by Mapping short reads","year":2016,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":68,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Computer science; Leverage (statistics); Sequence assembly; Source code; Reference genome; DNA sequencing; Algorithm; Data mining; Computational biology; Biology; Artificial intelligence; Genetics; Gene; Programming language","score_opus":0.014047876105899531,"score_gpt":0.22845864857618062,"score_spread":0.2144107724702811,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2518374209","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.010952056,0.001328376,0.9474032,0.00035350144,0.0006192889,0.00015436878,0.0023071258,0.03543375,0.0014483824],"genre_scores_gemma":[0.043339614,0.00070104224,0.9403022,0.00046253417,0.00016452183,0.00025388107,0.0047233403,0.005343836,0.0047089905],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9969811,0.0005169796,0.00009903767,0.0009825819,0.0013067807,0.00011360531],"domain_scores_gemma":[0.9950169,0.0019570035,0.0007706287,0.00088841555,0.0011672468,0.00019973773],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0031259924,0.0025757183,0.0010985683,0.0019906538,0.000982188,0.0016595229,0.0030296433,0.0014880154,0.0070332894],"category_scores_gemma":[0.0103670955,0.0009189209,0.0013437406,0.0025145018,0.0011467006,0.0019272555,0.0022352512,0.0024680474,0.0058829812],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008308241,0.0001822637,0.007315343,0.0026984357,0.0005953057,0.0005426055,0.000587651,0.05792108,0.1441288,0.0120383985,0.08429907,0.68886024],"study_design_scores_gemma":[0.00018224929,0.00040817272,0.0069420873,0.0003058905,0.00028932316,0.0017667905,0.00018966894,0.4453686,0.3480358,0.032233227,0.16384842,0.00042973133],"about_ca_topic_score_codex":0.0026153019,"about_ca_topic_score_gemma":0.0038441082,"teacher_disagreement_score":0.0070332894,"about_ca_system_score_codex":0.0006728897,"about_ca_system_score_gemma":0.0014612393,"threshold_uncertainty_score":0.023528695},"labels":[],"label_agreement":null},{"id":"W2520464109","doi":"10.1093/bioinformatics/btw597","title":"BOSS: a novel scaffolding algorithm based on an optimized scaffold graph","year":2016,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genome Rearrangement Algorithms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":39,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Saskatchewan","funders":"National Outstanding Youth Science Fund Project of National Natural Science Foundation of China; National Natural Science Foundation of China","keywords":"Boss; Scaffold; Computer science; Contig; Algorithm; Graph; Theoretical computer science; Genome; Database; Engineering; Biology","score_opus":0.014237031693811538,"score_gpt":0.2406790377023678,"score_spread":0.22644200600855624,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2520464109","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.012089511,0.00019404182,0.9725896,0.00012030429,0.00006991603,0.00012670219,0.00051921373,0.01247579,0.0018149485],"genre_scores_gemma":[0.058157958,0.00012299717,0.9344675,0.000113981936,0.000035824505,0.0001837106,0.0029560917,0.0014253063,0.0025367236],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99940395,0.00009096533,0.000034788383,0.00021700191,0.00017921475,0.00007416858],"domain_scores_gemma":[0.9987765,0.0004480672,0.00011548986,0.0002589875,0.00029611198,0.00010486349],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007116429,0.0016184472,0.0012871474,0.002369839,0.0013243563,0.0013017557,0.0022498746,0.001151252,0.0075406577],"category_scores_gemma":[0.0025262812,0.00073305954,0.0013678689,0.0021825067,0.0008879238,0.0018829824,0.0019712474,0.0010942926,0.0033802781],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00070096937,0.00023124104,0.0028703772,0.0006346715,0.00012972584,0.00048985577,0.00042544637,0.17187595,0.05167838,0.021609196,0.030888047,0.7184662],"study_design_scores_gemma":[0.00019285307,0.00018506532,0.0006158966,0.000044912587,0.00006270081,0.00027678037,0.00015186648,0.93876946,0.019820219,0.021323564,0.018506568,0.000050144434],"about_ca_topic_score_codex":0.004847378,"about_ca_topic_score_gemma":0.00661785,"teacher_disagreement_score":0.0075406577,"about_ca_system_score_codex":0.00089495577,"about_ca_system_score_gemma":0.0017585618,"threshold_uncertainty_score":0.025225937},"labels":[],"label_agreement":null},{"id":"W2530748177","doi":"10.1093/bioinformatics/btw634","title":"ARResT/Interrogate: an interactive immunoprofiler for IG/TR NGS data","year":2016,"lang":"en","type":"article","venue":"Bioinformatics","topic":"T-cell and B-cell Immunology","field":"Immunology and Microbiology","cited_by":106,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"OVC Pet Trust; Ministerstvo Školství, Mládeže a Tělovýchovy; Central European Institute of Technology; University of Guelph","keywords":"Computer science; Task (project management); Filter (signal processing); Interpretation (philosophy); Web application; Data science; Computational biology; World Wide Web; Biology; Programming language","score_opus":0.032016507568512785,"score_gpt":0.2756753969730611,"score_spread":0.2436588894045483,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2530748177","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.014100591,0.00041852664,0.3120578,0.00054350647,0.00018093226,0.00039583415,0.043083016,0.62653047,0.0026892745],"genre_scores_gemma":[0.1075299,0.00049633917,0.69560814,0.00087429036,0.00018361486,0.002275782,0.07268291,0.11471027,0.00563879],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99827325,0.00045468446,0.00017194118,0.00047816744,0.0004584571,0.00016349762],"domain_scores_gemma":[0.9944912,0.003869073,0.0003805336,0.0005763338,0.00037834144,0.00030455377],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00513996,0.0028401061,0.0014166388,0.00392534,0.0010163607,0.0029527547,0.0027176645,0.0013033499,0.031884044],"category_scores_gemma":[0.010105444,0.0015731765,0.0022160828,0.0018187758,0.0008397303,0.0022628403,0.003327834,0.0020742174,0.013739136],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.004222993,0.00037315598,0.022218423,0.0046697687,0.0013362955,0.0017651849,0.0019501881,0.022209372,0.0960913,0.0097594485,0.5745857,0.26081812],"study_design_scores_gemma":[0.001508719,0.00041165567,0.020636337,0.0006390138,0.00040318296,0.002379433,0.00057675986,0.52611214,0.13793194,0.03176584,0.27698386,0.0006510069],"about_ca_topic_score_codex":0.0015884028,"about_ca_topic_score_gemma":0.0032768613,"teacher_disagreement_score":0.031884044,"about_ca_system_score_codex":0.0007093651,"about_ca_system_score_gemma":0.0015688324,"threshold_uncertainty_score":0.10666275},"labels":[],"label_agreement":null},{"id":"W2531500048","doi":"10.1093/bioinformatics/btw639","title":"LDAP: a web server for lncRNA-disease association prediction","year":2016,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Cancer-related molecular mechanisms research","field":"Biochemistry, Genetics and Molecular Biology","cited_by":213,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Saskatchewan","funders":"National Natural Science Foundation of China","keywords":"Computer science; Lightweight Directory Access Protocol; World Wide Web; Association (psychology); Web server; Software; Database; Operating system; The Internet; Directory","score_opus":0.00993484447586988,"score_gpt":0.24962100539397397,"score_spread":0.2396861609181041,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2531500048","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.014430882,0.0013713259,0.28683984,0.0012421106,0.00046864443,0.000997824,0.19919716,0.48759377,0.007858407],"genre_scores_gemma":[0.10681873,0.0013829622,0.3535421,0.0010089725,0.00038259808,0.0034128951,0.50687206,0.01681826,0.009761454],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.998058,0.0005044589,0.00022257165,0.00054730824,0.0004907268,0.00017675305],"domain_scores_gemma":[0.9964988,0.0014058829,0.0002673328,0.00065194455,0.00069154357,0.00048455797],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0033265373,0.0021703194,0.0015138672,0.0042011407,0.001316312,0.0016780467,0.0029917832,0.0014691442,0.025226925],"category_scores_gemma":[0.008641293,0.0011819386,0.0017710313,0.004680112,0.00057841337,0.0023687389,0.002953076,0.0025110336,0.030876923],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0036475207,0.00065329054,0.019710423,0.002589135,0.00068350235,0.0010887859,0.0003758207,0.011019333,0.02222766,0.004558679,0.6538875,0.27955833],"study_design_scores_gemma":[0.0019110031,0.0005496283,0.030417608,0.0004818777,0.0005276008,0.0024081173,0.000398052,0.426927,0.050781332,0.050695844,0.4341411,0.0007608923],"about_ca_topic_score_codex":0.004216523,"about_ca_topic_score_gemma":0.0038099762,"teacher_disagreement_score":0.025226925,"about_ca_system_score_codex":0.00071193016,"about_ca_system_score_gemma":0.0028299894,"threshold_uncertainty_score":0.08439243},"labels":[],"label_agreement":null},{"id":"W2538873949","doi":"10.1093/bioinformatics/btw672","title":"Optimizing ChIP-seq peak detectors using visual labels and supervised machine learning","year":2016,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":39,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"McGill University","funders":"Canadian Institutes of Health Research","keywords":"Computer science; ENCODE; Artificial intelligence; Set (abstract data type); Genome browser; Pattern recognition (psychology); Supervised learning; Labeled data; Machine learning; Genome; Genomics; Artificial neural network; Gene; Biology","score_opus":0.009643948078409036,"score_gpt":0.22763551240346375,"score_spread":0.2179915643250547,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2538873949","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.081504844,0.00032258147,0.89804155,0.00029840652,0.00006839279,0.00019142107,0.0015917496,0.016599296,0.0013816573],"genre_scores_gemma":[0.27839217,0.00011659609,0.7125394,0.0003910659,0.00004970048,0.0005474219,0.004930894,0.0016994834,0.0013332352],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99781185,0.0006752771,0.00008760067,0.00093415537,0.0003513878,0.00013983829],"domain_scores_gemma":[0.9926278,0.0046529467,0.00052526477,0.0010540931,0.0009062995,0.00023365412],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0046919324,0.0019264916,0.0013059318,0.0011962543,0.00087440765,0.0014734735,0.0030015532,0.0018664153,0.0021883743],"category_scores_gemma":[0.012361246,0.0009272662,0.0014432246,0.0010039399,0.001663446,0.001669596,0.001349352,0.0020105115,0.0015283441],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00078405533,0.0005062353,0.012699674,0.00069727283,0.00028191545,0.00010275686,0.00018918236,0.7090918,0.051118325,0.004662102,0.011961319,0.20790541],"study_design_scores_gemma":[0.000040113155,0.000047031484,0.0010275001,0.000013785689,0.000017940876,0.000021763934,0.000020466117,0.97805417,0.014063121,0.006004483,0.000668406,0.000021183532],"about_ca_topic_score_codex":0.0046968763,"about_ca_topic_score_gemma":0.0069511384,"teacher_disagreement_score":0.0046968763,"about_ca_system_score_codex":0.0019344895,"about_ca_system_score_gemma":0.0019217756,"threshold_uncertainty_score":0.024813592},"labels":[],"label_agreement":null},{"id":"W2561907498","doi":"10.1093/bioinformatics/btw802","title":"Wright–Fisher exact solver (WFES): scalable analysis of population genetic models without simulation or diffusion theory","year":2016,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Evolution and Genetic Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Alberta Children's Hospital; University of Calgary","funders":"Natural Sciences and Engineering Research Council of Canada; Canada Foundation for Innovation","keywords":"Computer science; Markov chain; Population; Solver; Algorithm; Theoretical computer science; Machine learning; Programming language","score_opus":0.012193583736911793,"score_gpt":0.25581529432555206,"score_spread":0.24362171058864027,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2561907498","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0073971045,0.00024736926,0.9743129,0.00030806308,0.000086044805,0.000093742354,0.0011626346,0.011170918,0.0052212663],"genre_scores_gemma":[0.09495838,0.00026233224,0.8930123,0.0002725348,0.0000687317,0.0004322142,0.002073703,0.003101705,0.005818091],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9994407,0.00013957692,0.000045697805,0.00010262589,0.00019731559,0.00007401318],"domain_scores_gemma":[0.9979215,0.0014214352,0.00012638992,0.00017592794,0.0002543714,0.00010046438],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010797991,0.0014417769,0.0008068413,0.0005449849,0.00045986447,0.0009503791,0.002047243,0.0012279705,0.025541672],"category_scores_gemma":[0.0076488922,0.0006905674,0.001014906,0.0007232529,0.00066291407,0.0013123176,0.0015892582,0.0018521158,0.0055805496],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002444246,0.00016517946,0.005984544,0.0010440749,0.00025575346,0.00083685457,0.00036393848,0.5894354,0.006826939,0.08738519,0.057614606,0.24984299],"study_design_scores_gemma":[0.000098863544,0.000012568658,0.00016650553,0.0000341089,0.000012539231,0.00007031131,0.00002447698,0.9600608,0.0013619197,0.03159731,0.0065481653,0.000012449268],"about_ca_topic_score_codex":0.006951458,"about_ca_topic_score_gemma":0.010258561,"teacher_disagreement_score":0.025541672,"about_ca_system_score_codex":0.0005497398,"about_ca_system_score_gemma":0.0023972143,"threshold_uncertainty_score":0.085445404},"labels":[],"label_agreement":null},{"id":"W2562042122","doi":"10.1093/bioinformatics/btw810","title":"Defining the clonality of peripheral T cell lymphomas using RNA-seq","year":2016,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Lymphoma Diagnosis and Treatment","field":"Medicine","cited_by":12,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University; Canada's Michael Smith Genome Sciences Centre; Genome British Columbia; University of British Columbia; BC Cancer Agency","funders":"Canadian Institutes of Health Research","keywords":"RNA-Seq; Computational biology; RNA; Peripheral T-cell lymphoma; Peripheral; Biology; Virology; Computer science; T cell; Transcriptome; Genetics; Gene; Gene expression; Immune system","score_opus":0.02078213201240974,"score_gpt":0.26595058587321707,"score_spread":0.24516845386080732,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2562042122","genre_codex":"empirical","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.50562495,0.0025170785,0.33459988,0.00087324786,0.00026251544,0.00084571075,0.13063097,0.014739783,0.00990584],"genre_scores_gemma":[0.59652275,0.0010681006,0.30557546,0.0011007359,0.00019456602,0.001336995,0.086098105,0.004198074,0.00390524],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9986816,0.00017247409,0.00010125004,0.00061738846,0.0003454988,0.0000817996],"domain_scores_gemma":[0.9971649,0.0014902405,0.00033621973,0.00030612602,0.00058112293,0.000121436606],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00190215,0.00063067913,0.0006110846,0.0010753989,0.00067937654,0.0014815186,0.00072680885,0.0005616045,0.0042030243],"category_scores_gemma":[0.004922034,0.00035567998,0.0005342219,0.0009899291,0.00046350094,0.00044647962,0.0006188101,0.0007632934,0.0028136736],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008806159,0.00011451791,0.076995485,0.0016287809,0.00038099644,0.0005192729,0.0007295001,0.0068968395,0.8027952,0.0021163756,0.012820978,0.0941215],"study_design_scores_gemma":[0.00016161996,0.00045872416,0.19446038,0.00041533515,0.00066166074,0.0018195915,0.0007341592,0.076971725,0.6442277,0.009031773,0.07083625,0.00022111928],"about_ca_topic_score_codex":0.0017335981,"about_ca_topic_score_gemma":0.0041899104,"teacher_disagreement_score":0.0042030243,"about_ca_system_score_codex":0.00044025277,"about_ca_system_score_gemma":0.0005589051,"threshold_uncertainty_score":0.014060497},"labels":[],"label_agreement":null},{"id":"W2562850749","doi":"10.1093/bioinformatics/btw798","title":"switchde: inference of switch-like differential expression along single-cell trajectories","year":2016,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Single-cell and spatial transcriptomics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":44,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Medical Research Council; Medical Research Council Canada; Wellcome Trust; Wellcome","keywords":"Inference; Computer science; Expression (computer science); Differential (mechanical device); Computational biology; Software; Algorithm; Artificial intelligence; Biology; Programming language; Physics","score_opus":0.01616915510125579,"score_gpt":0.2229882719119528,"score_spread":0.206819116810697,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2562850749","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.067022555,0.000508248,0.9145452,0.00041960244,0.00010658762,0.000094539886,0.008885021,0.0071052723,0.0013131241],"genre_scores_gemma":[0.54753655,0.00044983294,0.41625503,0.00065737,0.00011540465,0.00076229055,0.02812622,0.0026376373,0.0034596676],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99874026,0.00045947736,0.00005622726,0.0005339697,0.00014485477,0.00006516936],"domain_scores_gemma":[0.98957014,0.008864381,0.0003900515,0.0006735278,0.00030229884,0.0001995908],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0042168573,0.0006258487,0.0010453679,0.0011526166,0.00065651676,0.0015593776,0.0017903757,0.0011050801,0.0058217477],"category_scores_gemma":[0.018136892,0.0005065558,0.001732496,0.001450604,0.0012204365,0.001140677,0.0013821345,0.0021280553,0.0015317898],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0027579535,0.00027845724,0.09359469,0.002272832,0.001960382,0.0007953423,0.0010433496,0.5578057,0.048372306,0.06366671,0.045790456,0.18166173],"study_design_scores_gemma":[0.00010292306,0.00013637794,0.00956179,0.0000561395,0.000104548446,0.00030957986,0.00012886811,0.91210365,0.009690389,0.05690563,0.010837925,0.00006218997],"about_ca_topic_score_codex":0.003366139,"about_ca_topic_score_gemma":0.0051448974,"teacher_disagreement_score":0.0058217477,"about_ca_system_score_codex":0.00081128743,"about_ca_system_score_gemma":0.0012162715,"threshold_uncertainty_score":0.022301197},"labels":[],"label_agreement":null},{"id":"W2566001817","doi":"10.1093/bioinformatics/btw793","title":"A new method for decontamination of <i>de novo</i> transcriptomes using a hierarchical clustering algorithm","year":2016,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":52,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Sherbrooke; Agriculture and Agri-Food Canada","funders":"Agriculture and Agri-Food Canada; U.S. Department of Agriculture","keywords":"Human decontamination; Computer science; Cluster analysis; Documentation; Sample (material); Scripting language; Identification (biology); Hierarchical clustering; Algorithm; Data mining; Biology; Artificial intelligence; Engineering; Programming language; Chemistry","score_opus":0.018058883457341895,"score_gpt":0.2906607890563882,"score_spread":0.2726019055990463,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2566001817","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.004590999,0.000100727535,0.9873467,0.00011259045,0.00012251522,0.00012819393,0.00044657866,0.006457195,0.000694486],"genre_scores_gemma":[0.012614456,0.000049412396,0.9829717,0.00007666269,0.000028012495,0.00014776697,0.0015657819,0.0011467212,0.0013994665],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9978765,0.00025089982,0.00018168944,0.0006766056,0.00088566815,0.00012864753],"domain_scores_gemma":[0.99742436,0.00057298125,0.00025289797,0.0006370388,0.0009683394,0.00014443592],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015044564,0.0017011233,0.00094365835,0.002227798,0.0019872324,0.0017066802,0.002390467,0.0010859118,0.004624928],"category_scores_gemma":[0.0042146575,0.0010243176,0.002269881,0.0021814136,0.00088794495,0.0011394595,0.0016998021,0.0030312396,0.0037349996],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003632161,0.00020686677,0.0049482114,0.0009190085,0.00057772954,0.00035636855,0.00085201423,0.040357422,0.19831489,0.01301235,0.034702826,0.7053892],"study_design_scores_gemma":[0.00013167389,0.00019028591,0.0072735823,0.000119598895,0.00022141184,0.00088542746,0.00020133858,0.71506757,0.16709097,0.020327192,0.08823625,0.00025472854],"about_ca_topic_score_codex":0.00884203,"about_ca_topic_score_gemma":0.012967548,"teacher_disagreement_score":0.00884203,"about_ca_system_score_codex":0.0015079129,"about_ca_system_score_gemma":0.002524512,"threshold_uncertainty_score":0.017581105},"labels":[],"label_agreement":null},{"id":"W2566610827","doi":"10.1093/bioinformatics/btw811","title":"Innovations and challenges in detecting long read overlaps: an evaluation of the state-of-the-art","year":2016,"lang":"en","type":"review","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":40,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University; Canada's Michael Smith Genome Sciences Centre; University of British Columbia","funders":"National Human Genome Research Institute; National Institutes of Health","keywords":"State (computer science); Computer science; Software; Data science; Programming language","score_opus":0.10441053174569143,"score_gpt":0.326231336371943,"score_spread":0.2218208046262516,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2566610827","genre_codex":"methods","genre_gemma":"review","domain_codex":null,"domain_gemma":"methods","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"review","genre_consensus":null,"domain_candidate":"methods","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.052316457,0.3068216,0.60611737,0.007958156,0.0018455746,0.00046097228,0.002972773,0.011333792,0.01017332],"genre_scores_gemma":[0.0851713,0.10327855,0.79035646,0.004143792,0.0009879082,0.00042861648,0.00904753,0.004068103,0.0025176827],"study_design_codex":"design_other","study_design_gemma":"systematic_review","domain_scores_codex":[0.9624735,0.012266316,0.0029136597,0.009239884,0.012318166,0.000788443],"domain_scores_gemma":[0.8771965,0.09171267,0.004145341,0.0074516046,0.017518505,0.0019753743],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.048320357,0.0033002365,0.003063447,0.008184874,0.0019982546,0.009706377,0.007468239,0.0040531405,0.0025473563],"category_scores_gemma":[0.1122527,0.0021083078,0.0035564022,0.008234879,0.0037569606,0.010727559,0.0051894295,0.0061437376,0.003006273],"study_design_candidate":"systematic_review","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010467857,0.00025515116,0.017548721,0.012157806,0.0013880163,0.000285671,0.0016126619,0.036565337,0.023074191,0.014714349,0.020715175,0.87063617],"study_design_scores_gemma":[0.0003740709,0.003262717,0.034279905,0.010500167,0.0024425262,0.004633582,0.0034350138,0.36157876,0.12125963,0.066146836,0.39065725,0.0014295377],"about_ca_topic_score_codex":0.006767411,"about_ca_topic_score_gemma":0.0060049193,"teacher_disagreement_score":0.95167965,"about_ca_system_score_codex":0.0028769032,"about_ca_system_score_gemma":0.0043241545,"threshold_uncertainty_score":0.2555455},"labels":[],"label_agreement":null},{"id":"W2567489417","doi":"10.1093/bioinformatics/btw781","title":"AnglerFish: a webserver for defining the geometry of α-helices in membrane proteins","year":2016,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Machine Learning in Bioinformatics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Biotechnology and Biological Sciences Research Council; Medical Research Council; Medical Research Council Canada; Pfizer","keywords":"Perl; Crystallography; Protein Data Bank (RCSB PDB); Helix (gastropod); Transmembrane protein; Orientation (vector space); Computer science; Membrane protein; Integral membrane protein; Protein structure; Transmembrane domain; Web server; Software; Computational biology; Biology; Membrane; Chemistry; Geometry; Biochemistry; Mathematics; Programming language; World Wide Web; The Internet","score_opus":0.010021768268550921,"score_gpt":0.24895022811555564,"score_spread":0.23892845984700473,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2567489417","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.009694684,0.002048402,0.08467102,0.0003853781,0.00039426997,0.00031742797,0.28181976,0.60159236,0.01907669],"genre_scores_gemma":[0.05191794,0.002440141,0.13837828,0.00083766016,0.00018667497,0.0014840406,0.6341897,0.15327686,0.017288692],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99897194,0.00014530489,0.00011333983,0.00021284343,0.00041601204,0.00014062309],"domain_scores_gemma":[0.99825853,0.0007226692,0.00017188193,0.00036873337,0.0002660286,0.00021219524],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0017228189,0.0033365758,0.0024968097,0.0028994265,0.0008168377,0.0027480703,0.0049983817,0.0018621013,0.12563327],"category_scores_gemma":[0.005640205,0.0014592778,0.0021023534,0.0030309747,0.00043655114,0.0044385623,0.0038034841,0.002503694,0.07615277],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014374194,0.00016433082,0.0034231476,0.0025219438,0.00022788589,0.00064705434,0.00031932982,0.0022062995,0.016978826,0.0059106424,0.8865086,0.079654366],"study_design_scores_gemma":[0.0011074846,0.00028405312,0.011376849,0.00080161483,0.00022124145,0.0016996221,0.00024205574,0.037052438,0.03646621,0.019404719,0.8908855,0.00045823256],"about_ca_topic_score_codex":0.0022538323,"about_ca_topic_score_gemma":0.0037809915,"teacher_disagreement_score":0.12563327,"about_ca_system_score_codex":0.0007996755,"about_ca_system_score_gemma":0.0012093552,"threshold_uncertainty_score":0.42028517},"labels":[],"label_agreement":null},{"id":"W2568061956","doi":"10.1093/bioinformatics/btw815","title":"Lineage-specific mutational clustering in protein structures predicts evolutionary shifts in function","year":2016,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Biology; Phylogenetic tree; Computational biology; Lineage (genetic); Genetics; Cluster analysis; Evolutionary biology; Phylogenetics; Protein family; Function (biology); Selection (genetic algorithm); Gene; Computer science","score_opus":0.009964921974192926,"score_gpt":0.20950603447699584,"score_spread":0.1995411125028029,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2568061956","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6739791,0.00063295243,0.20646624,0.0008612475,0.00009962245,0.0002562207,0.036388204,0.07446413,0.0068523274],"genre_scores_gemma":[0.6763055,0.00037056816,0.27922633,0.00020690344,0.00003794623,0.00030926007,0.03393405,0.008385805,0.0012235789],"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99941576,0.00009222338,0.00006607743,0.00022139237,0.00014147843,0.00006295605],"domain_scores_gemma":[0.9969471,0.0016069651,0.00064807,0.00026208453,0.00034851598,0.00018723478],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014806308,0.00083097257,0.00067549926,0.0020140784,0.00074838736,0.0011395491,0.0011438216,0.0008523348,0.008016945],"category_scores_gemma":[0.010280924,0.0006010239,0.0006403649,0.0022812234,0.00046586583,0.0012475627,0.000903938,0.00079614174,0.0027203474],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0031041673,0.0007522954,0.58947384,0.0020238394,0.0008893478,0.0008075856,0.0011494997,0.060569584,0.104447424,0.0067164185,0.051328838,0.17873713],"study_design_scores_gemma":[0.0003105061,0.00031397503,0.24909285,0.0002404211,0.00026263157,0.0008012443,0.00036144146,0.68700445,0.03459445,0.014161236,0.012690049,0.00016680128],"about_ca_topic_score_codex":0.0038158912,"about_ca_topic_score_gemma":0.0053216624,"teacher_disagreement_score":0.008016945,"about_ca_system_score_codex":0.0006641143,"about_ca_system_score_gemma":0.0011983794,"threshold_uncertainty_score":0.026819348},"labels":[],"label_agreement":null},{"id":"W2570372911","doi":"10.1093/bioinformatics/btw832","title":"ntCard: a streaming algorithm for cardinality estimation in genomics data","year":2017,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":80,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canada's Michael Smith Genome Sciences Centre; University of British Columbia","funders":"National Human Genome Research Institute; BC Cancer Agency; National Institutes of Health; Genome British Columbia; Genome Canada","keywords":"k-mer; Algorithm; Computer science; Hash function; Histogram; Cardinality (data modeling); Memory footprint; Genomics; Shotgun sequencing; Population; Hash table; Sequence assembly; Genome; Data mining; Biology; Artificial intelligence","score_opus":0.03918693709759102,"score_gpt":0.30188381076641124,"score_spread":0.2626968736688202,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2570372911","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.018495278,0.0003663426,0.9653764,0.0002314713,0.00013108234,0.000252491,0.0019417744,0.012444387,0.00076078885],"genre_scores_gemma":[0.07419363,0.00015084831,0.9170669,0.00016891841,0.00009791868,0.00034380803,0.006007592,0.0007564113,0.0012139542],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99759656,0.00038689276,0.00029132902,0.0006921129,0.0008970861,0.00013599098],"domain_scores_gemma":[0.9939336,0.0027732486,0.0006449481,0.0014198761,0.0009767435,0.0002515233],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0024446505,0.0011966274,0.0011670291,0.002387843,0.0008988711,0.0018021548,0.0024524739,0.0011897308,0.0033953423],"category_scores_gemma":[0.013803847,0.0006869291,0.0009623097,0.002349499,0.00085155386,0.0031590716,0.0022474686,0.0017928043,0.0019736306],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0015880984,0.0002982395,0.015464286,0.00061467715,0.00027050852,0.00044309377,0.00057773734,0.12729129,0.0432149,0.021427523,0.037882246,0.7509273],"study_design_scores_gemma":[0.00009716174,0.00006889483,0.0014319348,0.000030905197,0.0000196475,0.00030295894,0.00008736291,0.9560907,0.015665183,0.017435689,0.008727485,0.000042043455],"about_ca_topic_score_codex":0.0034225057,"about_ca_topic_score_gemma":0.0040682536,"teacher_disagreement_score":0.0034225057,"about_ca_system_score_codex":0.0014569543,"about_ca_system_score_gemma":0.0019935481,"threshold_uncertainty_score":0.012928724},"labels":[],"label_agreement":null},{"id":"W2573696390","doi":"10.1093/bioinformatics/btw773","title":"The super-n-motifs model: a novel alignment-free approach for representing and comparing RNA secondary structures","year":2016,"lang":"en","type":"article","venue":"Bioinformatics","topic":"RNA and protein synthesis mechanisms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Sherbrooke","funders":"Natural Sciences and Engineering Research Council of Canada; Canadian Network for Research and Innovation in Machining Technology, Natural Sciences and Engineering Research Council of Canada; Canada Foundation for Innovation","keywords":"Pseudoknot; Protein secondary structure; Computer science; Adjacency list; Structural motif; Nucleic acid secondary structure; RNA; Representation (politics); Theoretical computer science; Algorithm; Computational biology; Biology","score_opus":0.023506403898645757,"score_gpt":0.23715402200358585,"score_spread":0.2136476181049401,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2573696390","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.014231505,0.0002097793,0.98214394,0.000103523555,0.00002603097,0.00006272169,0.0005850128,0.001743135,0.0008942934],"genre_scores_gemma":[0.2275597,0.00029089017,0.7656003,0.00022112047,0.00006091987,0.00045996901,0.0026609073,0.0007649255,0.0023812137],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9991879,0.00021754176,0.000042597174,0.00023159252,0.0002502928,0.00007007071],"domain_scores_gemma":[0.9989139,0.000477863,0.00016701143,0.00018133207,0.00017560596,0.00008422598],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00078514125,0.00088751235,0.00096977275,0.0018261906,0.0005241163,0.0011646381,0.003081144,0.0018118204,0.003938169],"category_scores_gemma":[0.0040164306,0.0005259018,0.0015696095,0.001634475,0.0007371166,0.0025563147,0.0011461289,0.0013509382,0.0015112464],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00045905993,0.00019293185,0.0032479009,0.0002590995,0.00015893042,0.00029115123,0.00017574454,0.7794945,0.014952575,0.038874436,0.0045273122,0.15736635],"study_design_scores_gemma":[0.0000061097103,0.000024350016,0.00011114116,0.000007318421,0.000006432709,0.000046894995,0.0000075721205,0.985328,0.0007528307,0.012881345,0.00082085363,0.000007156653],"about_ca_topic_score_codex":0.007904842,"about_ca_topic_score_gemma":0.010063511,"teacher_disagreement_score":0.007904842,"about_ca_system_score_codex":0.0010657056,"about_ca_system_score_gemma":0.0014053964,"threshold_uncertainty_score":0.015717626},"labels":[],"label_agreement":null},{"id":"W2580812863","doi":"10.1093/bioinformatics/btw813","title":"BioPAXViz: a cytoscape application for the visual exploration of metabolic pathway evolution","year":2017,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Microbial Metabolic Engineering and Bioproduction","field":"Biochemistry, Genetics and Molecular Biology","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Visualization; Computer science; Documentation; Parsing; MIT License; Pathway analysis; Software; Metabolic pathway; Computational biology; Bioinformatics; Data mining; Biology; Programming language","score_opus":0.0147237461121498,"score_gpt":0.25835188617536103,"score_spread":0.24362814006321123,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2580812863","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.010161652,0.0015031167,0.20499362,0.00093950797,0.00071738136,0.000645209,0.19905604,0.56261945,0.019363973],"genre_scores_gemma":[0.0685538,0.0027180035,0.37247986,0.001835251,0.00037404068,0.0034804845,0.3916485,0.12767777,0.031232357],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9994129,0.0000679986,0.0000403472,0.00014469794,0.00026129448,0.000072724906],"domain_scores_gemma":[0.9986588,0.00067000074,0.0001001327,0.00015688989,0.00022831939,0.00018581665],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009965593,0.0018424909,0.0010606464,0.003977924,0.0010257268,0.001981305,0.0026171496,0.0014950379,0.08923596],"category_scores_gemma":[0.0037335553,0.0010581142,0.0015376579,0.0021680056,0.0003388581,0.0021082768,0.002396585,0.0027931014,0.027948266],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006788971,0.00015796915,0.0020707906,0.0043523298,0.000465373,0.0009702463,0.0006023621,0.0050379783,0.040517617,0.007333659,0.80170405,0.1361087],"study_design_scores_gemma":[0.00055336725,0.00013093157,0.009974903,0.0006931156,0.00018497992,0.001321327,0.00014368312,0.052189548,0.064270504,0.02714142,0.8429098,0.00048639785],"about_ca_topic_score_codex":0.00573123,"about_ca_topic_score_gemma":0.006618449,"teacher_disagreement_score":0.08923596,"about_ca_system_score_codex":0.0007758429,"about_ca_system_score_gemma":0.001664352,"threshold_uncertainty_score":0.29852402},"labels":[],"label_agreement":null},{"id":"W2587112482","doi":"10.1093/bioinformatics/btw826","title":"VEXOR: an integrative environment for prioritization of functional variants in fine-mapping analysis","year":2017,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université Laval; Centre hospitalier universitaire de Québec","funders":"National Human Genome Research Institute; Health Canada; National Cancer Institute; National Institutes of Health; Government of Canada; Fondation du cancer du sein du Québec; Compute Canada; Canadian Institutes of Health Research; Genome Canada","keywords":"Prioritization; Computer science; Genome-wide association study; Interface (matter); Identification (biology); Genetic variants; Association (psychology); User interface; Computational biology; Data science; Data mining; Biology; Single-nucleotide polymorphism; Genetics","score_opus":0.022914732973092038,"score_gpt":0.2727190200054641,"score_spread":0.24980428703237206,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2587112482","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.011550254,0.0016015751,0.4225962,0.0010137674,0.00038869982,0.0007099136,0.04239472,0.50947154,0.010273357],"genre_scores_gemma":[0.09183737,0.0018804838,0.7345407,0.0015319458,0.00027554875,0.0023067915,0.073654436,0.0828663,0.011106463],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99789387,0.00063832157,0.00020199805,0.00060924876,0.000475582,0.00018095439],"domain_scores_gemma":[0.9926111,0.00560078,0.00037941578,0.00066156837,0.00033861783,0.000408551],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.006723171,0.0024278192,0.0018423059,0.0035665063,0.00081034715,0.003164294,0.0037167545,0.0015133807,0.04252569],"category_scores_gemma":[0.012109532,0.001408065,0.0025019834,0.002238625,0.0008020931,0.0030451848,0.006129136,0.0019451206,0.016307855],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.007034477,0.0005402797,0.0153878685,0.0067746453,0.0015825928,0.0031969284,0.002761257,0.013961651,0.039184365,0.030171206,0.52149004,0.35791478],"study_design_scores_gemma":[0.0037823564,0.0009005397,0.02983261,0.0020547425,0.000879064,0.0038471776,0.0008171971,0.18362975,0.04876888,0.10444478,0.61992,0.0011229138],"about_ca_topic_score_codex":0.0024814196,"about_ca_topic_score_gemma":0.0030916561,"teacher_disagreement_score":0.04252569,"about_ca_system_score_codex":0.0007406771,"about_ca_system_score_gemma":0.0013089186,"threshold_uncertainty_score":0.14226258},"labels":[],"label_agreement":null},{"id":"W2587686048","doi":"10.1093/bioinformatics/btx078","title":"Kollector: transcript-informed, targeted <i>de novo</i> assembly of gene loci","year":2017,"lang":"en","type":"erratum","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":26,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University; Canada's Michael Smith Genome Sciences Centre; University of British Columbia","funders":"National Human Genome Research Institute; Genome British Columbia","keywords":"Sequence assembly; Computational biology; Genome; Biology; Genomics; Pipeline (software); Caenorhabditis elegans; Gene; Homo sapiens; Computer science; Genetics; Transcriptome; Gene expression","score_opus":0.015252239868434356,"score_gpt":0.24622726838113668,"score_spread":0.23097502851270232,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2587686048","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.032308273,0.0007549128,0.7966564,0.00020598977,0.0001402941,0.00035153676,0.0054660854,0.15926525,0.004851175],"genre_scores_gemma":[0.10391423,0.0006309748,0.8549218,0.0006117773,0.0000623978,0.00054251635,0.018851921,0.013797804,0.0066666035],"study_design_codex":"bench_or_experimental","study_design_gemma":"not_applicable","domain_scores_codex":[0.9990212,0.00011766024,0.00007814322,0.00033909178,0.00035157724,0.000092326154],"domain_scores_gemma":[0.99888927,0.0003632648,0.00019857819,0.0002971024,0.00015824485,0.00009351926],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013974813,0.001766663,0.00080950867,0.0008219289,0.00071008515,0.0013308681,0.0024248208,0.0012433833,0.008049193],"category_scores_gemma":[0.0019155572,0.0011200021,0.0012420949,0.00070277293,0.0007780085,0.0011625251,0.0017068207,0.0020720293,0.007080309],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0017861698,0.00018159267,0.004188067,0.0019611083,0.00028253562,0.00087727315,0.00056338514,0.029915528,0.6692915,0.006249498,0.05494447,0.2297588],"study_design_scores_gemma":[0.00032331125,0.00043005613,0.002619126,0.00015177894,0.0000948904,0.000963535,0.00011751005,0.21573131,0.7135344,0.0036178404,0.062185466,0.00023072961],"about_ca_topic_score_codex":0.0023606217,"about_ca_topic_score_gemma":0.0051778997,"teacher_disagreement_score":0.008049193,"about_ca_system_score_codex":0.0009944532,"about_ca_system_score_gemma":0.0013867519,"threshold_uncertainty_score":0.026927233},"labels":[],"label_agreement":null},{"id":"W2588631423","doi":"10.1093/bioinformatics/btx102","title":"<i>GARLIC</i>: Genomic Autozygosity Regions Likelihood-based Inference and Classification","year":2017,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":38,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Manitoba","funders":"National Human Genome Research Institute; National Institutes of Health","keywords":"Inference; Computer science; Artificial intelligence","score_opus":0.03143508679184543,"score_gpt":0.28629253257251003,"score_spread":0.2548574457806646,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2588631423","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.004318921,0.0003216111,0.47774565,0.0005011745,0.00031200674,0.00021077092,0.039246652,0.47137886,0.0059643895],"genre_scores_gemma":[0.054642774,0.00036833689,0.73077315,0.0011437712,0.00036813534,0.0010935285,0.07787255,0.12466588,0.009071844],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9986327,0.0003216571,0.00011937166,0.0004786004,0.00031499352,0.0001325686],"domain_scores_gemma":[0.99522805,0.0026342673,0.00045813323,0.0009916973,0.00044589664,0.00024189723],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003207931,0.0026051793,0.0019120023,0.0029020663,0.001295755,0.00320159,0.0046092244,0.0015544078,0.09167795],"category_scores_gemma":[0.016680002,0.0019396389,0.0024192468,0.0023123391,0.0011669124,0.0019816388,0.002497064,0.0029268265,0.049430266],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010391698,0.00018448857,0.009054825,0.0016692628,0.0007329743,0.0005086972,0.00050372886,0.024148742,0.008469777,0.015504715,0.7452904,0.19289322],"study_design_scores_gemma":[0.0010419271,0.00022526411,0.008649767,0.00054290803,0.00040908612,0.0013344174,0.00011542546,0.56138766,0.041164465,0.10154625,0.2831496,0.00043317417],"about_ca_topic_score_codex":0.0029171172,"about_ca_topic_score_gemma":0.0052883727,"teacher_disagreement_score":0.09167795,"about_ca_system_score_codex":0.0010510074,"about_ca_system_score_gemma":0.0019959926,"threshold_uncertainty_score":0.30669332},"labels":[],"label_agreement":null},{"id":"W2596012462","doi":"10.1093/bioinformatics/btw744","title":"node.dating: dating ancestors in phylogenetic trees in R","year":2016,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Evolution and Paleontology Studies","field":"Earth and Planetary Sciences","cited_by":21,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia; Western University; AIDS Vancouver; Simon Fraser University","funders":"Canadian Institutes of Health Research","keywords":"Phylogenetic tree; Node (physics); Divergence (linguistics); Tree (set theory); ENCODE; Computer science; Software; Phylogenetics; Evolutionary biology; Biology; Gene; Mathematics; Genetics; Combinatorics","score_opus":0.03289374500053088,"score_gpt":0.24061457772664732,"score_spread":0.20772083272611644,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2596012462","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0029765826,0.0013840784,0.45874357,0.00076224166,0.00094697,0.000339898,0.16809343,0.35788792,0.008865283],"genre_scores_gemma":[0.030211385,0.0010319466,0.47921222,0.0010126114,0.00043590108,0.002148068,0.1516563,0.32698077,0.0073108594],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99322915,0.002573729,0.00072291854,0.002092822,0.0010505391,0.00033073392],"domain_scores_gemma":[0.9804082,0.011589686,0.001905866,0.0035519407,0.001838571,0.00070579717],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.009970312,0.0036185277,0.0039448505,0.0046538594,0.0016920302,0.0041103996,0.006291231,0.0018098482,0.10519986],"category_scores_gemma":[0.060822416,0.00416192,0.0033895692,0.005023969,0.0015770737,0.0060164486,0.004203513,0.0073973257,0.11789892],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00065910805,0.000060993905,0.0061207614,0.0048243385,0.0012773962,0.00039926023,0.0013159037,0.005779939,0.0034590664,0.020137215,0.88495797,0.07100806],"study_design_scores_gemma":[0.00062081026,0.00012332064,0.0053361696,0.0010431992,0.0007721894,0.0010464541,0.00018791783,0.030009689,0.0070461463,0.07976736,0.8735326,0.00051401],"about_ca_topic_score_codex":0.0031203746,"about_ca_topic_score_gemma":0.004727678,"teacher_disagreement_score":0.10519986,"about_ca_system_score_codex":0.0008601594,"about_ca_system_score_gemma":0.0034186158,"threshold_uncertainty_score":0.3519286},"labels":[],"label_agreement":null},{"id":"W2604472817","doi":"10.1093/bioinformatics/btx202","title":"EzMap: a simple pipeline for reproducible analysis of the human virome","year":2017,"lang":"en","type":"article","venue":"Bioinformatics","topic":"vaccines and immunoinformatics approaches","field":"Biochemistry, Genetics and Molecular Biology","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary; Libin Cardiovascular Institute of Alberta; Alberta Children's Hospital","funders":"Alberta Children's Hospital Research Institute; Canada Foundation for Innovation","keywords":"Human virome; Computer science; Metagenomics; Computational biology; Pipeline (software); Workstation; Data mining; Biology; Programming language; Genetics","score_opus":0.030205647209669707,"score_gpt":0.28950098630569654,"score_spread":0.2592953390960268,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2604472817","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.009346857,0.00086866424,0.5106636,0.0009205642,0.0005380938,0.00078937656,0.11388907,0.3560205,0.0069632954],"genre_scores_gemma":[0.03965112,0.0008575753,0.7124132,0.00089900196,0.00019907956,0.0036678372,0.18292668,0.053920835,0.005464728],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99802655,0.00034678093,0.00014594056,0.00077783427,0.00049070007,0.00021221393],"domain_scores_gemma":[0.99771416,0.0010195202,0.00020322543,0.00043403427,0.0004253814,0.00020361286],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0036389453,0.0033672638,0.0020464435,0.00344624,0.0020490754,0.0034155957,0.0034707864,0.0014124345,0.039292235],"category_scores_gemma":[0.010070692,0.0020844205,0.002843763,0.0026820556,0.0007824875,0.0027205455,0.0047953264,0.003722255,0.030307375],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0024087762,0.00021793703,0.005913291,0.0038677962,0.0011732432,0.00079738663,0.0012721292,0.009607001,0.08292333,0.010116218,0.66009074,0.22161208],"study_design_scores_gemma":[0.0011960976,0.00031605267,0.018244004,0.00047813458,0.00052067044,0.0012467729,0.0005553294,0.15030771,0.10341838,0.060392957,0.66261774,0.0007062574],"about_ca_topic_score_codex":0.0029335287,"about_ca_topic_score_gemma":0.004284323,"teacher_disagreement_score":0.039292235,"about_ca_system_score_codex":0.00097016274,"about_ca_system_score_gemma":0.0027749913,"threshold_uncertainty_score":0.13144565},"labels":[],"label_agreement":null},{"id":"W2607390459","doi":"10.1093/bioinformatics/btx213","title":"BioCIDER: a Contextualisation InDEx for biological Resources discovery","year":2017,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Biomedical Text Mining and Ontologies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ontario Institute for Cancer Research","funders":"Biotechnology and Biological Sciences Research Council; Directorate for Biological Sciences; European Commission","keywords":"Contextualization; Documentation; Computer science; World Wide Web; Source code; Index (typography); Code (set theory); Open source; Software; Open source software; Data science; Programming language","score_opus":0.046255479885533336,"score_gpt":0.30591285235299864,"score_spread":0.2596573724674653,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2607390459","genre_codex":"dataset","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.011267122,0.00991121,0.21817979,0.0042079925,0.00095034484,0.0017533978,0.52129143,0.20651232,0.025926407],"genre_scores_gemma":[0.022955503,0.0044038203,0.34487003,0.0013993502,0.0002702727,0.0019361643,0.60733736,0.013172071,0.0036553754],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9958995,0.00074185577,0.00092152326,0.0009410179,0.0012918301,0.00020432695],"domain_scores_gemma":[0.99271303,0.0028446012,0.0009070405,0.0016215757,0.0010611379,0.00085257995],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004445867,0.0025234555,0.0018139577,0.016971184,0.0016846218,0.004349957,0.002077227,0.0017579978,0.017920205],"category_scores_gemma":[0.01929558,0.0011674522,0.0020598157,0.017449023,0.0007752111,0.0068325764,0.008079696,0.0023003893,0.015421417],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0012436857,0.0002065261,0.009852496,0.008305646,0.0005845839,0.00078225404,0.0012228529,0.0035144296,0.011293523,0.037547134,0.69615453,0.22929224],"study_design_scores_gemma":[0.00022770418,0.00009149533,0.0077537196,0.0010886556,0.00027501854,0.0005089843,0.0003713861,0.0081389155,0.00631328,0.036950745,0.93810993,0.00017017683],"about_ca_topic_score_codex":0.0073403628,"about_ca_topic_score_gemma":0.012302431,"teacher_disagreement_score":0.017920205,"about_ca_system_score_codex":0.002009777,"about_ca_system_score_gemma":0.0034298599,"threshold_uncertainty_score":0.05994904},"labels":[],"label_agreement":null},{"id":"W2611036498","doi":"10.1093/bioinformatics/btx281","title":"Assembling draft genomes using contiBAIT","year":2017,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canada's Michael Smith Genome Sciences Centre; University of British Columbia; BC Cancer Agency","funders":"Canadian Institutes of Health Research; Terry Fox Foundation; Terry Fox Research Institute; Michael Smith Health Research BC; Canadian Cancer Society; National Institutes of Health","keywords":"Bioconductor; Computer science; Sequence assembly; Orientation (vector space); Massively parallel; Data mining; Computational biology; Biology; Parallel computing; Genetics; Mathematics; Gene","score_opus":0.034056591695111035,"score_gpt":0.28557123785163047,"score_spread":0.2515146461565194,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2611036498","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0046379813,0.001125449,0.30305097,0.00051229884,0.0008968897,0.00073227513,0.19086063,0.4884851,0.009698395],"genre_scores_gemma":[0.011785697,0.0009315532,0.38829836,0.00033951912,0.00017459887,0.0018674354,0.45700827,0.13447548,0.0051190425],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99716777,0.0006411792,0.00036649912,0.0010669411,0.0005192583,0.00023838459],"domain_scores_gemma":[0.99333006,0.0028556709,0.0007871488,0.0013254646,0.0012983775,0.00040321986],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007939094,0.0066947886,0.0036355963,0.007854418,0.0025973346,0.00570244,0.005939565,0.0017130768,0.10084957],"category_scores_gemma":[0.023076797,0.0040623844,0.004939061,0.006981443,0.0010784279,0.0028595864,0.0042424295,0.0043208664,0.09424372],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0029599497,0.0001935654,0.0039910628,0.0095546115,0.0018762926,0.001147127,0.0017023761,0.008810724,0.03119303,0.010907391,0.7866485,0.14101537],"study_design_scores_gemma":[0.00093066826,0.00047305643,0.005732216,0.0015759083,0.0013656786,0.0012776214,0.0003076432,0.0497212,0.045565613,0.02262909,0.8698397,0.0005815315],"about_ca_topic_score_codex":0.0044308575,"about_ca_topic_score_gemma":0.006614738,"teacher_disagreement_score":0.10084957,"about_ca_system_score_codex":0.001807344,"about_ca_system_score_gemma":0.0035781036,"threshold_uncertainty_score":0.3373754},"labels":[],"label_agreement":null},{"id":"W2617255385","doi":"10.1093/bioinformatics/btx323","title":"Structurexplor: a platform for the exploration of structural features of RNA secondary structures","year":2017,"lang":"en","type":"article","venue":"Bioinformatics","topic":"RNA and protein synthesis mechanisms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Sherbrooke","funders":"","keywords":"Computer science; RNA; Software; Computational biology; Programming language; Biology; Genetics; Gene","score_opus":0.030411347495270892,"score_gpt":0.2749737539755978,"score_spread":0.24456240648032693,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2617255385","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.01328945,0.0010037316,0.29190856,0.0006934211,0.0002290519,0.0005107918,0.07687868,0.60813004,0.0073562996],"genre_scores_gemma":[0.07522246,0.0016081829,0.57749015,0.00076834287,0.00015452577,0.0032497435,0.23774476,0.09422655,0.009535318],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9990584,0.00015025871,0.00006470662,0.00022667693,0.00041574068,0.000084177926],"domain_scores_gemma":[0.99865997,0.0007416945,0.00012098036,0.00015547998,0.00015365085,0.0001681845],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002479249,0.0032388638,0.0020021114,0.0022109814,0.0011128898,0.0024384684,0.005829768,0.0016571743,0.06076733],"category_scores_gemma":[0.003988658,0.002044162,0.0019293269,0.0017015978,0.0005849423,0.002560699,0.0029207044,0.0042698416,0.024567556],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0022251338,0.0003903736,0.004722526,0.003918901,0.0008094895,0.0012802582,0.00071175495,0.007813931,0.09493771,0.013136625,0.7474724,0.12258088],"study_design_scores_gemma":[0.0035241426,0.0006023508,0.011833879,0.00051749666,0.0005504288,0.0023059822,0.00029779476,0.27388132,0.15595086,0.03477007,0.51516396,0.0006017031],"about_ca_topic_score_codex":0.0023729582,"about_ca_topic_score_gemma":0.0035155735,"teacher_disagreement_score":0.06076733,"about_ca_system_score_codex":0.0009636162,"about_ca_system_score_gemma":0.0022843273,"threshold_uncertainty_score":0.20328695},"labels":[],"label_agreement":null},{"id":"W2621457453","doi":"10.1093/bioinformatics/btx367","title":"MoDMaps3D: an interactive webtool for the quantification and 3D visualization of interrelationships in a dataset of DNA sequences","year":2017,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Fractal and DNA sequence analysis","field":"Biochemistry, Genetics and Molecular Biology","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo; Western University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Visualization; Computer science; Computational biology; DNA; Interactive visualization; Computer graphics (images); Artificial intelligence; Data mining; Biology; Genetics","score_opus":0.04514148420432668,"score_gpt":0.34420197863371754,"score_spread":0.29906049442939087,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2621457453","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0065469285,0.0008750074,0.20144182,0.00054095464,0.0004018872,0.00035101632,0.286896,0.4960755,0.006870918],"genre_scores_gemma":[0.043614317,0.0014052973,0.40535557,0.0009131616,0.00016703979,0.0025754352,0.4382215,0.09940675,0.008340893],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9989409,0.00012277247,0.00008521347,0.00026098083,0.0004805348,0.00010961533],"domain_scores_gemma":[0.9983423,0.00072960416,0.00013527289,0.00031881235,0.00027968612,0.00019430206],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016094437,0.0030312845,0.0021575934,0.0055075814,0.0011702497,0.0034869171,0.003283186,0.0015830706,0.073733516],"category_scores_gemma":[0.0050372775,0.0014772218,0.0025355213,0.0040277755,0.00060991565,0.0031827786,0.004589956,0.0026913788,0.02885479],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007020166,0.00018269202,0.0038612494,0.003046687,0.00051287847,0.00049182476,0.0006878907,0.0052440194,0.017243555,0.007921195,0.84534985,0.11475605],"study_design_scores_gemma":[0.00062442437,0.00011992995,0.0060004476,0.0005411152,0.0001760504,0.0010477608,0.000391822,0.09969979,0.031582363,0.045868766,0.81352776,0.00041979717],"about_ca_topic_score_codex":0.004089322,"about_ca_topic_score_gemma":0.008040049,"teacher_disagreement_score":0.073733516,"about_ca_system_score_codex":0.00078167586,"about_ca_system_score_gemma":0.001829657,"threshold_uncertainty_score":0.24666315},"labels":[],"label_agreement":null},{"id":"W2623917105","doi":"10.1093/bioinformatics/btx347","title":"COMPASS: the COMPletely Arbitrary Sequence Simulator","year":2017,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Carleton University","funders":"Natural Sciences and Engineering Research Council of Canada; Canadian Institutes of Health Research","keywords":"Compass; Sequence (biology); Computer science; Computer graphics (images); Simulation; Software; Programming language; Cartography; Geography; Biology","score_opus":0.04825999946533854,"score_gpt":0.28384619722334126,"score_spread":0.2355861977580027,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2623917105","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.04958721,0.0005074864,0.77920485,0.0010482052,0.0007544709,0.0005735463,0.013523486,0.13102178,0.023778956],"genre_scores_gemma":[0.5155262,0.00063566666,0.4229585,0.001225867,0.00017016279,0.0020294734,0.019967968,0.02366067,0.0138255125],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9989517,0.00033661313,0.00007612548,0.00022089022,0.0002815476,0.00013323055],"domain_scores_gemma":[0.99725574,0.0014043307,0.00015352659,0.00056005956,0.00039424057,0.00023218084],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015533292,0.0009451926,0.00082966866,0.00048205504,0.00073349767,0.0012658782,0.003642957,0.0016402847,0.019810114],"category_scores_gemma":[0.007930269,0.0008372056,0.0011469219,0.0005199757,0.0012051546,0.0017942591,0.0020438537,0.0020321307,0.0064824224],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001377148,0.00030465855,0.009212828,0.0012189153,0.00032925408,0.0006514843,0.00041993847,0.70164555,0.017061526,0.111503705,0.095714405,0.06056065],"study_design_scores_gemma":[0.00014142174,0.000087626795,0.00024742464,0.000036240217,0.0000237079,0.00009787706,0.000021659203,0.9313631,0.0068722228,0.02952247,0.03153323,0.000052999178],"about_ca_topic_score_codex":0.005579917,"about_ca_topic_score_gemma":0.0038197143,"teacher_disagreement_score":0.019810114,"about_ca_system_score_codex":0.0008157659,"about_ca_system_score_gemma":0.0026561792,"threshold_uncertainty_score":0.066271424},"labels":[],"label_agreement":null},{"id":"W2624426054","doi":"10.1093/bioinformatics/btx372","title":"Comparison of pre-processing methods for Infinium HumanMethylation450 BeadChip array","year":2017,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Epigenetics and DNA Methylation","field":"Biochemistry, Genetics and Molecular Biology","cited_by":31,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Princess Margaret Cancer Centre; University Health Network; University of Toronto; Ontario Institute for Cancer Research","funders":"Canadian Institutes of Health Research; Prostate Cancer Canada; Princess Margaret Cancer Foundation; Movember Foundation; Terry Fox Research Institute; Ontario Institute for Cancer Research","keywords":"Replicate; Computer science; Sample (material); DNA methylation; Computational biology; Data mining; Biology; Genetics; Statistics; Mathematics; Gene","score_opus":0.06380849064092295,"score_gpt":0.4320959425822919,"score_spread":0.3682874519413689,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2624426054","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.24047336,0.0075447652,0.7162632,0.0014200583,0.0009085296,0.0016094301,0.01307838,0.011328109,0.0073741917],"genre_scores_gemma":[0.2750731,0.0021970316,0.68837225,0.00069615874,0.00016661761,0.002663855,0.024159322,0.0029328782,0.0037388348],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99130857,0.0030500267,0.00069403456,0.0019123858,0.0027013414,0.00033359212],"domain_scores_gemma":[0.98457664,0.00916987,0.0006195154,0.0020987678,0.0033455444,0.00018969594],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01400287,0.0016220819,0.0010602939,0.0017034935,0.000805411,0.0018898939,0.0017238444,0.0010726235,0.0058717034],"category_scores_gemma":[0.032048967,0.0008196637,0.00200547,0.0016517878,0.000500703,0.0009202325,0.0014599131,0.0015073115,0.002664552],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.011136799,0.0009459735,0.04761197,0.0036427793,0.0024746105,0.000287016,0.0006749604,0.06859248,0.29338503,0.004165532,0.025553213,0.54152966],"study_design_scores_gemma":[0.00046524144,0.0028092924,0.107254855,0.00034237682,0.0013856996,0.00081948854,0.0004506772,0.23941329,0.5795209,0.00772858,0.059268724,0.00054094626],"about_ca_topic_score_codex":0.0032334419,"about_ca_topic_score_gemma":0.0053064534,"teacher_disagreement_score":0.01400287,"about_ca_system_score_codex":0.0009965497,"about_ca_system_score_gemma":0.0013268067,"threshold_uncertainty_score":0.074055135},"labels":[],"label_agreement":null},{"id":"W2626636543","doi":"10.1093/bioinformatics/btx327","title":"Detecting and removing multiplicative spatial bias in high-throughput screening technologies","year":2017,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Cell Image Analysis Techniques","field":"Biochemistry, Genetics and Molecular Biology","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; McGill University and Génome Québec Innovation Centre; Université du Québec à Montréal","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Computer science; False positive paradox; Multiplicative function; Protocol (science); Throughput; Data mining; Software; Machine learning; Mathematics; Medicine","score_opus":0.024382365776163813,"score_gpt":0.28665710386941745,"score_spread":0.26227473809325363,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2626636543","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.033206787,0.0013151725,0.9528561,0.0007721564,0.0002746347,0.00032883734,0.0018349341,0.00850296,0.00090837217],"genre_scores_gemma":[0.2042736,0.0012104247,0.7821615,0.0010122659,0.00021601049,0.0015692567,0.005262819,0.002840172,0.0014540001],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9672239,0.011678322,0.002318075,0.004784905,0.013214054,0.0007807441],"domain_scores_gemma":[0.8944542,0.06536973,0.010221066,0.017024813,0.012054207,0.00087597],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.030408774,0.0023494086,0.0022209394,0.0026788267,0.0012797414,0.0027215923,0.0043264297,0.001954198,0.0026301837],"category_scores_gemma":[0.08549479,0.0011775851,0.002783739,0.0040889145,0.0027647745,0.002364607,0.0037128418,0.0034885542,0.0020862948],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014087153,0.0005699734,0.0939944,0.007155422,0.0026932638,0.00083173555,0.0014149109,0.094173774,0.3196305,0.021940727,0.020133372,0.4360532],"study_design_scores_gemma":[0.0002190777,0.0007250638,0.04581769,0.0004317051,0.0009001068,0.0012023116,0.00022248844,0.22405556,0.6412161,0.04388774,0.04083476,0.00048739472],"about_ca_topic_score_codex":0.00249935,"about_ca_topic_score_gemma":0.0035099206,"teacher_disagreement_score":0.030408774,"about_ca_system_score_codex":0.0018707472,"about_ca_system_score_gemma":0.004005999,"threshold_uncertainty_score":0.16081887},"labels":[],"label_agreement":null},{"id":"W2660795308","doi":"10.1093/bioinformatics/btx388","title":"The International Society for Computational Biology and WikiProject Computational Biology: celebrating 10 years of collaboration towards open access","year":2017,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetics, Bioinformatics, and Biomedical Research","field":"Biochemistry, Genetics and Molecular Biology","cited_by":5,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canada's Michael Smith Genome Sciences Centre; University of British Columbia; BC Cancer Agency","funders":"","keywords":"Biology; Computational biology; Computer science","score_opus":0.04959906079315056,"score_gpt":0.40396770862636733,"score_spread":0.35436864783321675,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2660795308","genre_codex":"commentary","genre_gemma":"commentary","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"commentary","genre_consensus":"commentary","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.010402455,0.028771969,0.17233187,0.41499716,0.30657762,0.00029135257,0.004021319,0.016435066,0.04617126],"genre_scores_gemma":[0.06532931,0.030998517,0.47357345,0.058252867,0.084345944,0.0010198905,0.024045259,0.016282031,0.24615276],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.98557705,0.002918635,0.0007603092,0.0013442653,0.00745172,0.0019480805],"domain_scores_gemma":[0.8889613,0.016420837,0.0025962484,0.011064461,0.026454939,0.054502208],"candidate_categories":["scholarly_communication","open_science"],"consensus_categories":[],"category_scores_codex":[0.044621952,0.0017444289,0.0018623229,0.004604567,0.0031273598,0.0150569985,0.0030251734,0.006244232,0.01652847],"category_scores_gemma":[0.047580633,0.00075483066,0.0015806983,0.0034546414,0.0033518681,0.015918832,0.021052249,0.013627782,0.009549784],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00015194369,0.0001480541,0.00075153523,0.00028516955,0.00009724489,0.00012418137,0.00054771896,0.00039969594,0.0021092328,0.032318566,0.8449284,0.118138246],"study_design_scores_gemma":[0.00003395286,0.000039666986,0.0008132027,0.00017661216,0.000021618584,0.00013095107,0.00030002024,0.0015639629,0.00091041595,0.020415664,0.9755349,0.00005888742],"about_ca_topic_score_codex":0.0028123856,"about_ca_topic_score_gemma":0.0059545534,"teacher_disagreement_score":0.9969748,"about_ca_system_score_codex":0.0016682239,"about_ca_system_score_gemma":0.014041165,"threshold_uncertainty_score":0.23598629},"labels":[],"label_agreement":null},{"id":"W2668619786","doi":"10.1093/bioinformatics/btx421","title":"CoreTracker: accurate codon reassignment prediction, applied to mitochondrial genomes","year":2017,"lang":"en","type":"article","venue":"Bioinformatics","topic":"RNA and protein synthesis mechanisms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":14,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; Université de Montréal","funders":"Fonds de recherche du Québec – Nature et technologies","keywords":"Genome; Python (programming language); Computer science; Inference; False positive paradox; Pointer (user interface); Phylogenetic tree; Source code; Set (abstract data type); Computational biology; Data mining; Biology; Artificial intelligence; Genetics; Gene; Programming language","score_opus":0.017644792153679634,"score_gpt":0.24911727965160438,"score_spread":0.23147248749792473,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2668619786","genre_codex":"software","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.048565082,0.0014012343,0.4479233,0.0006714955,0.0005619648,0.0003616013,0.033718724,0.4620563,0.004740335],"genre_scores_gemma":[0.1573691,0.00062092446,0.7370002,0.0006671236,0.00019404555,0.0006864445,0.069318,0.029102983,0.0050411234],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9983935,0.00023193534,0.00012111325,0.00071064086,0.00040551592,0.00013727738],"domain_scores_gemma":[0.9970162,0.0015315311,0.00037599908,0.0004207028,0.00046613454,0.00018935798],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003226308,0.002530348,0.0014714279,0.002182922,0.0015550392,0.0022969139,0.0029134026,0.0018662286,0.014455358],"category_scores_gemma":[0.014615437,0.001415595,0.0018678105,0.0014712015,0.0008433693,0.0027545977,0.0022512462,0.002451711,0.010755838],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.002547926,0.00038735065,0.027694456,0.0030520344,0.0007971196,0.00083131425,0.0006578502,0.12268298,0.056161005,0.009626875,0.37439156,0.4011696],"study_design_scores_gemma":[0.00028158983,0.00020346475,0.0044094087,0.00024008646,0.00012480681,0.00044150781,0.00009228728,0.8876554,0.04659274,0.013923831,0.045864742,0.0001700297],"about_ca_topic_score_codex":0.0069241337,"about_ca_topic_score_gemma":0.009850591,"teacher_disagreement_score":0.014455358,"about_ca_system_score_codex":0.0012027078,"about_ca_system_score_gemma":0.002927272,"threshold_uncertainty_score":0.048357964},"labels":[],"label_agreement":null},{"id":"W2727134083","doi":"10.1093/bioinformatics/btx441","title":"Reactome enhanced pathway visualization","year":2017,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":187,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto; Ontario Institute for Cancer Research","funders":"National Institutes of Health; National Human Genome Research Institute; University of Toronto; European Bioinformatics Institute","keywords":"Scalable Vector Graphics; Computer science; Visualization; Hierarchy; Scalability; World Wide Web; Information retrieval; Data mining; Database","score_opus":0.01236784390598909,"score_gpt":0.26033991077438523,"score_spread":0.24797206686839615,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2727134083","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.021696813,0.0016421773,0.4593753,0.0024990092,0.0007972132,0.00028598378,0.122500524,0.368179,0.023024065],"genre_scores_gemma":[0.21238185,0.003651989,0.5001441,0.0013730071,0.00034091395,0.0014615981,0.21220411,0.043967493,0.024474911],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99914086,0.0001874808,0.00006363935,0.00019987815,0.0003368153,0.000071355666],"domain_scores_gemma":[0.99796104,0.00086282665,0.0001379085,0.00033301068,0.00053773937,0.00016746938],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016370503,0.0018452404,0.0010198283,0.0034178442,0.0006431552,0.0024266716,0.0018838922,0.0015152689,0.054535415],"category_scores_gemma":[0.005297044,0.0007189634,0.0019599611,0.0026953993,0.00039870042,0.0026113098,0.0026388892,0.0019264265,0.013135178],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0029954163,0.00026552615,0.007109234,0.007844351,0.0007242606,0.0016803677,0.0013726832,0.048691854,0.12240831,0.048688106,0.55522233,0.20299764],"study_design_scores_gemma":[0.00046367102,0.0001431643,0.004434444,0.0005037937,0.00024328417,0.0013833587,0.0002716056,0.17913914,0.113829136,0.05325707,0.646041,0.00029034558],"about_ca_topic_score_codex":0.002506479,"about_ca_topic_score_gemma":0.002689517,"teacher_disagreement_score":0.054535415,"about_ca_system_score_codex":0.00081036065,"about_ca_system_score_gemma":0.0013437791,"threshold_uncertainty_score":0.18243915},"labels":[],"label_agreement":null},{"id":"W2734886065","doi":"10.1093/bioinformatics/btx254","title":"Discovery and genotyping of novel sequence insertions in many sequenced individuals","year":2017,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":36,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia; Simon Fraser University","funders":"Türkiye Bilimsel ve Teknolojik Araştırma Kurumu; European Molecular Biology Organization","keywords":"Structural variation; Sequence assembly; Sequence (biology); Reference genome; Genome; Computational biology; Breakpoint; Computer science; Genotyping; Biology; Zygosity; Hybrid genome assembly; DNA sequencing; Genetics; Whole genome sequencing; Algorithm; Genotype; Gene","score_opus":0.04019711173590731,"score_gpt":0.2830124640901386,"score_spread":0.24281535235423127,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2734886065","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8065082,0.0005998168,0.18794543,0.00016167037,0.00006803646,0.00014608625,0.0021233384,0.0013185359,0.0011288924],"genre_scores_gemma":[0.74038905,0.0001516231,0.25549084,0.00015251136,0.000025032357,0.00009575078,0.0029424627,0.000095805124,0.0006568581],"study_design_codex":"observational","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99680626,0.0007586762,0.00018849138,0.0014499819,0.0006554528,0.00014117837],"domain_scores_gemma":[0.99437654,0.0028017063,0.0011649316,0.00086249056,0.00052116974,0.0002731484],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0038827846,0.00060326105,0.00083005376,0.001189796,0.0006740373,0.00091962726,0.0011268003,0.0013100217,0.00080977246],"category_scores_gemma":[0.0074263937,0.00043861847,0.00093402993,0.0011404484,0.0007212771,0.00076496997,0.001169244,0.0009200962,0.00046293528],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0015163737,0.0007065532,0.43754923,0.0007437523,0.0011820532,0.0013267875,0.00059266516,0.123307034,0.28456646,0.005386555,0.0015521528,0.1415704],"study_design_scores_gemma":[0.00012610035,0.000996003,0.114680156,0.00011670353,0.00035960146,0.0024988858,0.00032137928,0.6702757,0.19607994,0.008436041,0.005951539,0.00015804044],"about_ca_topic_score_codex":0.0011209912,"about_ca_topic_score_gemma":0.0021700829,"teacher_disagreement_score":0.0038827846,"about_ca_system_score_codex":0.00043910477,"about_ca_system_score_gemma":0.00048201345,"threshold_uncertainty_score":0.020534337},"labels":[],"label_agreement":null},{"id":"W2735438349","doi":"10.1093/bioinformatics/btx245","title":"Identification of associations between genotypes and longitudinal phenotypes via temporally-constrained group sparse canonical correlation analysis","year":2017,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Functional Brain Connectivity Studies","field":"Neuroscience","cited_by":56,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Institute on Aging; National Institute of Biomedical Imaging and Bioengineering; Canadian Institutes of Health Research; University of California, San Diego; National Institutes of Health; H. Lundbeck A/S; Servier; Nanjing University of Aeronautics and Astronautics; National Natural Science Foundation of China; Eisai; Genentech; U.S. National Library of Medicine; IXICO; Northern California Institute for Research and Education; Pfizer; Biogen; BioClinica; F. Hoffmann-La Roche; University of Southern California; Eli Lilly and Company; U.S. Department of Defense; National Center for Advancing Translational Sciences; Meso Scale Diagnostics; Alzheimer's Disease Neuroimaging Initiative; Scheme for Promotion of Academic and Research Collaboration; Novartis Pharmaceuticals Corporation; Bristol-Myers Squibb; Alzheimer's Association; Foundation for the National Institutes of Health","keywords":"Imaging genetics; Neuroimaging; Alzheimer's Disease Neuroimaging Initiative; Phenotype; Correlation; Computational biology; Computer science; Artificial intelligence; Biology; Machine learning; Cognition; Genetics; Neuroscience; Gene; Mathematics; Cognitive impairment","score_opus":0.04907499561906291,"score_gpt":0.2863039771638803,"score_spread":0.23722898154481736,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2735438349","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.026986912,0.0005011527,0.9702716,0.00043229468,0.000064682965,0.00007533671,0.0004474986,0.00048365883,0.0007369015],"genre_scores_gemma":[0.45522207,0.0013306102,0.5345205,0.0004579062,0.00025472543,0.0004373572,0.0031158742,0.0003995089,0.0042614564],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99930775,0.00025085526,0.000033132717,0.00020976101,0.0001316344,0.000066891706],"domain_scores_gemma":[0.99828607,0.0007365818,0.0002345331,0.0002065116,0.00041603993,0.00012026821],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015723918,0.0010706777,0.00090866233,0.001250581,0.00043657486,0.0007729491,0.001223263,0.00070117676,0.0022141784],"category_scores_gemma":[0.0063447147,0.00044973023,0.0011631183,0.0016644924,0.00087075727,0.0007978524,0.00093981903,0.0014000155,0.00057577423],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00038184912,0.00018429411,0.018266598,0.00039822218,0.0004883494,0.00067084195,0.0003375181,0.6356317,0.010792424,0.04323358,0.017814334,0.27180034],"study_design_scores_gemma":[0.000015373656,0.000022379998,0.0016160883,0.000020575013,0.000031422584,0.00010234557,0.00002490769,0.9852105,0.00083465007,0.010516103,0.0015818384,0.000023828086],"about_ca_topic_score_codex":0.018678907,"about_ca_topic_score_gemma":0.02028044,"teacher_disagreement_score":0.018678907,"about_ca_system_score_codex":0.00050106796,"about_ca_system_score_gemma":0.002917801,"threshold_uncertainty_score":0.03714037},"labels":[],"label_agreement":null},{"id":"W2736429011","doi":"10.1093/bioinformatics/btx459","title":"Phylotyper: <i>in silico</i> predictor of gene subtypes","year":2017,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Public Health Agency of Canada","funders":"","keywords":"Subtyping; In silico; Python (programming language); Computational biology; R package; Phylogenetic tree; Whole genome sequencing; Genome; Data mining; Biology; Computer science; Gene; Genetics; Programming language","score_opus":0.011763900604948708,"score_gpt":0.23193900143405483,"score_spread":0.22017510082910613,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2736429011","genre_codex":"dataset","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.048493158,0.00057970633,0.24643268,0.0012991601,0.00035681072,0.00049492595,0.36543798,0.32512817,0.011777304],"genre_scores_gemma":[0.14401595,0.000428646,0.31011745,0.0011018019,0.000135644,0.0012712794,0.4879005,0.04835862,0.006670144],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99883586,0.00024411816,0.00008604526,0.00041534452,0.00027813547,0.00014048036],"domain_scores_gemma":[0.9976628,0.00094157254,0.0004567103,0.0004294625,0.00030812956,0.00020138659],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0022185883,0.002267469,0.0014348816,0.0019703142,0.0009259255,0.0016320135,0.001964368,0.0009570335,0.02652017],"category_scores_gemma":[0.007063805,0.0008895039,0.0021888043,0.0015552391,0.0005020251,0.0011699543,0.0016470914,0.0015584871,0.024961645],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0030711095,0.00027680834,0.056861762,0.0031746405,0.0009879916,0.00086352014,0.00052337325,0.018496655,0.06940157,0.0069715586,0.7395085,0.099862516],"study_design_scores_gemma":[0.00091628195,0.000785102,0.061306246,0.00058289996,0.0008120163,0.0023743133,0.0003558917,0.29379147,0.15570703,0.021304283,0.461538,0.0005264154],"about_ca_topic_score_codex":0.002263566,"about_ca_topic_score_gemma":0.0028247007,"teacher_disagreement_score":0.02652017,"about_ca_system_score_codex":0.0007211438,"about_ca_system_score_gemma":0.0018038532,"threshold_uncertainty_score":0.08871877},"labels":[],"label_agreement":null},{"id":"W2742056220","doi":"10.1093/bioinformatics/btx489","title":"LRCstats, a tool for evaluating long reads correction methods","year":2017,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Computer science; Nanopore sequencing; Reference genome; Process (computing); Error detection and correction; Software; Field (mathematics); Data mining; Genome; Algorithm; Programming language; Biology; Genetics","score_opus":0.05481751731997311,"score_gpt":0.3875812066325948,"score_spread":0.33276368931262174,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2742056220","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":"methods","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":"methods","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.058800846,0.0037576002,0.5885376,0.0007683175,0.0009550183,0.0012898857,0.047896773,0.29141235,0.0065815933],"genre_scores_gemma":[0.1282164,0.0005624289,0.7931417,0.0007130477,0.0001859172,0.0024134612,0.043082654,0.030100299,0.0015839657],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9763141,0.008005926,0.003431027,0.0035471797,0.008119482,0.0005821689],"domain_scores_gemma":[0.8750007,0.09559292,0.0079189,0.008561515,0.011905229,0.0010208274],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.028965484,0.0043498124,0.0020253719,0.008318402,0.0022065141,0.004368446,0.004376546,0.0031046798,0.009299889],"category_scores_gemma":[0.1326532,0.0012949741,0.0033312743,0.006217694,0.001527605,0.0039691376,0.0036783533,0.0028758652,0.005212065],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0045705005,0.0009963681,0.091101445,0.011087246,0.0070150616,0.0008860509,0.0017319387,0.19068462,0.03881154,0.02252638,0.23823296,0.39235592],"study_design_scores_gemma":[0.0010140222,0.0017076827,0.02123869,0.0013443222,0.0012812983,0.000985496,0.00065194233,0.7424043,0.09250194,0.028564088,0.10745994,0.00084624224],"about_ca_topic_score_codex":0.005642117,"about_ca_topic_score_gemma":0.007744723,"teacher_disagreement_score":0.9710345,"about_ca_system_score_codex":0.0017270312,"about_ca_system_score_gemma":0.0036600812,"threshold_uncertainty_score":0.15318596},"labels":[],"label_agreement":null},{"id":"W2745257241","doi":"10.1093/bioinformatics/btx498","title":"Motif independent identification of potential RNA G-quadruplexes by G4RNA screener","year":2017,"lang":"en","type":"article","venue":"Bioinformatics","topic":"DNA and Nucleic Acid Chemistry","field":"Biochemistry, Genetics and Molecular Biology","cited_by":114,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Sherbrooke","funders":"Fonds de Recherche du Québec - Santé; Natural Sciences and Engineering Research Council of Canada; Fonds Québécois de la Recherche sur la Nature et les Technologies","keywords":"G-quadruplex; RNA; Computational biology; Nucleic acid structure; Computer science; Motif (music); Artificial intelligence; Biology; Genetics; Gene; DNA; Physics","score_opus":0.007000891683450851,"score_gpt":0.23653852590421526,"score_spread":0.2295376342207644,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2745257241","genre_codex":"empirical","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.82764626,0.00042563386,0.14035384,0.00021516511,0.000059141978,0.0003710866,0.008149189,0.01958726,0.003192397],"genre_scores_gemma":[0.7891824,0.00021907894,0.19532464,0.00021239105,0.0000150872675,0.0004470097,0.010955948,0.0005737512,0.0030697894],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99954945,0.00010120677,0.000044815635,0.00015201037,0.000107774234,0.000044831708],"domain_scores_gemma":[0.998936,0.00056658447,0.00016914282,0.00008268369,0.00019061996,0.000055022716],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00079979387,0.0009242187,0.0006944951,0.0009397281,0.00024586244,0.00045513493,0.0007765157,0.000727559,0.0036010358],"category_scores_gemma":[0.0023394946,0.00020839054,0.00051935436,0.00052364566,0.0002625125,0.00054153835,0.00047075396,0.00045026536,0.0009814173],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0022679013,0.0004968146,0.03392217,0.0011593164,0.0003123598,0.0005479814,0.0001758927,0.0835325,0.704076,0.0013277628,0.005090396,0.16709085],"study_design_scores_gemma":[0.00007101604,0.00068740745,0.013805816,0.000032804288,0.0001003929,0.0002607505,0.00006203469,0.5912476,0.38966715,0.0007715673,0.0032254015,0.00006805606],"about_ca_topic_score_codex":0.000967703,"about_ca_topic_score_gemma":0.0013626101,"teacher_disagreement_score":0.0036010358,"about_ca_system_score_codex":0.00033487537,"about_ca_system_score_gemma":0.00033891076,"threshold_uncertainty_score":0.012046695},"labels":[],"label_agreement":null},{"id":"W2756665674","doi":"10.1093/bioinformatics/btx613","title":"A collaborative filtering-based approach to biomedical knowledge discovery","year":2017,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Biomedical Text Mining and Ontologies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":30,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University; Canada's Michael Smith Genome Sciences Centre; University of British Columbia","funders":"Compute Canada","keywords":"Computer science; Knowledge extraction; Strengths and weaknesses; Data science; Singular value decomposition; Graph; Knowledge graph; Data mining; Information retrieval; Machine learning; Artificial intelligence; Theoretical computer science","score_opus":0.021150662301844248,"score_gpt":0.2974989414543733,"score_spread":0.276348279152529,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2756665674","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0042462465,0.00075358414,0.9896571,0.0010926193,0.00013692111,0.00035115422,0.00059230556,0.000713991,0.0024559833],"genre_scores_gemma":[0.08615848,0.0009443286,0.9058655,0.0005483309,0.00046653635,0.00056522066,0.0019595441,0.00010429509,0.0033877941],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.98372126,0.004803622,0.0013097593,0.004020558,0.005663681,0.0004810957],"domain_scores_gemma":[0.95820373,0.027674498,0.0020866243,0.004291433,0.006835119,0.0009086146],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.014804626,0.0013515255,0.0029545508,0.014925582,0.0032666652,0.005319368,0.004824879,0.0034306718,0.0046539325],"category_scores_gemma":[0.04358276,0.00091763865,0.0036633967,0.014400766,0.0022399162,0.0037813296,0.0036796168,0.0020875838,0.0025077977],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005121794,0.0007272445,0.010080834,0.001849904,0.0014617101,0.00084274914,0.0018905881,0.098322265,0.00951456,0.06435221,0.019759221,0.79068655],"study_design_scores_gemma":[0.00013488872,0.00025650606,0.0031953398,0.00036485313,0.0006159252,0.0010036892,0.00038428552,0.78396195,0.006957218,0.16799091,0.03496587,0.0001685728],"about_ca_topic_score_codex":0.01609169,"about_ca_topic_score_gemma":0.018303962,"teacher_disagreement_score":0.01609169,"about_ca_system_score_codex":0.002290527,"about_ca_system_score_gemma":0.0053052506,"threshold_uncertainty_score":0.07829529},"labels":[],"label_agreement":null},{"id":"W2761940999","doi":"10.1093/bioinformatics/btx609","title":"A new haplotype block detection method for dense genome sequencing data based on interval graph modeling of clusters of highly correlated SNPs","year":2017,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":79,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Sinai Health System; Lunenfeld-Tanenbaum Research Institute; Public Health Ontario; University of Toronto","funders":"National Research Foundation of Korea; Canadian Institutes of Health Research; National Research Foundation; Neurosciences Research Foundation","keywords":"Haplotype; Single-nucleotide polymorphism; Computational biology; Genetics; Graph; Interval (graph theory); Genome; Biology; Block (permutation group theory); Computer science; Gene; Combinatorics; Mathematics; Theoretical computer science; Genotype","score_opus":0.05327049569628053,"score_gpt":0.30984070304935896,"score_spread":0.25657020735307845,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2761940999","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0063602724,0.00009450627,0.9880391,0.00005485583,0.00002787604,0.00007976637,0.0004600793,0.0045431657,0.0003403905],"genre_scores_gemma":[0.04306633,0.00006527899,0.9521847,0.00006124533,0.00003864463,0.000154915,0.0022711554,0.00069434085,0.0014633896],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9993144,0.00011950488,0.00003978733,0.00021195205,0.00026453004,0.000049794915],"domain_scores_gemma":[0.998585,0.0006358663,0.00015718096,0.00023206015,0.00028255553,0.00010740821],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010660777,0.0007831204,0.0008906586,0.0022226095,0.0006094937,0.00084288523,0.0017412928,0.0006760435,0.004502345],"category_scores_gemma":[0.0031890117,0.000640596,0.0012559723,0.0014876513,0.00038061116,0.0011690602,0.0011767783,0.0010940743,0.0017578037],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00040177588,0.00020817078,0.013369101,0.0004050221,0.00043953743,0.00038746267,0.00043670918,0.14666931,0.028983625,0.013192903,0.020566845,0.77493954],"study_design_scores_gemma":[0.000051866893,0.000029311894,0.001388996,0.000012928964,0.000026708085,0.000148826,0.000028308425,0.98374385,0.0034147375,0.0067639058,0.0043687453,0.00002187176],"about_ca_topic_score_codex":0.006995023,"about_ca_topic_score_gemma":0.012166858,"teacher_disagreement_score":0.006995023,"about_ca_system_score_codex":0.00053415494,"about_ca_system_score_gemma":0.0010890799,"threshold_uncertainty_score":0.015061855},"labels":[],"label_agreement":null},{"id":"W2763456665","doi":"10.1093/bioinformatics/btx648","title":"SAGE2: parallel human genome assembly","year":2017,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genome Rearrangement Algorithms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University; Western University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Genome; Sequence assembly; Computer science; Human genome; License; Graph; MIT License; Computational biology; Biology; Theoretical computer science; Genetics; Gene","score_opus":0.022374859411655616,"score_gpt":0.2756007835409523,"score_spread":0.2532259241292967,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2763456665","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0084063765,0.0010332671,0.52363515,0.00073645846,0.0008232018,0.00040931682,0.11502985,0.33015454,0.01977189],"genre_scores_gemma":[0.038650606,0.000713334,0.60221994,0.000980521,0.0003339161,0.0015251136,0.27491,0.065529436,0.015137188],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99879825,0.00023292129,0.00009473481,0.00041583183,0.00036702192,0.00009117868],"domain_scores_gemma":[0.9988808,0.00035382403,0.00006834881,0.00029386635,0.00026926602,0.00013392503],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015172485,0.0023369596,0.0013543629,0.0013344657,0.00092948234,0.0014560963,0.0022568475,0.00075995695,0.04695857],"category_scores_gemma":[0.0035173246,0.0013637586,0.0020894282,0.0018435013,0.00047470018,0.0011302636,0.0023879418,0.0022391363,0.040228624],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0015402896,0.0000736753,0.0022382566,0.001377403,0.00052583084,0.00041750108,0.00030780697,0.011428392,0.033888616,0.01023131,0.8193929,0.11857802],"study_design_scores_gemma":[0.0015343715,0.00030696078,0.0062469356,0.00031501084,0.0003530788,0.0014351895,0.00014292853,0.1712724,0.095467985,0.060034845,0.6625089,0.00038134257],"about_ca_topic_score_codex":0.0021864625,"about_ca_topic_score_gemma":0.0045763548,"teacher_disagreement_score":0.04695857,"about_ca_system_score_codex":0.00037363157,"about_ca_system_score_gemma":0.001383068,"threshold_uncertainty_score":0.15709203},"labels":[],"label_agreement":null},{"id":"W2765753182","doi":"10.1093/bioinformatics/btx667","title":"DISC: DISulfide linkage Characterization from tandem mass spectra","year":2017,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Advanced Proteomics Techniques and Applications","field":"Chemistry","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Bioinformatics Solutions (Canada); Western University","funders":"Natural Sciences and Engineering Research Council of Canada; Mitacs","keywords":"Computer science; Disulfide Linkage; Tandem mass spectrometry; Mass spectrometry; Mass spectrum; Proteome; Matching (statistics); Chemistry; Characterization (materials science); Disulfide bond; Data mining; Biological system; Materials science; Chromatography; Mathematics; Cysteine; Nanotechnology; Statistics; Biochemistry; Biology","score_opus":0.015002807750675909,"score_gpt":0.2652042303954211,"score_spread":0.2502014226447452,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2765753182","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.11548463,0.0008714274,0.77911156,0.0006958072,0.00029371688,0.0009892734,0.043283176,0.0485268,0.01074362],"genre_scores_gemma":[0.22670466,0.00057159987,0.7072538,0.00034539527,0.00012133824,0.0007156633,0.05934248,0.0016199152,0.0033252037],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99921906,0.0001206017,0.000073393756,0.00018328684,0.00035948918,0.000044194083],"domain_scores_gemma":[0.9987589,0.0003846115,0.00024188744,0.00018932804,0.0003374973,0.00008767503],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0019270803,0.0013014866,0.0006578123,0.0024052365,0.00050058914,0.0011828116,0.0022191796,0.0005307252,0.0066708582],"category_scores_gemma":[0.0041821226,0.0002850754,0.0005640451,0.0015380137,0.00043049065,0.0010648769,0.0013406415,0.0006807703,0.002747553],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0032723446,0.00053374004,0.03743452,0.004437934,0.00039736845,0.0016714559,0.00035510052,0.0767231,0.19681846,0.0217714,0.18459487,0.47198966],"study_design_scores_gemma":[0.00025855805,0.00037051883,0.008289673,0.00010505734,0.000095468895,0.0014117605,0.00018319165,0.75344074,0.17223462,0.017774053,0.045707334,0.00012898316],"about_ca_topic_score_codex":0.0010394704,"about_ca_topic_score_gemma":0.001062775,"teacher_disagreement_score":0.0066708582,"about_ca_system_score_codex":0.0007044288,"about_ca_system_score_gemma":0.0010830861,"threshold_uncertainty_score":0.022316277},"labels":[],"label_agreement":null},{"id":"W2766641567","doi":"10.1093/bioinformatics/btx701","title":"Efficient comparative phylogenetics on large trees","year":2017,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":609,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"Natural Sciences and Engineering Research Council of Canada; University of British Columbia","keywords":"Phylogenetics; Computer science; Tree (set theory); Computational biology; Biology; Mathematics; Genetics; Combinatorics; Gene","score_opus":0.02820503632907191,"score_gpt":0.282642592489065,"score_spread":0.2544375561599931,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2766641567","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.013183119,0.0020545544,0.9425362,0.0011224211,0.00034627353,0.0002442995,0.010736665,0.02417617,0.005600352],"genre_scores_gemma":[0.049412303,0.0010554256,0.91946507,0.0003484875,0.00028797155,0.00057671947,0.021967681,0.005266751,0.0016196],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.993057,0.003210796,0.0006080299,0.0015078406,0.0013168361,0.000299402],"domain_scores_gemma":[0.9734771,0.017535558,0.0013267484,0.004423298,0.0023273649,0.00090998085],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007882093,0.0022898375,0.003319197,0.0048898915,0.0025110454,0.004414309,0.0037496595,0.001762732,0.017592987],"category_scores_gemma":[0.045305155,0.0017330073,0.0028716542,0.008991838,0.0019157068,0.0062345783,0.0046275673,0.0046173194,0.016101077],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0017682445,0.00032575685,0.0071989973,0.00619144,0.0012568858,0.0010348213,0.0020508035,0.06303164,0.03440362,0.08461438,0.17419024,0.6239332],"study_design_scores_gemma":[0.0006543246,0.00031671047,0.005343098,0.000884993,0.0005049124,0.0016375324,0.0006604679,0.36486304,0.016176105,0.42372814,0.1849866,0.00024411436],"about_ca_topic_score_codex":0.0013276245,"about_ca_topic_score_gemma":0.0028359487,"teacher_disagreement_score":0.017592987,"about_ca_system_score_codex":0.001527837,"about_ca_system_score_gemma":0.002440028,"threshold_uncertainty_score":0.05885446},"labels":[],"label_agreement":null},{"id":"W2766750745","doi":"10.1093/bioinformatics/btx707","title":"NanoStringNormCNV: pre-processing of NanoString CNV data","year":2017,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Cancer Genomics and Diagnostics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Princess Margaret Cancer Centre; University Health Network; University of Toronto; Ontario Institute for Cancer Research","funders":"Canadian Institutes of Health Research; Prostate Cancer Canada; Terry Fox Research Institute; Ontario Institute for Cancer Research; Movember Foundation","keywords":"Computer science; Normalization (sociology); Visualization; Software; Copy-number variation; Data mining; Data processing; Software package; R package; Database; Computational science; Operating system; Biology; Gene","score_opus":0.03027205727068864,"score_gpt":0.2912801544147864,"score_spread":0.26100809714409773,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2766750745","genre_codex":"software","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.03244101,0.00083403464,0.35654575,0.00087204424,0.00061130413,0.0014787064,0.23137121,0.36707506,0.008770925],"genre_scores_gemma":[0.09202236,0.00046226036,0.49965805,0.0014219603,0.00017580976,0.0064511457,0.2657697,0.122470275,0.011568501],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99708205,0.0004916276,0.0002848511,0.0010799089,0.0008085669,0.0002529279],"domain_scores_gemma":[0.9947266,0.0026798344,0.00049044116,0.0011398445,0.0007484899,0.00021479595],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0058508036,0.0020423138,0.0015318744,0.0036271184,0.0010798568,0.002123915,0.002860843,0.0011152867,0.047880635],"category_scores_gemma":[0.018581549,0.0014995856,0.0019786544,0.0021834716,0.00090510835,0.0015792496,0.0029908533,0.0021164021,0.021464607],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0023060022,0.00015953335,0.028468749,0.002501625,0.0009164166,0.0010328828,0.0011241516,0.012970654,0.048615698,0.010517039,0.69467765,0.19670959],"study_design_scores_gemma":[0.0011208515,0.00041327142,0.040490042,0.00047906596,0.0003747051,0.0019488347,0.00033555215,0.13053486,0.20032561,0.03480649,0.5885023,0.000668448],"about_ca_topic_score_codex":0.0045219623,"about_ca_topic_score_gemma":0.008605192,"teacher_disagreement_score":0.047880635,"about_ca_system_score_codex":0.0015741775,"about_ca_system_score_gemma":0.0027945666,"threshold_uncertainty_score":0.16017663},"labels":[],"label_agreement":null},{"id":"W2767132092","doi":"10.1093/bioinformatics/btx675","title":"ARCS: scaffolding genome drafts with linked reads","year":2017,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":234,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Human Genome Research Institute; National Institutes of Health; Genome British Columbia; Genome Canada","keywords":"Computer science; Genome; Scaffold; Computational biology; Programming language; Biology; Genetics; Gene","score_opus":0.015951966697740978,"score_gpt":0.23972494911462772,"score_spread":0.22377298241688673,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2767132092","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.007610732,0.0006426892,0.5080303,0.0005791365,0.0005349344,0.00083025213,0.08480255,0.38883796,0.008131449],"genre_scores_gemma":[0.027150288,0.0007281461,0.70560855,0.0002991158,0.0001930028,0.0010757033,0.1973094,0.062349714,0.005286036],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9962681,0.0009704576,0.00042465472,0.0013332809,0.0008426237,0.00016080581],"domain_scores_gemma":[0.98589766,0.007415901,0.0014713889,0.0030643356,0.0015590672,0.0005917239],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007793593,0.003695384,0.0014047568,0.0050357217,0.0019825688,0.0040501063,0.0037588973,0.0020818522,0.05152974],"category_scores_gemma":[0.032500286,0.0030541557,0.0026578445,0.006031229,0.0011396349,0.0032529968,0.004995862,0.003645599,0.039786078],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.004160422,0.00037457744,0.0069694556,0.0099157635,0.001387423,0.0018182904,0.002705104,0.025308482,0.05300394,0.024977976,0.5672627,0.30211592],"study_design_scores_gemma":[0.0011414648,0.0005251952,0.0071794754,0.001213729,0.00059069,0.0013811553,0.0007034118,0.111885294,0.09497113,0.038860466,0.74099135,0.0005567009],"about_ca_topic_score_codex":0.0027164838,"about_ca_topic_score_gemma":0.0038575672,"teacher_disagreement_score":0.05152974,"about_ca_system_score_codex":0.0008535181,"about_ca_system_score_gemma":0.0025043574,"threshold_uncertainty_score":0.17238414},"labels":[],"label_agreement":null},{"id":"W2768250222","doi":"10.1093/bioinformatics/bty498","title":"A descriptive marker gene approach to single-cell pseudotime inference","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Single-cell and spatial transcriptomics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":59,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Canadian Statistical Sciences Institute; Medical Research Council","keywords":"Computer science; Inference; Python (programming language); Trajectory; Bayesian probability; Artificial intelligence; Gene; Pattern recognition (psychology); Data mining; Computational biology; Biology; Genetics","score_opus":0.0261985328342391,"score_gpt":0.2282672827784173,"score_spread":0.2020687499441782,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2768250222","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0051809116,0.00007792887,0.99356717,0.00008160857,0.00001773032,0.000018048664,0.00035045404,0.00043396663,0.00027226945],"genre_scores_gemma":[0.29263207,0.0004127127,0.6974518,0.00031489445,0.0001629185,0.00041875965,0.004110474,0.00064503023,0.0038513194],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99919635,0.00023250647,0.000044518456,0.0003043927,0.00016919593,0.00005305178],"domain_scores_gemma":[0.99545825,0.0028167886,0.00048656028,0.0006111391,0.00045455783,0.000172708],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0020765846,0.0005943445,0.00091112626,0.0011269839,0.00042942376,0.0013306774,0.0027830335,0.0012486664,0.0029460771],"category_scores_gemma":[0.011969221,0.00059139007,0.0010665003,0.0013866805,0.0012592791,0.0014772831,0.0012587305,0.002191594,0.001062609],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00046131917,0.00009942799,0.014146687,0.000305303,0.00015478954,0.00030164886,0.0003250172,0.694424,0.022138497,0.15280208,0.004330712,0.11051054],"study_design_scores_gemma":[0.000009971144,0.0000241468,0.0007690337,0.000013084641,0.000013353965,0.00007267585,0.000013432889,0.94405323,0.0032829724,0.04943788,0.0022848404,0.000025363957],"about_ca_topic_score_codex":0.004641877,"about_ca_topic_score_gemma":0.0043741865,"teacher_disagreement_score":0.004641877,"about_ca_system_score_codex":0.0012890422,"about_ca_system_score_gemma":0.0015099304,"threshold_uncertainty_score":0.010982156},"labels":[],"label_agreement":null},{"id":"W2768349867","doi":"10.1093/bioinformatics/btx749","title":"glactools: a command-line toolset for the management of genotype likelihoods and allele counts","year":2017,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":27,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Natural Sciences and Engineering Research Council of Canada; Max-Planck-Gesellschaft","keywords":"Computer science; Population; Workflow; Set (abstract data type); Scripting language; Source code; Software; Bespoke; Database; Programming language","score_opus":0.028889968602558148,"score_gpt":0.29922432142460437,"score_spread":0.2703343528220462,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2768349867","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0016944151,0.00039229865,0.19038528,0.0005650402,0.00030954593,0.00023944596,0.1360037,0.66638905,0.0040211994],"genre_scores_gemma":[0.022399763,0.00066670234,0.30196747,0.0020223823,0.00036583646,0.002540031,0.32679376,0.33562437,0.0076197567],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9954657,0.0011255885,0.0004472093,0.001165447,0.0014533922,0.00034266562],"domain_scores_gemma":[0.9843693,0.009860712,0.0012769633,0.0022272822,0.001609581,0.00065617816],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007961882,0.006173361,0.0032718256,0.0059171245,0.0021421001,0.0055599264,0.008998035,0.0023330494,0.12969336],"category_scores_gemma":[0.031738583,0.0031867432,0.004390064,0.0046255034,0.0021178962,0.0046259575,0.0073959674,0.005588926,0.115282655],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00042086805,0.00007671304,0.0023245446,0.0017401924,0.00033039227,0.00040406757,0.00042886043,0.0050487453,0.0031536098,0.0048139137,0.9355189,0.04573916],"study_design_scores_gemma":[0.0011864934,0.00017899471,0.008642249,0.0012250647,0.0003187419,0.0016545474,0.0003233192,0.109169915,0.023945134,0.1111643,0.7412644,0.0009269188],"about_ca_topic_score_codex":0.0077914847,"about_ca_topic_score_gemma":0.0071029393,"teacher_disagreement_score":0.12969336,"about_ca_system_score_codex":0.0017398149,"about_ca_system_score_gemma":0.005278997,"threshold_uncertainty_score":0.4338675},"labels":[],"label_agreement":null},{"id":"W2769159716","doi":"10.1093/bioinformatics/btx764","title":"CytoCtrlAnalyser: a Cytoscape app for biomolecular network controllability analysis","year":2017,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":27,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Saskatchewan","funders":"National Natural Science Foundation of China","keywords":"Controllability; Computer science; Biological network; Upload; Network topology; Identification (biology); Network controllability; Complex network; Network analysis; Control (management); Distributed computing; Data mining; Bioinformatics; Artificial intelligence; Computer network; World Wide Web; Biology; Engineering; Mathematics","score_opus":0.011222578171884837,"score_gpt":0.261826191908094,"score_spread":0.25060361373620915,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2769159716","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.010614516,0.0019150886,0.33908448,0.0011554271,0.0008464817,0.0012962489,0.113492474,0.51133317,0.020262148],"genre_scores_gemma":[0.110380135,0.0033538442,0.622629,0.0031690723,0.00049266237,0.010668858,0.14680353,0.06231986,0.040183086],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9993309,0.00008354181,0.00004405477,0.00014949615,0.00033345888,0.00005864445],"domain_scores_gemma":[0.9968265,0.002084164,0.00015583375,0.00026075426,0.0004590323,0.00021370286],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014109543,0.0021197633,0.0011106166,0.0034519255,0.00079521374,0.0017576063,0.0027570324,0.0015059937,0.09254752],"category_scores_gemma":[0.005466443,0.0010939696,0.0011058736,0.0015600875,0.00039187158,0.0018872711,0.0015324507,0.00244268,0.020854654],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00077298115,0.00024470288,0.0032926209,0.005500391,0.00058678084,0.0008663362,0.0005376701,0.0155925695,0.0316278,0.013933205,0.7240952,0.20294969],"study_design_scores_gemma":[0.0009110745,0.00022383958,0.0076194457,0.000928616,0.00030525558,0.0010211217,0.00018908885,0.22102696,0.11341142,0.05095247,0.6028263,0.0005844302],"about_ca_topic_score_codex":0.0035454808,"about_ca_topic_score_gemma":0.0067749927,"teacher_disagreement_score":0.09254752,"about_ca_system_score_codex":0.00076408446,"about_ca_system_score_gemma":0.002359399,"threshold_uncertainty_score":0.30960232},"labels":[],"label_agreement":null},{"id":"W2770747919","doi":"10.1093/bioinformatics/btx752","title":"Reactome diagram viewer: data structures and strategies to boost performance","year":2017,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":235,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto; Ontario Institute for Cancer Research","funders":"National Institute of General Medical Sciences; National Institutes of Health; National Human Genome Research Institute; University of Toronto; European Bioinformatics Institute","keywords":"Computer science; Visualization; Usability; Scalability; HTML5; Rendering (computer graphics); Data visualization; Data mining; Information retrieval; World Wide Web; Human–computer interaction; Database; Artificial intelligence","score_opus":0.0251877592865337,"score_gpt":0.2801560565035076,"score_spread":0.2549682972169739,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2770747919","genre_codex":"software","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.004572371,0.0012121646,0.43382853,0.0015150663,0.0006197097,0.00049772806,0.024639638,0.5278954,0.0052194446],"genre_scores_gemma":[0.04369154,0.001974364,0.6808021,0.0012444418,0.00028797006,0.0032368347,0.08953246,0.17144918,0.0077811116],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9960795,0.0011621899,0.00047671617,0.000765779,0.0012607448,0.0002550092],"domain_scores_gemma":[0.99062,0.0042570205,0.00051185995,0.0017778791,0.0022484679,0.0005848507],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008475853,0.0043709814,0.0016508744,0.0036922006,0.0012039836,0.0047975862,0.005278369,0.0022500493,0.051156733],"category_scores_gemma":[0.021948101,0.0025591285,0.0032133546,0.0028780922,0.0007555946,0.0067450535,0.0047316486,0.0053215763,0.031212304],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0035657124,0.00045257242,0.0048522362,0.0048892368,0.00054119696,0.0006227071,0.0013633015,0.011670505,0.061103724,0.02182385,0.68112266,0.20799235],"study_design_scores_gemma":[0.0012745287,0.00038126457,0.002009907,0.00077701424,0.00036593163,0.0007587086,0.00031782914,0.09926974,0.15804893,0.046433054,0.68977565,0.0005874585],"about_ca_topic_score_codex":0.0027609554,"about_ca_topic_score_gemma":0.0031100963,"teacher_disagreement_score":0.051156733,"about_ca_system_score_codex":0.0012149266,"about_ca_system_score_gemma":0.0024457462,"threshold_uncertainty_score":0.17113632},"labels":[],"label_agreement":null},{"id":"W2773943265","doi":"10.1093/bioinformatics/btx807","title":"Computational identification of micro-structural variations and their proteogenomic consequences in cancer","year":2017,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Advanced Proteomics Techniques and Applications","field":"Chemistry","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia; Simon Fraser University","funders":"National Cancer Institute; National Institutes of Health; National Institute of General Medical Sciences; Natural Sciences and Engineering Research Council of Canada; Indiana University","keywords":"Identification (biology); Computer science; Computational biology; Proteogenomics; Artificial intelligence; Biology; Genetics; Genome; Genomics; Gene","score_opus":0.02001339280471797,"score_gpt":0.2992022984635289,"score_spread":0.27918890565881094,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2773943265","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.830136,0.0019632287,0.15601173,0.0016277857,0.000082883525,0.0001281972,0.004248824,0.0032183612,0.002582962],"genre_scores_gemma":[0.89653295,0.00048277754,0.09442969,0.0004371617,0.000080255,0.00014350291,0.0070209387,0.00017454072,0.0006982354],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99963045,0.000097757984,0.000025337093,0.00012322144,0.00007594189,0.000047282185],"domain_scores_gemma":[0.9989085,0.00075113395,0.00013053467,0.000072751114,0.000070963564,0.0000661095],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012006195,0.0006744109,0.00084697554,0.0015841328,0.00052437565,0.0011451347,0.0012823085,0.0009424162,0.0010651108],"category_scores_gemma":[0.003013363,0.0003947658,0.0012761232,0.0011630551,0.0006116233,0.0005882339,0.0008689546,0.00067790225,0.0001696534],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00038267096,0.00015601498,0.032526784,0.00019919444,0.00033272785,0.00034730954,0.000060801645,0.93503326,0.002914395,0.0026829736,0.0019897772,0.023373978],"study_design_scores_gemma":[0.000012636584,0.000024005376,0.0022262842,0.000005214978,0.000025284608,0.000054172117,0.000015736505,0.99385166,0.000476541,0.0030712937,0.0002316727,0.00000546066],"about_ca_topic_score_codex":0.007026113,"about_ca_topic_score_gemma":0.009247428,"teacher_disagreement_score":0.007026113,"about_ca_system_score_codex":0.0008616717,"about_ca_system_score_gemma":0.0010635082,"threshold_uncertainty_score":0.013970435},"labels":[],"label_agreement":null},{"id":"W2774750904","doi":"10.1093/bioinformatics/btx769","title":"Model-based design of bistable cell factories for metabolic engineering","year":2017,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Microbial Metabolic Engineering and Bioproduction","field":"Biochemistry, Genetics and Molecular Biology","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"Max Planck Institute for Dynamics of Complex Technical Systems Magdeburg; Ontario Ministry of Research and Innovation; Natural Sciences and Engineering Research Council of Canada","keywords":"Bistability; Metabolic engineering; Computer science; Biology; Materials science; Genetics","score_opus":0.02065703111434475,"score_gpt":0.22984291311245614,"score_spread":0.2091858819981114,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2774750904","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.056646653,0.00044400344,0.9365843,0.00019033787,0.000037205744,0.0001264007,0.00013777573,0.0004922475,0.005341027],"genre_scores_gemma":[0.7942483,0.00060232176,0.20217228,0.000071351795,0.000015766365,0.0003906372,0.00021601311,0.00010322459,0.0021801328],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99983084,0.000036766913,0.000007946217,0.000039120936,0.000063621505,0.000021564636],"domain_scores_gemma":[0.9995858,0.00022590635,0.00009099812,0.000024461511,0.000049284332,0.000023515724],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00046948082,0.0009636189,0.0006287253,0.00040853044,0.00022083575,0.0006617116,0.0007215803,0.0007892291,0.001982801],"category_scores_gemma":[0.0015451006,0.00042520816,0.0006188013,0.0002790627,0.00054657174,0.0006316081,0.0005255415,0.0006081965,0.0002937235],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000032519252,0.000026025948,0.00019765153,0.00009270695,0.000011535207,0.000028709088,0.000013374925,0.97108465,0.015251144,0.0067866566,0.0001408304,0.006334208],"study_design_scores_gemma":[0.00001583012,0.0000512279,0.000060365306,0.00000821787,0.000008621202,0.000011100767,0.000003966012,0.99253076,0.0036769002,0.0028571035,0.0007711572,0.0000047963313],"about_ca_topic_score_codex":0.0017438037,"about_ca_topic_score_gemma":0.0024524762,"teacher_disagreement_score":0.001982801,"about_ca_system_score_codex":0.0010817101,"about_ca_system_score_gemma":0.0008965783,"threshold_uncertainty_score":0.007848382},"labels":[],"label_agreement":null},{"id":"W2779222065","doi":"10.1093/bioinformatics/btx821","title":"DyNetViewer: a Cytoscape app for dynamic network construction, analysis and visualization","year":2017,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":44,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Saskatchewan","funders":"Central South University; National Natural Science Foundation of China","keywords":"Visualization; Computer science; Cluster analysis; Network analysis; Data mining; Data visualization; Graph drawing; Machine learning","score_opus":0.006517248778063056,"score_gpt":0.25594796557713834,"score_spread":0.24943071679907527,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2779222065","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.006615988,0.001477048,0.320615,0.0010637972,0.0009331275,0.0008119222,0.1585149,0.49350023,0.01646799],"genre_scores_gemma":[0.060883358,0.0031974015,0.5220713,0.0024293715,0.00039945453,0.0074872198,0.24155319,0.12722637,0.034752287],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9992482,0.000110083456,0.00006700481,0.00018654476,0.0002953111,0.00009276341],"domain_scores_gemma":[0.99797386,0.0011353856,0.00010105245,0.00020553825,0.00038619546,0.00019810049],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013868654,0.0020995382,0.0014813337,0.0033065022,0.0010216906,0.001958541,0.0037323749,0.0016075546,0.122109696],"category_scores_gemma":[0.0053777713,0.0015120194,0.001379691,0.002050358,0.00037953685,0.0029230565,0.0024084377,0.0029963013,0.035619747],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00051063887,0.00013193408,0.0017214671,0.0041178735,0.0005344574,0.0007561634,0.00058397267,0.0065996535,0.022751184,0.010197751,0.8351095,0.11698539],"study_design_scores_gemma":[0.0006735431,0.00009452132,0.003829635,0.0006208143,0.00020959757,0.0009464869,0.00017910072,0.08793007,0.06280431,0.033047065,0.80913645,0.0005283523],"about_ca_topic_score_codex":0.005149341,"about_ca_topic_score_gemma":0.00778346,"teacher_disagreement_score":0.122109696,"about_ca_system_score_codex":0.0006821787,"about_ca_system_score_gemma":0.002273118,"threshold_uncertainty_score":0.40849763},"labels":[],"label_agreement":null},{"id":"W2780622732","doi":"10.1093/bioinformatics/btx839","title":"ChopStitch: exon annotation and splice graph construction using transcriptome assembly and whole genome sequencing data","year":2017,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canada's Michael Smith Genome Sciences Centre","funders":"National Human Genome Research Institute; National Institutes of Health","keywords":"Annotation; splice; Sequence assembly; Exon; Computational biology; Genome; Transcriptome; Computer science; Whole genome sequencing; Genome project; Graph; RNA splicing; Genetics; Exon skipping; Gene Annotation; Biology; Alternative splicing; Gene; Theoretical computer science; RNA; Gene expression","score_opus":0.05435405250254151,"score_gpt":0.2784282938971421,"score_spread":0.22407424139460058,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2780622732","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.015670886,0.00048395,0.47962672,0.0004457476,0.00042376158,0.00066410244,0.05175209,0.4462771,0.0046555223],"genre_scores_gemma":[0.04494203,0.00041430027,0.75753343,0.00064620894,0.00013817649,0.0012143713,0.13687,0.053135544,0.0051058824],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9984816,0.000207068,0.000110729656,0.0006052184,0.0004473923,0.0001480184],"domain_scores_gemma":[0.9971571,0.001261498,0.00031267325,0.00069151947,0.0003577337,0.00021952407],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002848815,0.0020612953,0.0011801764,0.0026741198,0.0013393788,0.0021952488,0.0031047491,0.0014883144,0.020044364],"category_scores_gemma":[0.008393278,0.0018242655,0.0019919807,0.0023447992,0.0010958867,0.0018486001,0.003087259,0.0028399983,0.014244075],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0015609999,0.00024778725,0.009650181,0.0040614074,0.00093030836,0.0015371257,0.0011161296,0.021046497,0.07376696,0.0073011136,0.5066823,0.37209934],"study_design_scores_gemma":[0.00094392465,0.0005720584,0.021094121,0.0007886203,0.00042476022,0.0028469567,0.00044007128,0.38652393,0.20614032,0.043158416,0.33642003,0.0006469381],"about_ca_topic_score_codex":0.0030284368,"about_ca_topic_score_gemma":0.0057535353,"teacher_disagreement_score":0.020044364,"about_ca_system_score_codex":0.0008010215,"about_ca_system_score_gemma":0.0021152564,"threshold_uncertainty_score":0.06705505},"labels":[],"label_agreement":null},{"id":"W2784883284","doi":"10.1093/bioinformatics/bty494","title":"Deep convolutional networks for quality assessment of protein folds","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Protein Structure and Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":112,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal; Concordia University","funders":"Natural Sciences and Engineering Research Council of Canada; Canada Research Chairs; Canadian Institute for Advanced Research","keywords":"Computer science; Convolutional neural network; Quality assessment; Quality (philosophy); Artificial intelligence; Reliability engineering; Evaluation methods; Engineering","score_opus":0.014708584649960195,"score_gpt":0.30088106583445734,"score_spread":0.2861724811844971,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2784883284","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.21827483,0.008359189,0.72660786,0.002217853,0.0002968636,0.00024669498,0.00870621,0.024847535,0.01044293],"genre_scores_gemma":[0.8420535,0.0019320407,0.13384885,0.0004754587,0.00011421995,0.0001922726,0.014678753,0.0005411154,0.0061639054],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9992587,0.00012741839,0.00004161979,0.00020221683,0.00026869285,0.000101296384],"domain_scores_gemma":[0.9985409,0.00048029612,0.00023127289,0.00023910472,0.0004110977,0.00009728305],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015202514,0.001411359,0.00077609584,0.0014264922,0.00040141903,0.0010385811,0.0016066801,0.0010729817,0.003664962],"category_scores_gemma":[0.0048766374,0.00041516815,0.0006011792,0.001090177,0.0005822498,0.0012306343,0.0011726483,0.0016871339,0.0014310592],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000749798,0.00027725945,0.010581449,0.0005416286,0.00033768726,0.00016924561,0.000108093256,0.5614789,0.0151272435,0.0147516,0.03182412,0.36405295],"study_design_scores_gemma":[0.0000106651405,0.000024737958,0.00082568097,0.000021084319,0.0000136557355,0.000021208547,0.000005614927,0.9893439,0.0033680748,0.0055082,0.0008489485,0.000008244558],"about_ca_topic_score_codex":0.01013884,"about_ca_topic_score_gemma":0.011398495,"teacher_disagreement_score":0.01013884,"about_ca_system_score_codex":0.0023708472,"about_ca_system_score_gemma":0.0010326672,"threshold_uncertainty_score":0.020159662},"labels":[],"label_agreement":null},{"id":"W2788915007","doi":"10.1093/bioinformatics/bty082","title":"flowLearn: fast and precise identification and quality checking of cell populations in flow cytometry","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Single-cell and spatial transcriptomics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":49,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University; Terry Fox Research Institute; University of British Columbia","funders":"National Institute of Allergy and Infectious Diseases; National Institutes of Health; Wellcome Trust; California Department of Fish and Game; National Institute of General Medical Sciences; Wellcome","keywords":"Identification (biology); Flow cytometry; Computer science; Flow (mathematics); Quality (philosophy); Computational biology; Biology; Mathematics; Molecular biology; Physics","score_opus":0.04216526388673059,"score_gpt":0.292363036621574,"score_spread":0.2501977727348434,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2788915007","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.011571385,0.0017493011,0.73539513,0.0010767309,0.00046265853,0.00064733555,0.015694888,0.23005125,0.003351326],"genre_scores_gemma":[0.07508325,0.0011213268,0.8517733,0.00201283,0.0003533759,0.0024968565,0.038963348,0.024022333,0.0041733524],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.991426,0.0019516654,0.0008056043,0.002609976,0.002806378,0.00040042747],"domain_scores_gemma":[0.98009664,0.01141386,0.001537201,0.0036111793,0.0025147323,0.00082636793],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.014900596,0.0034218633,0.0030417512,0.0046299663,0.002123163,0.004494845,0.0070168707,0.003774906,0.015928784],"category_scores_gemma":[0.046946492,0.0022370545,0.0025988063,0.0022255236,0.0024076563,0.005426099,0.005610547,0.005683018,0.020385],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0017301741,0.0004110225,0.017166002,0.0046109078,0.0006937788,0.00050934404,0.0008560597,0.02460364,0.11528116,0.012303536,0.38913006,0.4327043],"study_design_scores_gemma":[0.00086825463,0.00040060942,0.012518277,0.0011308768,0.000328699,0.0016264978,0.00022150311,0.5250295,0.22892046,0.080077544,0.1481956,0.0006822682],"about_ca_topic_score_codex":0.0034468903,"about_ca_topic_score_gemma":0.0053050215,"teacher_disagreement_score":0.015928784,"about_ca_system_score_codex":0.0023002832,"about_ca_system_score_gemma":0.0041418183,"threshold_uncertainty_score":0.078802824},"labels":[],"label_agreement":null},{"id":"W2789853766","doi":"10.1093/bioinformatics/bty075","title":"chromswitch: a flexible method to detect chromatin state switches","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; Jewish General Hospital","funders":"Natural Sciences and Engineering Research Council of Canada; Canadian Institutes of Health Research; Compute Canada","keywords":"Bioconductor; Epigenomics; Chromatin; Computer science; Benchmarking; Key (lock); Epigenetics; Source code; State (computer science); Data mining; Computational biology; Biology; Gene; DNA methylation; Gene expression; Programming language; Genetics; Operating system","score_opus":0.011509805674919115,"score_gpt":0.2669648109639451,"score_spread":0.255455005289026,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2789853766","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.036942434,0.0014717795,0.5570491,0.00037668715,0.0006283306,0.00048756285,0.09869397,0.29767272,0.0066773314],"genre_scores_gemma":[0.2446364,0.0006980908,0.5501976,0.0011807665,0.00034537475,0.0028301033,0.13429558,0.054609966,0.011206194],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9987532,0.00014804851,0.00007855722,0.00056040194,0.00033824003,0.00012148559],"domain_scores_gemma":[0.9983248,0.000927782,0.00016458632,0.00029342587,0.0001776796,0.00011172902],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0017679377,0.0015680746,0.0012792301,0.0033239948,0.0010847832,0.0020496752,0.0020829171,0.0011244045,0.027060313],"category_scores_gemma":[0.0063427547,0.0011594164,0.0019746828,0.0022259008,0.00063149707,0.0013516382,0.0025304866,0.0016938872,0.0072596944],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0033064305,0.00032309478,0.037722573,0.0031461394,0.0022373484,0.00086608663,0.00071890996,0.030337878,0.09822592,0.013883478,0.39495122,0.414281],"study_design_scores_gemma":[0.0009858328,0.0005545529,0.05454778,0.00031098403,0.0007332717,0.0013857194,0.00027100928,0.47919366,0.14366561,0.06605093,0.25155064,0.00074997527],"about_ca_topic_score_codex":0.0036339264,"about_ca_topic_score_gemma":0.011437791,"teacher_disagreement_score":0.027060313,"about_ca_system_score_codex":0.00080775673,"about_ca_system_score_gemma":0.0016995773,"threshold_uncertainty_score":0.09052575},"labels":[],"label_agreement":null},{"id":"W2790921906","doi":"10.1093/bioinformatics/bty095","title":"Improved genomic island predictions with IslandPath-DIMOB","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Machine Learning in Bioinformatics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":223,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung","keywords":"Computer science; Computational biology; Biology","score_opus":0.004793943520078781,"score_gpt":0.2206221854113326,"score_spread":0.21582824189125382,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2790921906","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.20193516,0.0061156084,0.4204929,0.00087875745,0.0006068297,0.00044536762,0.080512755,0.28012928,0.008883328],"genre_scores_gemma":[0.23494919,0.0012644343,0.54204816,0.00045397982,0.00016285737,0.00036203678,0.20977497,0.0082122935,0.0027720614],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99918634,0.00015868853,0.000059461112,0.0002816817,0.00024851758,0.00006539955],"domain_scores_gemma":[0.9979613,0.0011631502,0.00015705553,0.00021930649,0.00036551795,0.00013371423],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016175898,0.0021913273,0.0012030046,0.0045286724,0.00051932817,0.0016060317,0.0014190078,0.00097656157,0.0046627144],"category_scores_gemma":[0.0058763647,0.00060701685,0.0014076768,0.0021611524,0.00020412743,0.0016111096,0.0016038723,0.0008956212,0.0031310308],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.002716044,0.00055812736,0.067140825,0.0055219927,0.0012187315,0.00094442,0.00051985774,0.06550548,0.06034166,0.0034064974,0.19675723,0.59536916],"study_design_scores_gemma":[0.0003167163,0.00039113796,0.018968258,0.00044448237,0.00034634562,0.0007860551,0.00026221896,0.83087623,0.05137121,0.0073691015,0.088621795,0.0002464635],"about_ca_topic_score_codex":0.003907816,"about_ca_topic_score_gemma":0.0059478628,"teacher_disagreement_score":0.0046627144,"about_ca_system_score_codex":0.00037564768,"about_ca_system_score_gemma":0.0008682085,"threshold_uncertainty_score":0.015598357},"labels":[],"label_agreement":null},{"id":"W2793111149","doi":"10.1093/bioinformatics/bty153","title":"Data mining of effect sizes from PubMed abstracts: a cross-study conceptual replication","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Scientific Computing and Data Management","field":"Decision Sciences","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"Université de Toulouse; Agence Nationale de la Recherche","keywords":"Replication (statistics); Computer science; Data mining; Computational biology; Data science; Statistics; Biology; Mathematics","score_opus":0.23823442753928825,"score_gpt":0.4248347766624087,"score_spread":0.18660034912312046,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2793111149","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.3944324,0.07601158,0.45471704,0.0067286403,0.0036434333,0.03952974,0.013150434,0.0013403371,0.010446439],"genre_scores_gemma":[0.7818496,0.0042031794,0.17491414,0.0021806494,0.0006202118,0.030045282,0.0044952165,0.0005493142,0.0011423596],"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.6753691,0.19272995,0.061562993,0.035649605,0.033034235,0.0016541745],"domain_scores_gemma":[0.192081,0.5797159,0.05180742,0.11043321,0.06459804,0.0013644659],"candidate_categories":["metaresearch"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.31840754,0.0020745846,0.00350246,0.014527215,0.002640569,0.0060480637,0.0053612436,0.0028839246,0.007034388],"category_scores_gemma":[0.68515414,0.001870148,0.010076527,0.010608913,0.0045259907,0.008653442,0.0076664225,0.0026939218,0.001546273],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.007766883,0.0021427472,0.43873817,0.09858943,0.1221998,0.002246535,0.061115816,0.0032763635,0.007587024,0.022165522,0.0097545115,0.22441721],"study_design_scores_gemma":[0.005740423,0.017403884,0.5015044,0.051531184,0.13594653,0.0054284222,0.039059203,0.021361196,0.024287887,0.06814627,0.1281404,0.0014501417],"about_ca_topic_score_codex":0.002798254,"about_ca_topic_score_gemma":0.0028791337,"teacher_disagreement_score":0.68159246,"about_ca_system_score_codex":0.003003432,"about_ca_system_score_gemma":0.005518721,"threshold_uncertainty_score":0.84052485},"labels":[],"label_agreement":null},{"id":"W2793136329","doi":"10.1093/bioinformatics/bty208","title":"MechRNA: prediction of lncRNA mechanisms from RNA–RNA and RNA–protein interactions","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Cancer-related molecular mechanisms research","field":"Biochemistry, Genetics and Molecular Biology","cited_by":111,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"National Institute of General Medical Sciences; Deutsche Forschungsgemeinschaft; Baden-Württemberg Stiftung","keywords":"RNA; Computational biology; Non-coding RNA; Biology; RNA editing; Genetics; Gene","score_opus":0.012646024646255302,"score_gpt":0.25746365466075644,"score_spread":0.24481763001450113,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2793136329","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.07592431,0.0021683383,0.60569066,0.00052036846,0.00018224907,0.0003688619,0.062127408,0.24894442,0.00407343],"genre_scores_gemma":[0.31126648,0.0011688326,0.50768733,0.00034887742,0.00013766403,0.00080757006,0.1676291,0.0067393756,0.0042147934],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99959654,0.000058602225,0.000026378424,0.00019439186,0.000090162466,0.00003390359],"domain_scores_gemma":[0.9994253,0.00029236532,0.00008541218,0.00008274612,0.00006345351,0.000050708957],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007975853,0.002636764,0.0008040438,0.0020959317,0.0006488798,0.0011989977,0.0014982119,0.0011534233,0.0068156025],"category_scores_gemma":[0.0013404378,0.000828063,0.002394208,0.0007792146,0.00045063402,0.0011012743,0.0010682096,0.00093137234,0.0036123577],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0020834056,0.00056475104,0.05814897,0.00824817,0.0015846207,0.0032952933,0.00038696537,0.19533974,0.3502744,0.020232659,0.13739616,0.22244483],"study_design_scores_gemma":[0.00018211095,0.00032441344,0.015225187,0.00012226317,0.00025321415,0.0015314359,0.000086608314,0.82243156,0.094515026,0.018449528,0.04671637,0.00016223239],"about_ca_topic_score_codex":0.0016927167,"about_ca_topic_score_gemma":0.0027290625,"teacher_disagreement_score":0.0068156025,"about_ca_system_score_codex":0.00058091275,"about_ca_system_score_gemma":0.0011343819,"threshold_uncertainty_score":0.022800505},"labels":[],"label_agreement":null},{"id":"W2793622104","doi":"10.1093/bioinformatics/bty136","title":"ddPCRclust: an R package and Shiny app for automated analysis of multiplexed ddPCR data","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Innovative Microfluidic and Catalytic Techniques Innovation","field":"Engineering","cited_by":20,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia; BC Cancer Agency","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Bottleneck; Digital polymerase chain reaction; Computer science; Throughput; Multiplexing; Interface (matter); Operating system; Embedded system; Polymerase chain reaction; Biology","score_opus":0.035184368106834216,"score_gpt":0.2955873917552118,"score_spread":0.2604030236483776,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2793622104","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.003388251,0.0019496225,0.24292827,0.0011588125,0.0007794293,0.00061229995,0.22113456,0.52294815,0.0051006363],"genre_scores_gemma":[0.023401579,0.0014698558,0.4895466,0.0019958084,0.00030496359,0.005430356,0.19385745,0.27476674,0.009226733],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99453086,0.0019390364,0.00045400704,0.001458773,0.0011755114,0.0004419426],"domain_scores_gemma":[0.9884552,0.007653534,0.0010350433,0.001266781,0.001116722,0.000472697],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007983243,0.004986834,0.004894137,0.0044925436,0.0016483652,0.0043418068,0.0061486633,0.0026089202,0.11210147],"category_scores_gemma":[0.031622592,0.0033042498,0.0040908586,0.0036173156,0.0016949076,0.0031155488,0.004968408,0.005053178,0.108020596],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009888157,0.000116480915,0.0035989543,0.0062424433,0.0012303852,0.00069833186,0.0005784112,0.0046113776,0.0067766863,0.008024045,0.92616665,0.04096736],"study_design_scores_gemma":[0.0011978877,0.0002228735,0.0060992376,0.0015670651,0.0006836436,0.0013225103,0.00029326917,0.048197515,0.02028127,0.052084945,0.8674389,0.0006108866],"about_ca_topic_score_codex":0.0028723902,"about_ca_topic_score_gemma":0.0041864146,"teacher_disagreement_score":0.11210147,"about_ca_system_score_codex":0.0011152682,"about_ca_system_score_gemma":0.005221795,"threshold_uncertainty_score":0.37501675},"labels":[],"label_agreement":null},{"id":"W2793784779","doi":"10.1093/bioinformatics/bty046","title":"pymzML v2.0: introducing a highly compressed and seekable gzip format","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Metabolomics and Mass Spectrometry Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":90,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; Jewish General Hospital","funders":"","keywords":"Computer science; Random access; Scripting language; File format; Documentation; Code (set theory); Source code; Scheme (mathematics); Field (mathematics); MIT License; Data file; Data compression; Algorithm; Software; Computational science; Computer engineering; Database; Programming language; Set (abstract data type)","score_opus":0.008613553688783167,"score_gpt":0.22951924981960134,"score_spread":0.22090569613081817,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2793784779","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.006920279,0.00040125192,0.26402053,0.0009296236,0.0004975255,0.00042896825,0.04652307,0.67230296,0.007975727],"genre_scores_gemma":[0.0526028,0.0007220872,0.40284997,0.0017926918,0.00038299308,0.0035456035,0.15771665,0.3648424,0.015544793],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99789965,0.00035309946,0.0002492555,0.00040722004,0.00081362226,0.00027715994],"domain_scores_gemma":[0.99577063,0.0012974933,0.00035232728,0.0011920205,0.0010727554,0.00031473063],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0025866614,0.0022682154,0.0012145552,0.0032080882,0.001384848,0.0027307083,0.004109424,0.0012574124,0.05557166],"category_scores_gemma":[0.015575456,0.0014050531,0.0013416659,0.0036576302,0.0011291368,0.0034210877,0.004717299,0.0036940845,0.054461285],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0016671531,0.00015711809,0.0028120335,0.0015523465,0.00015044589,0.00069508184,0.0006778729,0.004162794,0.03073213,0.014890381,0.78595364,0.15654907],"study_design_scores_gemma":[0.00044409768,0.00017094066,0.004014869,0.0005363113,0.00009371448,0.00084237085,0.00021950345,0.04639741,0.12044982,0.03646672,0.7899707,0.00039355532],"about_ca_topic_score_codex":0.0020867288,"about_ca_topic_score_gemma":0.0016894105,"teacher_disagreement_score":0.05557166,"about_ca_system_score_codex":0.0011149457,"about_ca_system_score_gemma":0.0020044413,"threshold_uncertainty_score":0.1859057},"labels":[],"label_agreement":null},{"id":"W2796130155","doi":"10.1093/bioinformatics/bty242","title":"Accurate prediction of orthologs in the presence of divergence after duplication","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":29,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Ottawa; Université de Sherbrooke","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Gene duplication; Phylogenetic tree; Phylogenetics; Computational biology; Gene; Biology; Divergence (linguistics); Similarity (geometry); Function (biology); Computer science; Evolutionary biology; Genetics; Artificial intelligence","score_opus":0.015191951221622946,"score_gpt":0.2430373068675879,"score_spread":0.22784535564596495,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2796130155","genre_codex":"empirical","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.5102183,0.005144583,0.45500183,0.0017688858,0.0002920885,0.00014098866,0.011059309,0.009483151,0.0068907943],"genre_scores_gemma":[0.88380826,0.00067341945,0.09637156,0.00047465658,0.00012577618,0.0000956144,0.016693993,0.0007641708,0.0009925466],"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9977502,0.00061210507,0.00015393001,0.00077382976,0.0005211121,0.00018879081],"domain_scores_gemma":[0.9908443,0.005740593,0.0009227042,0.0011085066,0.00087500457,0.0005088739],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003979825,0.00085291977,0.0017670365,0.0025244937,0.0013625362,0.0017747089,0.0019487453,0.0019253212,0.0034656264],"category_scores_gemma":[0.023097698,0.00057376677,0.0011492918,0.0026189275,0.0013232676,0.0039677615,0.002875313,0.00232749,0.002495665],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0022082457,0.00047174154,0.37898332,0.0038661489,0.00087302167,0.0036013226,0.0013367699,0.24358855,0.08065751,0.03992687,0.03674225,0.2077442],"study_design_scores_gemma":[0.00019224307,0.000310721,0.052221138,0.00027043166,0.00025944857,0.0027206358,0.00058297155,0.7851653,0.022355327,0.11097073,0.024805548,0.00014552665],"about_ca_topic_score_codex":0.0021290297,"about_ca_topic_score_gemma":0.0026188777,"teacher_disagreement_score":0.003979825,"about_ca_system_score_codex":0.000798857,"about_ca_system_score_gemma":0.0012748896,"threshold_uncertainty_score":0.021047592},"labels":[],"label_agreement":null},{"id":"W2801106948","doi":"10.1093/bioinformatics/bty368","title":"Improved enzyme annotation with EC-specific cutoffs using DETECT v2","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":31,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary; Hospital for Sick Children; University of Toronto","funders":"Natural Sciences and Engineering Research Council of Canada; National Institutes of Health; National Institute of Allergy and Infectious Diseases; Government of Ontario","keywords":"Annotation; Computer science; Enzyme; Artificial intelligence; Chemistry; Biochemistry","score_opus":0.012649891050087679,"score_gpt":0.22722516158106076,"score_spread":0.21457527053097308,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2801106948","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.073482014,0.0016528124,0.42959753,0.0010241612,0.0010335054,0.00031428027,0.055557072,0.4271707,0.0101680085],"genre_scores_gemma":[0.1821842,0.00043200058,0.6028826,0.0010016297,0.00028403674,0.0005768834,0.15490612,0.050173633,0.0075588413],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99370736,0.0010315183,0.0006751816,0.002127329,0.0020467427,0.00041182246],"domain_scores_gemma":[0.9899018,0.0042568725,0.0008476138,0.002318493,0.0021806848,0.0004944219],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0064355927,0.0034945903,0.0019630333,0.0053588846,0.0015450977,0.0050453856,0.0032150305,0.0016963321,0.013206432],"category_scores_gemma":[0.022994878,0.0015126278,0.0021620772,0.0038367023,0.00055620895,0.0027017018,0.0041010138,0.0023876943,0.010724165],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0059427163,0.0008703234,0.06082074,0.0033797717,0.0016002295,0.0009244136,0.00089355744,0.014938037,0.101788625,0.00898929,0.40630725,0.39354497],"study_design_scores_gemma":[0.0016543493,0.0009518475,0.024936378,0.0006396194,0.0008268335,0.0031342697,0.0004626705,0.34468073,0.21696864,0.03391288,0.37096387,0.000867997],"about_ca_topic_score_codex":0.003788505,"about_ca_topic_score_gemma":0.007832454,"teacher_disagreement_score":0.013206432,"about_ca_system_score_codex":0.0013114817,"about_ca_system_score_gemma":0.0021387842,"threshold_uncertainty_score":0.044179857},"labels":[],"label_agreement":null},{"id":"W2801288141","doi":"10.1093/bioinformatics/bty331","title":"MetaBridge: enabling network-based integrative analysis via direct protein interactors of metabolites","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":20,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"Canadian Institutes of Health Research; Killam Trusts; University of British Columbia; Canada Research Chairs","keywords":"Workflow; Computer science; Source code; Computational biology; Complement (music); Systems biology; Metabolomics; Bioinformatics; Biology; Database; Biochemistry; Programming language; Mutant","score_opus":0.008295472400709979,"score_gpt":0.2388291522548754,"score_spread":0.2305336798541654,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2801288141","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.020877067,0.0027702448,0.57568187,0.0009968554,0.0004570893,0.00038571178,0.087903194,0.30404264,0.0068853972],"genre_scores_gemma":[0.10969608,0.0027443538,0.67685264,0.0008091633,0.00031212677,0.0013429961,0.17514335,0.026487537,0.006611736],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9985763,0.0002312339,0.00008620932,0.00055154576,0.00045619256,0.00009859262],"domain_scores_gemma":[0.99782723,0.0009811306,0.00030232884,0.00039500894,0.00027765965,0.00021656908],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0024399038,0.0027730984,0.0016751225,0.0045455787,0.0011764505,0.0034437778,0.0022304505,0.0008386236,0.015018662],"category_scores_gemma":[0.0054184464,0.0013356686,0.0022299986,0.002909709,0.00058197434,0.0028876502,0.003819413,0.00195352,0.008408734],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.002622768,0.00039350093,0.030373365,0.009696775,0.00360857,0.0023153827,0.001994024,0.034836482,0.12877436,0.04094401,0.46266168,0.2817791],"study_design_scores_gemma":[0.00046505537,0.00032469627,0.024390655,0.00081677927,0.0009022038,0.0014864184,0.0004846097,0.3481965,0.07794247,0.082522705,0.4620108,0.00045709018],"about_ca_topic_score_codex":0.0033222006,"about_ca_topic_score_gemma":0.0067979214,"teacher_disagreement_score":0.015018662,"about_ca_system_score_codex":0.0006395768,"about_ca_system_score_gemma":0.0017936409,"threshold_uncertainty_score":0.050242484},"labels":[],"label_agreement":null},{"id":"W2801332475","doi":"10.1093/bioinformatics/bty363","title":"NGScloud: RNA-seq analysis of non-model species using cloud computing","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Ministerio de Economía y Competitividad; Ministry of Agriculture - Saskatchewan","keywords":"Cloud computing; RNA-Seq; Computer science; Computational biology; Software; Data science; Biology; Genetics; Operating system; Gene; Transcriptome","score_opus":0.025178827820963704,"score_gpt":0.26661059269504833,"score_spread":0.24143176487408463,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2801332475","genre_codex":"software","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.031032218,0.001022473,0.325182,0.00065368885,0.00076052145,0.0012216632,0.098254375,0.5206228,0.021250255],"genre_scores_gemma":[0.09627979,0.00073588046,0.50978136,0.0012571631,0.00017325133,0.00465191,0.21634431,0.15261011,0.018166203],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99862385,0.00015939982,0.00008852699,0.0004685919,0.00047503228,0.00018459269],"domain_scores_gemma":[0.99857783,0.0005108212,0.00010815419,0.00038161984,0.00023594002,0.00018558606],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0020806042,0.0016105345,0.0012446482,0.0014361126,0.0021709164,0.0017794839,0.0040660906,0.0009858876,0.032081824],"category_scores_gemma":[0.0032113867,0.0011810788,0.00166037,0.0016167399,0.0009491741,0.001676165,0.0026030876,0.0022226104,0.020949574],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0027854086,0.0002960527,0.008538352,0.0034009393,0.0004827271,0.00092653994,0.0011698988,0.009314521,0.14403544,0.008133852,0.71127576,0.109640464],"study_design_scores_gemma":[0.00086944114,0.00026528534,0.019733477,0.0005418969,0.00026190103,0.001085251,0.0003144067,0.16790356,0.23452666,0.017210538,0.5565653,0.0007223169],"about_ca_topic_score_codex":0.008207675,"about_ca_topic_score_gemma":0.010448389,"teacher_disagreement_score":0.032081824,"about_ca_system_score_codex":0.0013417745,"about_ca_system_score_gemma":0.0020757231,"threshold_uncertainty_score":0.10732436},"labels":[],"label_agreement":null},{"id":"W2803639923","doi":"10.1093/bioinformatics/bty385","title":"MRMAssayDB: an integrated resource for validated targeted proteomics assays","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Advanced Proteomics Techniques and Applications","field":"Chemistry","cited_by":25,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; Jewish General Hospital; Genome British Columbia; University of Victoria","funders":"Fondation De Famille Alvin Segal; Leids Universitair Medisch Centrum; Genome British Columbia; McGill University; University of Victoria; Jewish General Hospital; Leading Edge Endowment Fund; Genome Canada","keywords":"Proteomics; UniProt; KEGG; Computational biology; Quantitative proteomics; Computer science; Resource (disambiguation); Bioinformatics; Biology; Gene ontology; Gene; Biochemistry; Gene expression","score_opus":0.021240597739801293,"score_gpt":0.2857756677972513,"score_spread":0.26453507005745,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2803639923","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0067506176,0.0056533525,0.17166099,0.0010847357,0.0003925866,0.00080999685,0.33160865,0.46476436,0.017274698],"genre_scores_gemma":[0.026713973,0.003948185,0.19988228,0.0016233162,0.0002899045,0.0016859644,0.7186434,0.039028324,0.008184673],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99529546,0.0007150902,0.00078677444,0.0010132013,0.0019006706,0.00028880493],"domain_scores_gemma":[0.9927288,0.002064043,0.0012633593,0.001657294,0.0016709439,0.00061557064],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0068098824,0.004840292,0.0037466874,0.00863683,0.0013790467,0.00737698,0.0069614053,0.002363089,0.037644774],"category_scores_gemma":[0.00967742,0.0024530792,0.0018892734,0.0062739085,0.0007808871,0.0053496812,0.0063503236,0.0028532979,0.056527264],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00400626,0.0004656806,0.0062522464,0.007849227,0.0009913337,0.0012911354,0.00041648498,0.003757533,0.0631244,0.011246078,0.767262,0.1333375],"study_design_scores_gemma":[0.00094232004,0.0003149092,0.007892588,0.0011618236,0.0004452432,0.0018663618,0.00018630139,0.033372886,0.12286408,0.018408075,0.8119812,0.0005641363],"about_ca_topic_score_codex":0.0027084025,"about_ca_topic_score_gemma":0.0020270573,"teacher_disagreement_score":0.037644774,"about_ca_system_score_codex":0.0016983779,"about_ca_system_score_gemma":0.0034138497,"threshold_uncertainty_score":0.1259343},"labels":[],"label_agreement":null},{"id":"W2806701851","doi":"10.1093/bioinformatics/bty430","title":"GateFinder: projection-based gating strategy optimization for flow and mass cytometry","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Single-cell and spatial transcriptomics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":26,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Terry Fox Research Institute","funders":"National Institute of Allergy and Infectious Diseases; National Institute of General Medical Sciences; U.S. Food and Drug Administration; Cure Starts Now Foundation; National Cancer Institute; Entertainment Industry Foundation; International Society for Advancement of Cytometry; National Institutes of Health; Ovarian Cancer Research Fund; Terry Fox Foundation; Canadian Institutes of Health Research; Damon Runyon Cancer Research Foundation; Stanford Bio-X; Hamilton Health Sciences Foundation; European Commission; Stanford University; March of Dimes Foundation; California Institute for Regenerative Medicine","keywords":"Bioconductor; Computer science; Mass cytometry; R package; Projection (relational algebra); Simple (philosophy); Polygon (computer graphics); Algorithm; Data mining; Computational biology; Theoretical computer science; Biology; Computational science","score_opus":0.019318678120541957,"score_gpt":0.24989795749647944,"score_spread":0.23057927937593747,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2806701851","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.005964116,0.00025591246,0.9687118,0.00023810368,0.00006774457,0.00015476631,0.0009909337,0.021670515,0.0019461102],"genre_scores_gemma":[0.050819706,0.000261805,0.93907,0.00023242777,0.000031684616,0.00088206044,0.0019149287,0.004701791,0.0020856012],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9989436,0.00027579215,0.000046523197,0.00021773283,0.00040775703,0.00010859137],"domain_scores_gemma":[0.9985461,0.00090519415,0.00009266129,0.0001196346,0.00024323155,0.00009326095],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0028715804,0.0023082208,0.0012091894,0.0014614662,0.0008881637,0.0020454992,0.0031966753,0.0015095313,0.016215684],"category_scores_gemma":[0.0070775445,0.00077179505,0.0010718732,0.0015605926,0.00082245184,0.0014094791,0.0022106732,0.0025252046,0.0056536216],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010460778,0.0003044984,0.0043700533,0.00090299005,0.00025343453,0.00031660352,0.00034878304,0.22308718,0.050562255,0.043923065,0.09711048,0.5777746],"study_design_scores_gemma":[0.00011729623,0.00006661708,0.00045638013,0.000038491784,0.00002164141,0.00010971021,0.000030165507,0.93450934,0.022507565,0.020029854,0.02206065,0.00005231184],"about_ca_topic_score_codex":0.0038529434,"about_ca_topic_score_gemma":0.004955664,"teacher_disagreement_score":0.016215684,"about_ca_system_score_codex":0.0014320469,"about_ca_system_score_gemma":0.0034949488,"threshold_uncertainty_score":0.054246843},"labels":[],"label_agreement":null},{"id":"W2807635396","doi":"10.1093/bioinformatics/bty420","title":"Knotty: efficient and accurate prediction of complex RNA pseudoknot structures","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"RNA and protein synthesis mechanisms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":62,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Pseudoknot; Benchmark (surveying); Computer science; Minification; Algorithm; Space (punctuation); RNA; Biology; Genetics","score_opus":0.02181555103800082,"score_gpt":0.2504419687821222,"score_spread":0.2286264177441214,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2807635396","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.23208037,0.0013435514,0.6701355,0.00041442076,0.00037277085,0.00029828522,0.0031344758,0.08669356,0.0055270195],"genre_scores_gemma":[0.4070214,0.00036676685,0.5751059,0.00017057202,0.00013632182,0.00017367733,0.01008962,0.0035703292,0.0033653837],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99937004,0.00007962491,0.000033275126,0.00017856892,0.0002560652,0.000082573475],"domain_scores_gemma":[0.99793917,0.0008677159,0.000250701,0.00035665423,0.00038887252,0.00019686946],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00085318304,0.0013219084,0.00094496197,0.0014269033,0.00064370706,0.00166173,0.0022768357,0.0015238654,0.0050615952],"category_scores_gemma":[0.006164234,0.0006111696,0.00077260047,0.0009871742,0.0007013882,0.0018214926,0.0014983553,0.0014285627,0.002266727],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0018815257,0.00030785523,0.020477226,0.0013511705,0.00026192723,0.0007799258,0.00041607127,0.4167643,0.0666864,0.012476901,0.05479523,0.42380148],"study_design_scores_gemma":[0.000060218725,0.0000983497,0.0014428718,0.00002118536,0.000015861593,0.00018440676,0.00003088858,0.9796779,0.011402186,0.0038354397,0.003187377,0.000043152522],"about_ca_topic_score_codex":0.003814813,"about_ca_topic_score_gemma":0.0047950074,"teacher_disagreement_score":0.0050615952,"about_ca_system_score_codex":0.0006445211,"about_ca_system_score_gemma":0.0015790323,"threshold_uncertainty_score":0.016932726},"labels":[],"label_agreement":null},{"id":"W2807745747","doi":"10.1093/bioinformatics/bty234","title":"aliFreeFold: an alignment-free approach to predict secondary structure from homologous RNA sequences","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"RNA and protein synthesis mechanisms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Sherbrooke","funders":"Natural Sciences and Engineering Research Council of Canada; Canada Research Chairs; Université de Sherbrooke","keywords":"RNA; Computer science; Weighting; Computational biology; Sequence (biology); Sequence alignment; Set (abstract data type); Nucleic acid secondary structure; Protein secondary structure; Structural alignment; Homologous chromosome; Nucleic acid structure; Multiple sequence alignment; Algorithm; Pseudoknot; Protein superfamily; Theoretical computer science; Biology; Genetics; Peptide sequence; Physics; Gene","score_opus":0.011423693410995715,"score_gpt":0.22554404988826676,"score_spread":0.21412035647727104,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2807745747","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.027102018,0.0010895245,0.839725,0.0002537777,0.00026235735,0.00033670713,0.0056975633,0.1226841,0.0028489418],"genre_scores_gemma":[0.11234916,0.00051020744,0.84838325,0.00037955196,0.0000882167,0.00069660717,0.018650848,0.015615554,0.0033266535],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99936706,0.00012637097,0.000040011306,0.00020502294,0.00020148701,0.000060024115],"domain_scores_gemma":[0.9990715,0.0004374172,0.00011178412,0.00015562656,0.00013719752,0.00008640067],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015012888,0.0022470956,0.0017049839,0.0019289239,0.0011454026,0.001283904,0.0028076689,0.0016780467,0.011933466],"category_scores_gemma":[0.0040935925,0.001116639,0.001961922,0.00094959297,0.0006004542,0.0019530794,0.0013823707,0.0021217887,0.0056639267],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0020722034,0.0006001602,0.011811998,0.0036838786,0.0010531675,0.0010608743,0.0005578659,0.22285546,0.1366001,0.01724753,0.13649844,0.46595824],"study_design_scores_gemma":[0.00025246458,0.0002866066,0.0019581805,0.00014085097,0.0001244303,0.000680991,0.00007602493,0.8981546,0.047506865,0.016936753,0.03374983,0.00013223462],"about_ca_topic_score_codex":0.0021419483,"about_ca_topic_score_gemma":0.004019583,"teacher_disagreement_score":0.011933466,"about_ca_system_score_codex":0.0008172702,"about_ca_system_score_gemma":0.0014830232,"threshold_uncertainty_score":0.039921463},"labels":[],"label_agreement":null},{"id":"W2808973977","doi":"10.1093/bioinformatics/bty466","title":"iMetaLab 1.0: a web platform for metaproteomics data analysis","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gut microbiota and health","field":"Biochemistry, Genetics and Molecular Biology","cited_by":50,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canadian Institute for Advanced Research; University of Ottawa","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Metaproteomics; Computer science; Pipeline (software); Task (project management); Data mining; Data science; Database; World Wide Web; Metagenomics; Engineering; Biology; Systems engineering","score_opus":0.041602311727796375,"score_gpt":0.3111446754283295,"score_spread":0.26954236370053314,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2808973977","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0067640822,0.001335864,0.22791983,0.0010108235,0.00045693383,0.00097571197,0.104492754,0.651241,0.005803082],"genre_scores_gemma":[0.04532464,0.0015466993,0.501721,0.0017141452,0.00033206824,0.0050698193,0.28519398,0.15153587,0.007561804],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99669504,0.00073304045,0.00039655823,0.00069170713,0.001108162,0.00037540882],"domain_scores_gemma":[0.9939507,0.0021948752,0.00084612705,0.0009909129,0.0012147033,0.00080260425],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008928393,0.004493981,0.0027278764,0.006828277,0.001526896,0.0056871204,0.0056045405,0.0015991048,0.04257342],"category_scores_gemma":[0.017406791,0.0026605786,0.002723361,0.0051378296,0.0012349819,0.0050736256,0.008975625,0.00544657,0.040805917],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0071492614,0.0009037758,0.011452156,0.008409425,0.003132797,0.002183047,0.0023762947,0.0097597465,0.0726248,0.015988171,0.6835537,0.18246688],"study_design_scores_gemma":[0.0014882401,0.00043118125,0.015582513,0.001854995,0.0008982287,0.0016790989,0.0006233757,0.13732149,0.09310615,0.059682343,0.6861906,0.0011417487],"about_ca_topic_score_codex":0.003033396,"about_ca_topic_score_gemma":0.0024562073,"teacher_disagreement_score":0.04257342,"about_ca_system_score_codex":0.0012469663,"about_ca_system_score_gemma":0.003854111,"threshold_uncertainty_score":0.14242226},"labels":[],"label_agreement":null},{"id":"W2809066811","doi":"10.1093/bioinformatics/bty482","title":"Metaxa2 Database Builder: enabling taxonomic identification from metagenomic or metabarcoding data using any genetic marker","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Environmental DNA in Biodiversity Studies","field":"Environmental Science","cited_by":48,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canadian Food Inspection Agency","funders":"Vetenskapsrådet; Svenska Forskningsrådet Formas; Project Apis m.","keywords":"Metagenomics; Identification (biology); Biology; Amplicon; Taxonomic rank; DNA barcoding; Software; Computational biology; Genetic marker; Computer science; Data mining; Evolutionary biology; Gene; Genetics; Ecology; Taxon; Polymerase chain reaction","score_opus":0.0907229681115546,"score_gpt":0.2702370529449247,"score_spread":0.1795140848333701,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2809066811","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.024177186,0.0017074227,0.44696933,0.0009370147,0.00063994655,0.00081532006,0.073666185,0.44483745,0.0062500834],"genre_scores_gemma":[0.07555152,0.0016427927,0.6814625,0.0010660638,0.0002992524,0.0025433518,0.17062993,0.060458407,0.0063461154],"study_design_codex":"not_applicable","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99668056,0.00057868764,0.00052574713,0.0009361709,0.0010454903,0.00023341393],"domain_scores_gemma":[0.99214935,0.0026795247,0.0013475006,0.0018631205,0.0013316208,0.0006288979],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.009385317,0.0028722195,0.00147839,0.0037110555,0.0014441863,0.004360156,0.005012067,0.001609705,0.015540408],"category_scores_gemma":[0.019886434,0.0027031782,0.00220155,0.0030191634,0.0008522657,0.0043515596,0.00494776,0.0031484612,0.014348853],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0070366496,0.0007142279,0.04082838,0.0075554973,0.0020770365,0.0016301604,0.0031497348,0.008615921,0.16714826,0.015558184,0.42317277,0.32251316],"study_design_scores_gemma":[0.001109841,0.0007508185,0.029547643,0.0016515708,0.0008571175,0.0023443163,0.0006991361,0.06898145,0.23578669,0.022154354,0.6351811,0.00093596324],"about_ca_topic_score_codex":0.0021298332,"about_ca_topic_score_gemma":0.002253784,"teacher_disagreement_score":0.015540408,"about_ca_system_score_codex":0.0008024953,"about_ca_system_score_gemma":0.0023458796,"threshold_uncertainty_score":0.051987886},"labels":[],"label_agreement":null},{"id":"W2809263774","doi":"10.1093/bioinformatics/bty475","title":"HUME: large-scale detection of causal genetic factors of adverse drug reactions","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Pharmacovigilance and Adverse Drug Reactions","field":"Pharmacology, Toxicology and Pharmaceutics","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia; Child and Family Research Institute; Children's & Women's Health Centre of British Columbia; Simon Fraser University","funders":"Institute of Cancer Research; Canada Foundation for Innovation; Genome Canada","keywords":"Confounding; Drug reaction; Drug; Data mining; Adverse drug reaction; Computer science; Adverse effect; Medicine; Bioinformatics; Intensive care medicine; Internal medicine; Pharmacology; Biology","score_opus":0.05633522496170181,"score_gpt":0.38342296736908704,"score_spread":0.32708774240738525,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2809263774","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.1422997,0.0015902757,0.8287697,0.0026515073,0.00014833681,0.00134408,0.01296107,0.008572947,0.0016623723],"genre_scores_gemma":[0.43687496,0.00037631,0.5481468,0.00050753815,0.00016889903,0.00085483294,0.011437036,0.00019044656,0.0014431684],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9972652,0.001538626,0.00016067427,0.00062486104,0.000327925,0.00008268839],"domain_scores_gemma":[0.9662929,0.02966309,0.0014091491,0.0014051213,0.0007438974,0.00048580486],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008814045,0.0010246636,0.0012082424,0.0031830932,0.0006688668,0.0012123632,0.001727929,0.0014705816,0.0058196066],"category_scores_gemma":[0.034925807,0.0005991408,0.0015312772,0.0011879614,0.00058699585,0.001094994,0.0018096933,0.0011630822,0.0006679116],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0022945018,0.0009451678,0.31919795,0.0020028502,0.0034583155,0.0037125212,0.0006085846,0.22641706,0.0071036094,0.013306783,0.031821888,0.3891309],"study_design_scores_gemma":[0.0005288216,0.00031674467,0.027259672,0.0001166426,0.000454037,0.0012237005,0.00011601512,0.93003947,0.0030036427,0.031379506,0.00550571,0.00005595074],"about_ca_topic_score_codex":0.0022627756,"about_ca_topic_score_gemma":0.005302159,"teacher_disagreement_score":0.008814045,"about_ca_system_score_codex":0.00048428023,"about_ca_system_score_gemma":0.0014894032,"threshold_uncertainty_score":0.046613693},"labels":[],"label_agreement":null},{"id":"W2809895957","doi":"10.1093/bioinformatics/bty528","title":"MetaboAnalystR: an R package for flexible and reproducible analysis of metabolomics data","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Metabolomics and Mass Spectrometry Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":802,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"Natural Sciences and Engineering Research Council of Canada; Canada Research Chairs; McGill University; Chesapeake Research Consortium","keywords":"Computer science; Workflow; Flexibility (engineering); Interface (matter); Web server; Web application; World Wide Web; Database; Operating system; The Internet","score_opus":0.049798772644403,"score_gpt":0.32536535111417814,"score_spread":0.2755665784697751,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2809895957","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0038888198,0.0018031029,0.5410796,0.0012311446,0.0005899334,0.00059353566,0.13026878,0.31643403,0.0041111647],"genre_scores_gemma":[0.030014118,0.0017191477,0.68851095,0.0010668524,0.0003477087,0.0030372962,0.11946453,0.15125215,0.0045872126],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99622154,0.0014859425,0.00043048072,0.0008142542,0.00084941916,0.00019838376],"domain_scores_gemma":[0.98631394,0.007062486,0.0018976128,0.002236909,0.002036995,0.00045200097],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007509256,0.0034298485,0.0030254552,0.004037924,0.0008225412,0.0040576146,0.0047614346,0.0012831136,0.045148943],"category_scores_gemma":[0.0339763,0.0017570399,0.0031510151,0.0037666555,0.0010211932,0.0023906708,0.003547555,0.0040812595,0.040368527],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013252005,0.0000861774,0.0073845475,0.008780267,0.0028482107,0.00089047,0.0005470171,0.011459958,0.014770238,0.017393969,0.8248138,0.10970016],"study_design_scores_gemma":[0.0007574585,0.00028344078,0.0112366695,0.0012857816,0.0012466343,0.0017956787,0.00014324013,0.07466522,0.032743353,0.052967608,0.8223073,0.0005677286],"about_ca_topic_score_codex":0.0029627809,"about_ca_topic_score_gemma":0.0030145966,"teacher_disagreement_score":0.045148943,"about_ca_system_score_codex":0.0007413148,"about_ca_system_score_gemma":0.0046897796,"threshold_uncertainty_score":0.15103823},"labels":[],"label_agreement":null},{"id":"W2809990446","doi":"10.1093/bioinformatics/bty537","title":"Multiomics modeling of the immunome, transcriptome, microbiome, proteome and metabolome adaptations during human pregnancy","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Pregnancy and preeclampsia studies","field":"Medicine","cited_by":197,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Polytechnique Montréal; Group for Research in Decision Analysis; Transport Canada","funders":"U.S. National Library of Medicine; National Institute of General Medical Sciences; National Institute of Diabetes and Digestive and Kidney Diseases; National Heart, Lung, and Blood Institute; Canadian Institutes of Health Research; National Institutes of Health; NOMIS Stiftung; Wellcome Trust; Ovarian Cancer Research Fund; National Institute of Allergy and Infectious Diseases; Bill and Melinda Gates Foundation","keywords":"Metabolome; Transcriptome; Proteome; Biology; Computational biology; Microbiome; Pregnancy; Bioinformatics; Physiology; Metabolomics; Genetics; Gene; Gene expression","score_opus":0.029459298037730824,"score_gpt":0.26237827660429675,"score_spread":0.23291897856656593,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2809990446","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.79502594,0.0013229798,0.18948287,0.0018753548,0.00007789054,0.00010231686,0.009770548,0.00064898975,0.0016931564],"genre_scores_gemma":[0.95747894,0.00045891159,0.033999704,0.00022267461,0.000049541683,0.00021129932,0.006319812,0.000071771574,0.0011874128],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9996729,0.00016038226,0.000014426916,0.00009776894,0.000023966966,0.000030557287],"domain_scores_gemma":[0.99901307,0.00069759303,0.00009715156,0.00007589182,0.00006429532,0.00005196936],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001956083,0.00065951154,0.00054192793,0.00051340077,0.0003389618,0.00083378074,0.000858813,0.0007458779,0.0011179944],"category_scores_gemma":[0.0036360521,0.0003529544,0.0014414713,0.00069978053,0.0003900035,0.00047907504,0.0007698301,0.00082306727,0.00020931155],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00071492343,0.00012955564,0.06803073,0.00015332382,0.00043312827,0.0003475951,0.00017572728,0.8994982,0.0047277873,0.0024325934,0.002476875,0.020879546],"study_design_scores_gemma":[0.000016007909,0.000039919087,0.012813738,0.000015763157,0.000044016902,0.000068401154,0.000033428223,0.9824326,0.0007278908,0.0033541494,0.0004391768,0.000014788908],"about_ca_topic_score_codex":0.010631707,"about_ca_topic_score_gemma":0.00638029,"teacher_disagreement_score":0.010631707,"about_ca_system_score_codex":0.00058321236,"about_ca_system_score_gemma":0.0007729181,"threshold_uncertainty_score":0.021139681},"labels":[],"label_agreement":null},{"id":"W2810425915","doi":"10.1093/bioinformatics/bty544","title":"lordFAST: sensitive and Fast Alignment Search Tool for LOng noisy Read sequencing Data","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":46,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia; Simon Fraser University","funders":"National Institute of General Medical Sciences; National Institutes of Health; National Science Foundation","keywords":"Nanopore sequencing; Computer science; Reference genome; Genomics; DNA sequencing; Throughput; Genome; Data mining; Computational biology; Biology; Wireless; Genetics; Gene; Telecommunications","score_opus":0.0409620581989204,"score_gpt":0.28350774083588093,"score_spread":0.24254568263696052,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2810425915","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.019424953,0.0028905126,0.5374782,0.00086181384,0.0011134406,0.00080399436,0.06158083,0.3718878,0.0039585317],"genre_scores_gemma":[0.04460206,0.0006708212,0.8179567,0.0009097562,0.00012568335,0.0013199374,0.10210122,0.027689705,0.004624104],"study_design_codex":"not_applicable","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99608755,0.0009255192,0.00048718718,0.0011256956,0.0010431379,0.00033085485],"domain_scores_gemma":[0.996788,0.0017528256,0.00037579905,0.00038071902,0.00047473703,0.00022793125],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005356895,0.0032783411,0.0022707647,0.0025284763,0.0022912943,0.0023533683,0.0048492965,0.002658474,0.025762027],"category_scores_gemma":[0.015241334,0.0021862264,0.0024889507,0.0030503164,0.0009019248,0.0035686756,0.0031970902,0.00381106,0.023576494],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0061437944,0.00038017356,0.0065542455,0.008176181,0.001541724,0.002529674,0.0012281531,0.017744191,0.14780737,0.0103469845,0.54584354,0.2517039],"study_design_scores_gemma":[0.0024460007,0.0011759416,0.010446641,0.0012154124,0.00051327393,0.0046022674,0.0007660409,0.36166957,0.19339602,0.0403044,0.38235447,0.0011100004],"about_ca_topic_score_codex":0.0025971185,"about_ca_topic_score_gemma":0.005613445,"teacher_disagreement_score":0.025762027,"about_ca_system_score_codex":0.0010260595,"about_ca_system_score_gemma":0.0031073862,"threshold_uncertainty_score":0.086182535},"labels":[],"label_agreement":null},{"id":"W2810570362","doi":"10.1093/bioinformatics/bty586","title":"Fast characterization of segmental duplications in genome assemblies","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genome Rearrangement Algorithms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":107,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"National Institute of General Medical Sciences; National Institutes of Health","keywords":"Genome; Structural variation; Computer science; Sequence assembly; Computational biology; Segmental duplication; Gene duplication; Jaccard index; Biology; Genetics; Artificial intelligence; Pattern recognition (psychology); Gene","score_opus":0.010153928465806603,"score_gpt":0.23674114807972357,"score_spread":0.22658721961391698,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2810570362","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.22556297,0.0017000336,0.7333973,0.00022564363,0.00012559355,0.00026288646,0.0044405013,0.030882992,0.0034019696],"genre_scores_gemma":[0.3197565,0.00043613857,0.6655418,0.000101574966,0.00004500455,0.0001373831,0.010815787,0.0017299886,0.0014357801],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9984819,0.00021765585,0.00011093971,0.00042234594,0.00063970353,0.0001275336],"domain_scores_gemma":[0.9959656,0.0013968223,0.0005739781,0.0004895352,0.0013751172,0.00019897976],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0021361124,0.0011035716,0.0008439557,0.0033219638,0.00069336983,0.0014160534,0.0011120293,0.0007670383,0.0023242636],"category_scores_gemma":[0.007840788,0.00043197355,0.00081107987,0.0018367278,0.00038082473,0.0015810152,0.0010776817,0.00089894823,0.001527762],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00088508934,0.00020379605,0.051784266,0.0013472558,0.00042720174,0.0007789863,0.00087668526,0.04545311,0.30840045,0.0074365893,0.016449006,0.5659575],"study_design_scores_gemma":[0.00006132034,0.00038693618,0.04536889,0.00015139242,0.00014981341,0.0014088219,0.000387303,0.6456215,0.27399686,0.007135583,0.025158407,0.00017317117],"about_ca_topic_score_codex":0.0027934466,"about_ca_topic_score_gemma":0.0038700362,"teacher_disagreement_score":0.0033219638,"about_ca_system_score_codex":0.0005919633,"about_ca_system_score_gemma":0.0008254691,"threshold_uncertainty_score":0.0112969875},"labels":[],"label_agreement":null},{"id":"W2810954907","doi":"10.1093/bioinformatics/bty538","title":"Differential coexpression in human tissues and the confounding effect of mean expression levels","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":32,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canada's Michael Smith Genome Sciences Centre; University of British Columbia","funders":"National Institute of Mental Health; National Institute of General Medical Sciences; Natural Sciences and Engineering Research Council of Canada","keywords":"Confounding; Biology; Differential (mechanical device); Gene expression; Gene; Expression (computer science); Differential diagnosis; Gene expression profiling; Computational biology; Genetics; Internal medicine; Computer science; Pathology; Medicine","score_opus":0.011072941190757163,"score_gpt":0.2735351256649175,"score_spread":0.26246218447416036,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2810954907","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8512421,0.006463814,0.1221013,0.0005908407,0.00009236682,0.00010851352,0.0149892345,0.0006578117,0.0037540924],"genre_scores_gemma":[0.9689349,0.0007911662,0.020609189,0.00018081412,0.000045477373,0.00012322176,0.008662235,0.00010667576,0.00054638594],"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9965654,0.0007728703,0.00029193074,0.0017322785,0.00044552473,0.00019212504],"domain_scores_gemma":[0.98952776,0.00686047,0.0012926498,0.0015612787,0.00059424026,0.0001635206],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0031537875,0.0004924064,0.0006982606,0.0013559911,0.0006891188,0.0012432382,0.0006837307,0.0006046753,0.0021462822],"category_scores_gemma":[0.012799691,0.00030016157,0.00073063356,0.0035333384,0.0009977482,0.0009230882,0.0011015991,0.00075924466,0.00059563067],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014620055,0.00018187918,0.6470863,0.0033646352,0.0019019746,0.0014135004,0.00095349055,0.03655842,0.18577287,0.009976721,0.0052099745,0.10611829],"study_design_scores_gemma":[0.0000873702,0.0003372142,0.72891194,0.0002681248,0.0013271003,0.004650696,0.00039121223,0.08478704,0.11461916,0.042643342,0.02186173,0.00011511285],"about_ca_topic_score_codex":0.0016725431,"about_ca_topic_score_gemma":0.002386175,"teacher_disagreement_score":0.0031537875,"about_ca_system_score_codex":0.0006960048,"about_ca_system_score_gemma":0.00055927417,"threshold_uncertainty_score":0.016678989},"labels":[],"label_agreement":null},{"id":"W2884325264","doi":"10.1093/bioinformatics/bty621","title":"MAVIS: merging, annotation, validation, and illustration of structural variants","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Rare Diseases","field":"Biochemistry, Genetics and Molecular Biology","cited_by":33,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University; University of British Columbia; Canada's Michael Smith Genome Sciences Centre","funders":"Genome British Columbia; BC Cancer Foundation; Genome Canada","keywords":"Annotation; Context (archaeology); Computer science; Key (lock); Computational biology; Process (computing); Genome; Data mining; Data science; Artificial intelligence; Biology; Genetics; Programming language; Gene","score_opus":0.009028182947586842,"score_gpt":0.24153045101335732,"score_spread":0.23250226806577048,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2884325264","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.029545216,0.0010490421,0.65355635,0.001111151,0.00070083456,0.00039272525,0.05093163,0.2534624,0.009250619],"genre_scores_gemma":[0.09214147,0.00049227173,0.7766775,0.0005397017,0.00017910941,0.00074748945,0.09359702,0.03176343,0.0038620562],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99705184,0.00057871494,0.00031636504,0.0011294859,0.00075244793,0.00017110452],"domain_scores_gemma":[0.9945209,0.0026972815,0.0005685736,0.0010846833,0.0008263858,0.0003022002],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0054053487,0.0019731985,0.001125402,0.006093763,0.00201707,0.0036908798,0.004086189,0.0019131901,0.030153617],"category_scores_gemma":[0.01775135,0.0012709717,0.00236975,0.004011072,0.0009254302,0.0019481481,0.004037036,0.0020835954,0.0127755655],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0034946413,0.0003387465,0.026600225,0.003598976,0.00113734,0.0026942552,0.0026832165,0.022113966,0.05679732,0.0416533,0.43483394,0.40405405],"study_design_scores_gemma":[0.000535323,0.00048033037,0.013678923,0.0009568954,0.00054957694,0.0029460965,0.0006723149,0.25080144,0.11062204,0.07381814,0.5444763,0.0004627275],"about_ca_topic_score_codex":0.0036077676,"about_ca_topic_score_gemma":0.004804278,"teacher_disagreement_score":0.030153617,"about_ca_system_score_codex":0.0008197262,"about_ca_system_score_gemma":0.002442763,"threshold_uncertainty_score":0.10087383},"labels":[],"label_agreement":null},{"id":"W2884335298","doi":"10.1093/bioinformatics/bty655","title":"The epiGenomic Efficient Correlator (epiGeEC) tool allows fast comparison of user datasets with thousands of public epigenomic datasets","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Epigenetics and DNA Methylation","field":"Biochemistry, Genetics and Molecular Biology","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; McGill University and Génome Québec Innovation Centre; Université de Sherbrooke","funders":"Natural Sciences and Engineering Research Council of Canada; Compute Canada; Canarie","keywords":"Epigenomics; Computer science; Epigenome; Source code; Variety (cybernetics); R package; Data mining; Biology; Artificial intelligence; DNA methylation; Computational science; Genetics","score_opus":0.017196482251295987,"score_gpt":0.2739516866853645,"score_spread":0.2567552044340685,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2884335298","genre_codex":"software","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.01715452,0.00089333544,0.3205574,0.0005712867,0.00029784144,0.0004213991,0.22501117,0.4288727,0.0062202956],"genre_scores_gemma":[0.07540864,0.0005114238,0.5015629,0.0010105083,0.0001549653,0.0028593605,0.35768324,0.055526484,0.005282429],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9978605,0.0004700946,0.00016219809,0.0007196732,0.00059296,0.0001945236],"domain_scores_gemma":[0.9952793,0.002342891,0.00037814432,0.0011801503,0.00054675783,0.00027279896],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005467778,0.001907287,0.0017872875,0.0038569134,0.0011586153,0.0019582121,0.0027546736,0.0012782376,0.03130593],"category_scores_gemma":[0.014350291,0.0015419605,0.0015929177,0.0035241786,0.0006604191,0.0021138075,0.0042607444,0.002310439,0.013380492],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0025167642,0.00020324474,0.027020857,0.0033286903,0.0020626385,0.0008806922,0.0008274545,0.007215879,0.058819134,0.011172916,0.75685585,0.12909591],"study_design_scores_gemma":[0.0012966681,0.00031816322,0.051270466,0.00039533683,0.0006951382,0.0016321781,0.00036752614,0.11481766,0.17141344,0.034322806,0.6228043,0.00066638726],"about_ca_topic_score_codex":0.0037431812,"about_ca_topic_score_gemma":0.008370153,"teacher_disagreement_score":0.03130593,"about_ca_system_score_codex":0.00088706595,"about_ca_system_score_gemma":0.002306539,"threshold_uncertainty_score":0.10472882},"labels":[],"label_agreement":null},{"id":"W2889339419","doi":"10.1093/bioinformatics/bty780","title":"Co-registration and analysis of multiple imaging mass spectrometry datasets targeting different analytes","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Mass Spectrometry Techniques and Applications","field":"Chemistry","cited_by":29,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Computer science; Workflow; Mass spectrometry imaging; Source code; Sample (material); Data mining; Multivariate statistics; R package; Analyte; Code (set theory); Pattern recognition (psychology); Mass spectrometry; Artificial intelligence; Database; Machine learning; Computational science; Chemistry; Programming language","score_opus":0.011835380283237948,"score_gpt":0.2730484600134204,"score_spread":0.2612130797301825,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2889339419","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.057654105,0.0022215145,0.8611079,0.0014107001,0.00045312304,0.0005844399,0.026309991,0.045648567,0.004609674],"genre_scores_gemma":[0.09643242,0.0008157832,0.8478668,0.0005372603,0.000196439,0.0008215842,0.044442534,0.0067261984,0.0021610053],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99245524,0.0012903234,0.0007757764,0.002929755,0.0021719993,0.00037689254],"domain_scores_gemma":[0.99091876,0.0023726157,0.0013857598,0.0029808155,0.0020135168,0.0003286324],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.009965042,0.0024428742,0.0021999038,0.006249617,0.0015357528,0.0044823014,0.0033734848,0.0016099947,0.0042904443],"category_scores_gemma":[0.018609414,0.0012097913,0.0029719127,0.006273846,0.0014215357,0.0029439686,0.006298444,0.0023932185,0.006594822],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0028415634,0.00079084863,0.05643102,0.006417712,0.004786862,0.0020626674,0.0020951824,0.04238911,0.33841845,0.017853249,0.118867315,0.4070461],"study_design_scores_gemma":[0.00031812972,0.00042194888,0.056051794,0.000536588,0.0016199801,0.004661573,0.0011327709,0.26940352,0.41427222,0.054468293,0.19641858,0.0006945799],"about_ca_topic_score_codex":0.0023286075,"about_ca_topic_score_gemma":0.004441351,"teacher_disagreement_score":0.009965042,"about_ca_system_score_codex":0.0009454961,"about_ca_system_score_gemma":0.0027974837,"threshold_uncertainty_score":0.052700818},"labels":[],"label_agreement":null},{"id":"W2889433641","doi":"10.1093/bioinformatics/bty725","title":"Efficient computation of the kinship coefficients","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic and phenotypic traits in livestock","field":"Biochemistry, Genetics and Molecular Biology","cited_by":12,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Kinship; Inbreeding; Pedigree chart; Population; Mendelian inheritance; Biology; Mathematics; Genetics; Combinatorics; Genealogy; Statistics; Computer science; Demography; Sociology","score_opus":0.013356671602386468,"score_gpt":0.24442406810928294,"score_spread":0.23106739650689648,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2889433641","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.06753359,0.00051848777,0.9189317,0.000262242,0.000055466327,0.00014618952,0.0020577693,0.004317805,0.00617675],"genre_scores_gemma":[0.31454408,0.0003230807,0.67334557,0.000119202436,0.000040132665,0.00020088966,0.005401514,0.0006395162,0.005385967],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9989692,0.00015630787,0.00006587862,0.00027775456,0.000356183,0.00017468596],"domain_scores_gemma":[0.99701846,0.0013579007,0.00024261151,0.0005360247,0.0006921886,0.0001529042],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008374102,0.00078860397,0.0008592547,0.0013308985,0.000631696,0.0015956599,0.0015406815,0.0006021265,0.010765548],"category_scores_gemma":[0.009701007,0.0004276771,0.00064209657,0.0017626482,0.0004673726,0.0020028376,0.0016341988,0.0011385948,0.0039402605],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005392643,0.0001864876,0.016088063,0.00075482053,0.00014526398,0.0003237078,0.0005207809,0.23994787,0.010514077,0.023517532,0.021846104,0.6856161],"study_design_scores_gemma":[0.00012839324,0.00009156076,0.0070162863,0.00008277618,0.000050062667,0.00045078795,0.00019655602,0.89313465,0.008901439,0.0774196,0.012480786,0.00004716467],"about_ca_topic_score_codex":0.0075823707,"about_ca_topic_score_gemma":0.0096069975,"teacher_disagreement_score":0.010765548,"about_ca_system_score_codex":0.0010155026,"about_ca_system_score_gemma":0.002599961,"threshold_uncertainty_score":0.036014378},"labels":[],"label_agreement":null},{"id":"W2890693027","doi":"10.1093/bioinformatics/bty812","title":"PRESM: personalized reference editor for somatic mutation discovery in cancer genomics","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Cancer Genomics and Diagnostics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Alberta Children's Hospital; University of Calgary","funders":"Canada Foundation for Innovation; Arkansas Children’s Hospital Research Institute","keywords":"Somatic cell; Genomics; Computational biology; Mutation; Cancer genetics; Personalized medicine; Germline mutation; Cancer; Computer science; Genetics; Biology; Genome; Gene","score_opus":0.016820696563665267,"score_gpt":0.2798510192886188,"score_spread":0.2630303227249535,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2890693027","genre_codex":"software","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.002849476,0.0010949326,0.33571544,0.0010248218,0.0013364775,0.00032006006,0.07500693,0.5764764,0.0061754766],"genre_scores_gemma":[0.033771943,0.0012271308,0.65460217,0.0028540613,0.00066565257,0.0019096143,0.17988482,0.107529365,0.017555337],"study_design_codex":"not_applicable","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9965951,0.00076658174,0.000370903,0.0009741942,0.0010849753,0.00020815399],"domain_scores_gemma":[0.99193805,0.004640446,0.0005540458,0.0014696014,0.001035071,0.0003628105],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0058770142,0.0024254206,0.0016486533,0.0027264825,0.0010318161,0.002437803,0.0049793296,0.0029345066,0.09181378],"category_scores_gemma":[0.02574891,0.0017145183,0.0018669306,0.0025089383,0.000751908,0.0028011561,0.004136295,0.0031204608,0.05449223],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013385037,0.000099120436,0.0031959154,0.0033263573,0.00032730887,0.0011486812,0.00049285154,0.0036544616,0.020842109,0.012610225,0.7759935,0.17697091],"study_design_scores_gemma":[0.0005855076,0.00026955822,0.0034724222,0.0005490472,0.00017270524,0.0014661421,0.00014279272,0.04008477,0.06597247,0.028563969,0.858513,0.00020755306],"about_ca_topic_score_codex":0.0011469868,"about_ca_topic_score_gemma":0.0024515348,"teacher_disagreement_score":0.09181378,"about_ca_system_score_codex":0.0008947461,"about_ca_system_score_gemma":0.0022759184,"threshold_uncertainty_score":0.30714768},"labels":[],"label_agreement":null},{"id":"W2890720538","doi":"10.1093/bioinformatics/bty773","title":"SCOP: a novel scaffolding algorithm based on contig classification and optimization","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":16,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Saskatchewan","funders":"Higher Education Discipline Innovation Project; National Natural Science Foundation of China","keywords":"Contig; Computer science; Scaffold; Spurious relationship; Graph; Sequence assembly; Algorithm; Cuboid; Pattern recognition (psychology); Artificial intelligence; Theoretical computer science; Genome; Machine learning; Biology; Mathematics; Genetics; Programming language","score_opus":0.018884952096979497,"score_gpt":0.24170945418960468,"score_spread":0.2228245020926252,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2890720538","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.021888789,0.000499521,0.948112,0.00020944183,0.0001288231,0.0002690066,0.0015201163,0.024680927,0.002691276],"genre_scores_gemma":[0.077610046,0.00031020993,0.9021803,0.00024809706,0.00010288276,0.00047123674,0.011702089,0.0035415331,0.0038335414],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99889165,0.00013235181,0.0000768109,0.0003846795,0.00038493564,0.0001295038],"domain_scores_gemma":[0.99819034,0.0005609104,0.00017709575,0.00037138016,0.0005526787,0.00014753813],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011603086,0.0027590208,0.0019074361,0.0040405425,0.0022396562,0.0017153731,0.0027058213,0.0015925006,0.0063745514],"category_scores_gemma":[0.0034981489,0.0009909293,0.0020837171,0.0043684584,0.0011575982,0.0023512265,0.0025221426,0.0021264409,0.0033111367],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006439191,0.00035623758,0.006354223,0.00075050874,0.00020195718,0.0004486068,0.00042132137,0.15500927,0.028827416,0.0099365115,0.050879933,0.7461701],"study_design_scores_gemma":[0.00014628164,0.0001886773,0.0014701176,0.000050346076,0.00007747769,0.0002658339,0.00016068702,0.9500801,0.015101808,0.014854914,0.017542394,0.00006135171],"about_ca_topic_score_codex":0.007491695,"about_ca_topic_score_gemma":0.009343939,"teacher_disagreement_score":0.007491695,"about_ca_system_score_codex":0.0009405396,"about_ca_system_score_gemma":0.0028102603,"threshold_uncertainty_score":0.021324933},"labels":[],"label_agreement":null},{"id":"W2892158700","doi":"10.1093/bioinformatics/bty740","title":"ECCB 2018: The 17th European Conference on Computational Biology","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetics, Bioinformatics, and Biomedical Research","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Computer science; Computational biology; Biology","score_opus":0.037408995564868465,"score_gpt":0.302256421269661,"score_spread":0.2648474257047925,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2892158700","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0035462945,0.13013895,0.22867286,0.06502716,0.21517563,0.002357441,0.023817644,0.023257755,0.30800626],"genre_scores_gemma":[0.015304084,0.11688567,0.12047902,0.018421523,0.059721123,0.004261873,0.092551485,0.023361258,0.54901403],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9863253,0.0036126254,0.0010599919,0.0023766346,0.005498974,0.0011265707],"domain_scores_gemma":[0.97415257,0.00787085,0.0007979464,0.00534663,0.0074804877,0.004351474],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.014453252,0.0037266994,0.005603406,0.0049707186,0.0023018245,0.015023305,0.0071560033,0.006420481,0.3190319],"category_scores_gemma":[0.029823981,0.0014522155,0.0029249638,0.0050379084,0.0036610307,0.013482443,0.01032729,0.009898728,0.2774241],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00009269883,0.000051589202,0.00012301852,0.00047081558,0.000033770135,0.000069544585,0.00006516743,0.00058375637,0.00056019187,0.012090639,0.91169214,0.07416658],"study_design_scores_gemma":[0.000016530274,0.00001859891,0.00016211053,0.00028101986,0.0000057212096,0.00007083591,0.00002262381,0.00071174727,0.00014511745,0.0050515966,0.993497,0.000017119353],"about_ca_topic_score_codex":0.0024886983,"about_ca_topic_score_gemma":0.0016855003,"teacher_disagreement_score":0.3190319,"about_ca_system_score_codex":0.0030207585,"about_ca_system_score_gemma":0.007354318,"threshold_uncertainty_score":0.97131836},"labels":[],"label_agreement":null},{"id":"W2894104460","doi":"10.1093/bioinformatics/bty832","title":"A systematic method for surveying data visualizations and a resulting genomic epidemiology visualization typology: GEViT","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Data Visualization and Analytics","field":"Computer Science","cited_by":23,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"BC Centre for Disease Control; University of British Columbia","funders":"Natural Sciences and Engineering Research Council of Canada; Canadian Institutes of Health Research; University of British Columbia; Canada Research Chairs; Michael Smith Health Research BC; Fred Hutchinson Cancer Research Center","keywords":"Typology; Visualization; Data science; Computer science; Data visualization; Computational biology; Data mining; Biology; Geography; Archaeology","score_opus":0.1553546717332538,"score_gpt":0.43647967099048524,"score_spread":0.28112499925723144,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2894104460","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.024717662,0.0006126554,0.93046176,0.0026435875,0.00027942893,0.0069110305,0.0119797895,0.01250087,0.009893285],"genre_scores_gemma":[0.035491176,0.00018693057,0.9468973,0.00021137694,0.00004044412,0.008718905,0.004905977,0.0014601445,0.002087808],"study_design_codex":"design_other","study_design_gemma":"design_other","domain_scores_codex":[0.954291,0.027426522,0.0069224685,0.004668974,0.0059687304,0.0007223277],"domain_scores_gemma":[0.77060395,0.14408407,0.013697806,0.03495865,0.03401721,0.002638353],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0469923,0.0021902358,0.0012236725,0.019866405,0.00308383,0.010363593,0.0026332464,0.0022056135,0.0142935375],"category_scores_gemma":[0.17640656,0.0018673268,0.0028947117,0.012495658,0.0034173962,0.008957474,0.009340911,0.0035362912,0.0034124856],"study_design_candidate":"design_other","study_design_consensus":"design_other","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008526677,0.0007344036,0.02993949,0.010961582,0.00046278653,0.0007319599,0.11228345,0.004222898,0.01908953,0.15934172,0.10591612,0.5554634],"study_design_scores_gemma":[0.0006014394,0.0005955076,0.018867873,0.007580303,0.00043926356,0.0015153992,0.04718575,0.07240035,0.029710058,0.20929511,0.6110748,0.0007341583],"about_ca_topic_score_codex":0.0024700088,"about_ca_topic_score_gemma":0.005154498,"teacher_disagreement_score":0.0469923,"about_ca_system_score_codex":0.0034009363,"about_ca_system_score_gemma":0.0065169306,"threshold_uncertainty_score":0.24852204},"labels":[],"label_agreement":null},{"id":"W2895908163","doi":"10.1093/bioinformatics/bty883","title":"Landscape of multi-tissue global gene expression reveals the regulatory signatures of feed efficiency in beef cattle","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic and phenotypic traits in livestock","field":"Biochemistry, Genetics and Molecular Biology","cited_by":25,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Agriculture and Agri-Food Canada; Alberta Livestock and Meat Agency; Alberta Agriculture and Forestry","keywords":"Transcriptome; Biology; Residual feed intake; Gene; Feed conversion ratio; Rumen; Gene expression; Beef cattle; Phenotype; Genetics; Regulator gene; Gene expression profiling; Trait; Computational biology; Body weight; Biochemistry","score_opus":0.009546474809135003,"score_gpt":0.2493989742368527,"score_spread":0.2398524994277177,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2895908163","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99092287,0.0009488362,0.0028857454,0.000038832924,0.0000057814636,0.000007213149,0.0046246466,0.00005653855,0.0005095375],"genre_scores_gemma":[0.9886819,0.00036873185,0.0025633657,0.00008616973,0.000012026645,0.000028905171,0.007386053,0.00004322049,0.0008295695],"study_design_codex":"bench_or_experimental","study_design_gemma":"observational","domain_scores_codex":[0.9997904,0.000024080746,0.0000072051375,0.00010331001,0.00002880582,0.000046266767],"domain_scores_gemma":[0.999808,0.00005829514,0.000059777023,0.000012347903,0.000033466444,0.000028028391],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00027666995,0.00017832061,0.00039035725,0.0006203103,0.0002244093,0.0004489267,0.00013177519,0.0002044878,0.0012099278],"category_scores_gemma":[0.0002783649,0.00013461152,0.00031087743,0.0009914752,0.00020739566,0.00020799729,0.00031286417,0.00020805086,0.00019914306],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007650753,0.000047044006,0.23791549,0.00039689805,0.00034498397,0.00016041989,0.0006283991,0.0019635095,0.73522544,0.00032913388,0.000795422,0.02142827],"study_design_scores_gemma":[0.000005543418,0.00009212204,0.98860455,0.000014827459,0.00010096904,0.00015596939,0.0002198831,0.0027674912,0.0065517356,0.00024934587,0.0012230229,0.000014478209],"about_ca_topic_score_codex":0.0017004812,"about_ca_topic_score_gemma":0.0029277923,"teacher_disagreement_score":0.0017004812,"about_ca_system_score_codex":0.00022756135,"about_ca_system_score_gemma":0.00017644944,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W2896090507","doi":"10.1093/bioinformatics/bty825","title":"MetaOmics: analysis pipeline and browser-based software suite for transcriptomic meta-analysis","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":33,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto; Centre for Addiction and Mental Health","funders":"National Cancer Institute; National Institutes of Health; U.S. National Library of Medicine; National Natural Science Foundation of China","keywords":"Suite; Pipeline (software); Computer science; Software; Software suite; Software engineering; World Wide Web; Operating system","score_opus":0.04819263447781635,"score_gpt":0.2974941623703049,"score_spread":0.24930152789248855,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2896090507","genre_codex":"software","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.005599884,0.0008302147,0.42487988,0.00063389435,0.00034012806,0.00094259664,0.12857674,0.4344132,0.003783539],"genre_scores_gemma":[0.028114177,0.0009510847,0.6791664,0.000929796,0.00015385312,0.0053123487,0.20416139,0.07718825,0.0040226686],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9981945,0.00039260008,0.00025714902,0.00046236528,0.0005188683,0.00017456565],"domain_scores_gemma":[0.9975091,0.0010247255,0.00030808884,0.0003952809,0.00050509494,0.00025761063],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0059283148,0.0034121333,0.0018811174,0.0041953754,0.0011409345,0.0028405343,0.003382468,0.0012443092,0.027308619],"category_scores_gemma":[0.0068232864,0.0025203943,0.0035422896,0.0028969888,0.00065895676,0.0019783878,0.0034493413,0.0039002504,0.01855621],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0029988224,0.0004757826,0.015366931,0.0074648056,0.0038310576,0.0011904098,0.0015403527,0.012332557,0.13791624,0.017930912,0.6258482,0.17310391],"study_design_scores_gemma":[0.001229784,0.00042925047,0.022667484,0.0010032381,0.0015318022,0.0016451229,0.00044589207,0.10219927,0.12031088,0.05456733,0.6930436,0.00092632906],"about_ca_topic_score_codex":0.003303435,"about_ca_topic_score_gemma":0.0042294,"teacher_disagreement_score":0.027308619,"about_ca_system_score_codex":0.0010831233,"about_ca_system_score_gemma":0.0042561567,"threshold_uncertainty_score":0.0913564},"labels":[],"label_agreement":null},{"id":"W2897133367","doi":"10.1093/bioinformatics/bty878","title":"BMDExpress 2: enhanced transcriptomic dose-response analysis workflow","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":227,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Health Canada","funders":"National Institute of Environmental Health Sciences; National Institutes of Health","keywords":"Workflow; Computer science; Software; Transcriptome; Software engineering; Data mining; Data science; Database; Operating system; Biology; Gene","score_opus":0.012719542668610605,"score_gpt":0.2668319319701167,"score_spread":0.2541123893015061,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2897133367","genre_codex":"software","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0054629263,0.0005293731,0.38306063,0.00084097835,0.00050456997,0.00085472804,0.20705847,0.39199182,0.009696511],"genre_scores_gemma":[0.02592133,0.00083216326,0.5117707,0.0033074177,0.00021527392,0.0067382897,0.31809956,0.112198256,0.020917004],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9971654,0.00038705982,0.00030926004,0.00081143214,0.001112213,0.00021465705],"domain_scores_gemma":[0.9967468,0.001525474,0.00024630973,0.0005473868,0.00077462714,0.00015939938],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0052812058,0.0022862093,0.0018853642,0.0023687594,0.001063291,0.0031893603,0.0032946216,0.0014113642,0.06499657],"category_scores_gemma":[0.0065535023,0.0022939893,0.0027409343,0.001638665,0.00073777157,0.0014936976,0.0030657072,0.0033298938,0.05087113],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0017768389,0.00022273902,0.0057656257,0.003781361,0.00044212324,0.00086279557,0.0008289178,0.008913109,0.08538679,0.008693162,0.77212346,0.11120302],"study_design_scores_gemma":[0.00060228043,0.00024580985,0.009848394,0.0005684613,0.00021904144,0.0007115974,0.00021243063,0.04551249,0.13244031,0.021060435,0.78802085,0.00055792794],"about_ca_topic_score_codex":0.0031694495,"about_ca_topic_score_gemma":0.0047636093,"teacher_disagreement_score":0.06499657,"about_ca_system_score_codex":0.0012149,"about_ca_system_score_gemma":0.0032982847,"threshold_uncertainty_score":0.21743518},"labels":[],"label_agreement":null},{"id":"W2898449026","doi":"10.1093/bioinformatics/bty888","title":"Alignment-free clustering of UMI tagged DNA molecules","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Cancer Genomics and Diagnostics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":44,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia; Simon Fraser University","funders":"","keywords":"Barcode; Cluster analysis; Computer science; DNA sequencing; Computational biology; Genomics; Data mining; Biology; DNA; Genetics; Genome; Gene; Artificial intelligence","score_opus":0.00933349121835504,"score_gpt":0.23147020391928735,"score_spread":0.22213671270093233,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2898449026","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.042012755,0.0004067015,0.94127184,0.00018955584,0.00009699494,0.00024379526,0.0018825271,0.011112023,0.0027837646],"genre_scores_gemma":[0.16957553,0.00031625506,0.8118967,0.00022171359,0.000045537923,0.00042880772,0.0123589765,0.0017163533,0.0034402085],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99872154,0.00020682128,0.000062948806,0.0004860449,0.0004198805,0.0001027063],"domain_scores_gemma":[0.9980848,0.00055562507,0.00020950568,0.0003492073,0.0006772246,0.00012361661],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010636309,0.0013840485,0.0013102476,0.0022493075,0.0014680275,0.0015978492,0.0038777993,0.0014178441,0.004059702],"category_scores_gemma":[0.0047923755,0.0006795187,0.0013422295,0.0030356122,0.00093043,0.0012038334,0.0014447345,0.0012442613,0.0030958415],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00067064515,0.00016758057,0.006443489,0.00089361856,0.0002534208,0.00040954142,0.00043101446,0.75233847,0.04344477,0.011571113,0.014437831,0.16893849],"study_design_scores_gemma":[0.000027457236,0.000040900668,0.00082232355,0.00002040608,0.00001944217,0.000067938265,0.000052100153,0.9687345,0.018265815,0.008170489,0.0037536537,0.000025013715],"about_ca_topic_score_codex":0.011510091,"about_ca_topic_score_gemma":0.013596551,"teacher_disagreement_score":0.011510091,"about_ca_system_score_codex":0.0018536071,"about_ca_system_score_gemma":0.0019954313,"threshold_uncertainty_score":0.022886217},"labels":[],"label_agreement":null},{"id":"W2899545492","doi":"10.1093/bioinformatics/bty908","title":"BiXGBoost: a scalable, flexible boosting-based method for reconstructing gene regulatory networks","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene Regulatory Network Analysis","field":"Biochemistry, Genetics and Molecular Biology","cited_by":86,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Saskatchewan","funders":"Higher Education Discipline Innovation Project; National Natural Science Foundation of China","keywords":"Computer science; Scalability; Gene regulatory network; Inference; Data mining; Python (programming language); Boosting (machine learning); Machine learning; Artificial intelligence; Gene; Gene expression; Biology; Database","score_opus":0.014920243380440353,"score_gpt":0.2667671860171329,"score_spread":0.25184694263669255,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2899545492","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.015680935,0.00080193294,0.9621008,0.00025881318,0.00013439707,0.00013931969,0.001496002,0.017858263,0.0015294756],"genre_scores_gemma":[0.13254571,0.00037106167,0.8525573,0.00045094333,0.000100084086,0.0004262485,0.0070878696,0.0027669938,0.003693874],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9995797,0.000100339035,0.00001540925,0.00011443646,0.00013992368,0.00005023393],"domain_scores_gemma":[0.99941707,0.0002533235,0.000050630402,0.00009984146,0.00011355873,0.00006559738],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015432268,0.0011974804,0.0014623062,0.0014319093,0.00072541833,0.0010765059,0.0023182826,0.0012137439,0.005777263],"category_scores_gemma":[0.0033495273,0.0008563159,0.0013722661,0.0011665539,0.00057036977,0.0008979544,0.0013707405,0.0018333835,0.0022904235],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00050081906,0.00026309086,0.0065881233,0.0007029379,0.0005533982,0.00032591156,0.000254544,0.47306237,0.022257734,0.013830943,0.04433692,0.43732315],"study_design_scores_gemma":[0.000059595306,0.000021680746,0.00047489625,0.00001947678,0.000022827846,0.000035186793,0.00001442227,0.9865371,0.0015376994,0.0071828444,0.0040814867,0.00001274324],"about_ca_topic_score_codex":0.008282182,"about_ca_topic_score_gemma":0.01306779,"teacher_disagreement_score":0.008282182,"about_ca_system_score_codex":0.00069635775,"about_ca_system_score_gemma":0.0018332177,"threshold_uncertainty_score":0.019326866},"labels":[],"label_agreement":null},{"id":"W2899971763","doi":"10.1093/bioinformatics/bty911","title":"Discovering network phenotype between genetic risk factors and disease status via diagnosis-aligned multi-modality regression method in Alzheimer’s disease","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Functional Brain Connectivity Studies","field":"Neuroscience","cited_by":28,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Institute of Biomedical Imaging and Bioengineering; National Institute on Aging; Canadian Institutes of Health Research; National Institutes of Health; Stanley Medical Research Institute; National Natural Science Foundation of China; U.S. Department of Defense","keywords":"Disease; Phenotype; Regression; Clinical phenotype; Computational biology; Computer science; Artificial intelligence; Medicine; Bioinformatics; Genetics; Biology; Statistics; Pathology; Gene; Mathematics","score_opus":0.06549029352753541,"score_gpt":0.32384701547034633,"score_spread":0.2583567219428109,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2899971763","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.07032734,0.0017202527,0.92265695,0.00084716734,0.00008523485,0.00007084733,0.0011363497,0.0019865173,0.0011693373],"genre_scores_gemma":[0.7114759,0.001339637,0.2768021,0.00047422422,0.0002564105,0.00020433431,0.005313988,0.00015968333,0.0039737793],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9995042,0.00015203063,0.000024111674,0.00019344299,0.00007584495,0.000050360217],"domain_scores_gemma":[0.99931324,0.0003368882,0.00008613171,0.000088646,0.00013268134,0.000042350315],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010413652,0.0010704778,0.0009712161,0.0009224416,0.00037081938,0.00057405606,0.0013543764,0.0010261759,0.0016794995],"category_scores_gemma":[0.003224571,0.00033173308,0.0011316027,0.00088741333,0.000290395,0.0007704969,0.0006750662,0.0011642921,0.0006897617],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006812292,0.0005181856,0.019653868,0.00035743485,0.00043351125,0.001216387,0.00027526103,0.43997428,0.016538063,0.005624283,0.015417018,0.4993105],"study_design_scores_gemma":[0.000010654277,0.000021143385,0.001107278,0.0000075810904,0.000027865599,0.000098271266,0.000011699068,0.995535,0.00064950215,0.002073241,0.00044830737,0.000009335845],"about_ca_topic_score_codex":0.009020288,"about_ca_topic_score_gemma":0.00774177,"teacher_disagreement_score":0.009020288,"about_ca_system_score_codex":0.00036723597,"about_ca_system_score_gemma":0.00072525174,"threshold_uncertainty_score":0.017935574},"labels":[],"label_agreement":null},{"id":"W2903232627","doi":"10.1093/bioinformatics/bty976","title":"Detection of rare disease variants in extended pedigrees using RVS","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université Laval","funders":"National Institute of Dental and Craniofacial Research","keywords":"Pedigree chart; Bioconductor; Genetics; Biology; Genetic linkage; Linkage (software); Computational biology; Computer science; Gene","score_opus":0.018304788989561893,"score_gpt":0.27315039777806716,"score_spread":0.25484560878850526,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2903232627","genre_codex":"empirical","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.44132954,0.0028544485,0.3795831,0.0008235928,0.0003122767,0.00027166127,0.12394164,0.038347237,0.012536515],"genre_scores_gemma":[0.6391685,0.0010681973,0.2581505,0.00030822773,0.00019083808,0.00038503011,0.09206916,0.005168833,0.0034906613],"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9973369,0.0008059682,0.00026398644,0.0010542301,0.00041766788,0.000121296085],"domain_scores_gemma":[0.98951507,0.006248713,0.0018931114,0.0013785859,0.00059923,0.0003652781],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004381274,0.0005994844,0.0010961932,0.0053498033,0.00053753785,0.0015766871,0.0011328835,0.0005043165,0.014436491],"category_scores_gemma":[0.016989307,0.00059804204,0.0010961451,0.0052201482,0.0003732446,0.0006302522,0.0015835017,0.00055088906,0.004762782],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0021937382,0.0001845845,0.43491155,0.0031421392,0.0036128035,0.0052081384,0.0023349638,0.027747853,0.03362853,0.015367973,0.116188996,0.35547876],"study_design_scores_gemma":[0.0011518772,0.0006626355,0.4606315,0.0013201206,0.0027183397,0.012224398,0.00078647526,0.23133388,0.02446391,0.060186192,0.20408525,0.0004353153],"about_ca_topic_score_codex":0.002640721,"about_ca_topic_score_gemma":0.004865182,"teacher_disagreement_score":0.014436491,"about_ca_system_score_codex":0.00027628272,"about_ca_system_score_gemma":0.00053146225,"threshold_uncertainty_score":0.048294842},"labels":[],"label_agreement":null},{"id":"W2904816621","doi":"10.1093/bioinformatics/bty1019","title":"ModL: exploring and restoring regularity when testing for positive selection","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bayesian Methods and Mixture Models","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Dalhousie University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Type I and type II errors; Selection (genetic algorithm); Context (archaeology); Statistics; Statistical power; Null hypothesis; Statistical hypothesis testing; Mathematics; Mixing (physics); Likelihood-ratio test; Power (physics); Model selection; Type (biology); Variety (cybernetics); Chi-square test; Computer science; Artificial intelligence; Biology","score_opus":0.09692745390537694,"score_gpt":0.2875576846226645,"score_spread":0.19063023071728757,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2904816621","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.014643839,0.0003842359,0.9498072,0.0015949833,0.0001856961,0.00035059574,0.0024161148,0.028574081,0.0020432284],"genre_scores_gemma":[0.1004244,0.00020701425,0.88606954,0.0011219895,0.00027079068,0.0010711543,0.0034022874,0.006360627,0.0010722422],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9811859,0.011662931,0.0013143823,0.002692128,0.0026221443,0.0005224533],"domain_scores_gemma":[0.85208863,0.12686105,0.005549079,0.010876335,0.0033252854,0.0012995617],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.035437923,0.0020259598,0.0019248712,0.0033514618,0.0014081361,0.0036556234,0.007167456,0.002696879,0.012792662],"category_scores_gemma":[0.18649426,0.0018190945,0.002254598,0.0033068717,0.0047663995,0.005518235,0.005999514,0.0041760206,0.0046775537],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0025198031,0.00041374296,0.07749614,0.003006368,0.0009657564,0.0020803837,0.0014077749,0.10131056,0.017773159,0.103024185,0.12468439,0.56531775],"study_design_scores_gemma":[0.00076766295,0.00036522964,0.0057678735,0.00038602858,0.00022606323,0.0016924437,0.00021741034,0.6872228,0.0193074,0.2511315,0.032673653,0.00024200066],"about_ca_topic_score_codex":0.0018337362,"about_ca_topic_score_gemma":0.0025515836,"teacher_disagreement_score":0.035437923,"about_ca_system_score_codex":0.0014512416,"about_ca_system_score_gemma":0.004141608,"threshold_uncertainty_score":0.1874159},"labels":[],"label_agreement":null},{"id":"W2904871025","doi":"10.1093/bioinformatics/bty992","title":"Gene expression models based on transcription factor binding events confer insight into functional <i>cis</i> -regulatory variants","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":14,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"BC Children's Hospital; University of British Columbia","funders":"National Institute of General Medical Sciences; Natural Sciences and Engineering Research Council of Canada; Genome British Columbia; BC Children’s Hospital Foundation; BC Children's Hospital; National Institutes of Health; Children's Hospital Foundation; Canadian Institutes of Health Research; China Scholarship Council; Genome Canada","keywords":"CTCF; Computational biology; Transcription factor; Gene; Biology; Genetics; DNA binding site; Regulation of gene expression; Regulatory sequence; Gene expression; Enhancer; Promoter","score_opus":0.015847253369992757,"score_gpt":0.21414458782602006,"score_spread":0.1982973344560273,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2904871025","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.36285678,0.00093792926,0.6149159,0.00080999616,0.00011456126,0.000101820435,0.011954728,0.00522901,0.003079255],"genre_scores_gemma":[0.887974,0.0003922708,0.096623644,0.0002584828,0.00004351568,0.00022025865,0.01175402,0.0005396237,0.002194149],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9997148,0.00009976212,0.000013132137,0.000098815144,0.00004310837,0.000030318384],"domain_scores_gemma":[0.9984536,0.0012487895,0.000071790026,0.00007575374,0.00011648304,0.00003357906],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011408167,0.0010948239,0.0006175669,0.000559174,0.00024696594,0.0007353041,0.0009168159,0.0008968919,0.0029547855],"category_scores_gemma":[0.004059999,0.00029843807,0.0012954207,0.0006717002,0.00039381318,0.0005072595,0.00047525731,0.0012497999,0.00092057325],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00025897048,0.00005278311,0.012880933,0.00010668086,0.00015119999,0.00010047444,0.000041764717,0.963594,0.0057825614,0.001985413,0.0017098068,0.013335402],"study_design_scores_gemma":[0.000011117664,0.000022710228,0.001248846,0.000008952646,0.000030645842,0.000026736128,0.000008993139,0.99380565,0.0013299431,0.0028651021,0.00063385826,0.0000075456164],"about_ca_topic_score_codex":0.008462147,"about_ca_topic_score_gemma":0.0078117736,"teacher_disagreement_score":0.008462147,"about_ca_system_score_codex":0.000692415,"about_ca_system_score_gemma":0.00081815146,"threshold_uncertainty_score":0.016825795},"labels":[],"label_agreement":null},{"id":"W2905275624","doi":"10.1093/bioinformatics/bty1003","title":"Pathogen–Host Analysis Tool (PHAT): an integrative platform to analyze next-generation sequencing data","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bacteriophages and microbial interactions","field":"Environmental Science","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"NOSM University; Lakehead University; Thunder Bay Regional Research Institute","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Host (biology); Pathogen; Computer science; Computational biology; Data science; Biology; Genetics","score_opus":0.07464074968777398,"score_gpt":0.29671725205821314,"score_spread":0.22207650237043916,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2905275624","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.006620966,0.00073276326,0.47762358,0.0008492311,0.0006029996,0.0008263545,0.083010875,0.42457575,0.0051575736],"genre_scores_gemma":[0.029952556,0.00083121634,0.7336065,0.0010167371,0.00024668322,0.003118905,0.14111368,0.084402256,0.005711349],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9977719,0.0005364058,0.00020560113,0.00062743714,0.00064336567,0.00021518413],"domain_scores_gemma":[0.9956418,0.0020375664,0.00050573994,0.00068686495,0.0006461416,0.00048186205],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.006784451,0.0025758457,0.001515689,0.0043869494,0.0018338742,0.003175062,0.0026460271,0.0013558452,0.04302726],"category_scores_gemma":[0.01283755,0.0021233573,0.0021690053,0.0031760037,0.0008601058,0.0030710427,0.0045640445,0.0036836471,0.028029881],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.002124304,0.00028202194,0.00800406,0.0036339317,0.0009487437,0.001358484,0.001401593,0.004557285,0.06044143,0.009429829,0.7408465,0.16697182],"study_design_scores_gemma":[0.0012697442,0.00070183486,0.014950775,0.001131134,0.00062865624,0.0030535287,0.0004749441,0.091429286,0.11712038,0.045312855,0.72293615,0.0009906652],"about_ca_topic_score_codex":0.001634254,"about_ca_topic_score_gemma":0.0017266185,"teacher_disagreement_score":0.04302726,"about_ca_system_score_codex":0.00077898556,"about_ca_system_score_gemma":0.0029772273,"threshold_uncertainty_score":0.14394051},"labels":[],"label_agreement":null},{"id":"W2905575949","doi":"10.1093/bioinformatics/bty1015","title":"SPRING: a next-generation compressor for FASTQ data","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Algorithms and Data Compression","field":"Computer Science","cited_by":91,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Cancer Institute; University of Illinois at Urbana-Champaign; National Institutes of Health; Sunnybrook Research Institute; Silicon Valley Community Foundation","keywords":"Computer science; Lossless compression; Scalability; Lossy compression; Data compression; Data mining; Redundancy (engineering); Identifier; Compression (physics); Database; Artificial intelligence; Computer network; Operating system","score_opus":0.1701892097079504,"score_gpt":0.3126197987859841,"score_spread":0.14243058907803371,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2905575949","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.03717218,0.001932525,0.7046961,0.0020845023,0.0012806546,0.00093226664,0.019589502,0.2201468,0.012165411],"genre_scores_gemma":[0.16683571,0.0014459696,0.7144243,0.0018313781,0.00070378656,0.0019957437,0.06849612,0.014386315,0.029880723],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99819845,0.00016157361,0.00016794253,0.00031195616,0.0010142236,0.00014596805],"domain_scores_gemma":[0.99660695,0.0009193017,0.00018962083,0.000920224,0.001223359,0.00014063029],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002101411,0.0016274801,0.0006736658,0.0017276222,0.0012820471,0.0015120462,0.0030153408,0.0011907711,0.02184419],"category_scores_gemma":[0.008034871,0.00059503916,0.0008668639,0.0025093667,0.0011980913,0.0034657235,0.0023743731,0.0018180202,0.009098105],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0029308214,0.00030501318,0.004287962,0.0012346874,0.00026627202,0.0007021545,0.0007338182,0.014102874,0.06632976,0.021567788,0.37592146,0.5116174],"study_design_scores_gemma":[0.00095584383,0.000643967,0.0031417296,0.00032352266,0.00013791923,0.0013045961,0.00032122486,0.26977253,0.35615087,0.02458731,0.3423069,0.0003536525],"about_ca_topic_score_codex":0.0030788467,"about_ca_topic_score_gemma":0.0032420333,"teacher_disagreement_score":0.02184419,"about_ca_system_score_codex":0.0010699354,"about_ca_system_score_gemma":0.0016290754,"threshold_uncertainty_score":0.07307607},"labels":[],"label_agreement":null},{"id":"W2906587988","doi":"10.1093/bioinformatics/bty1035","title":"Systematic selection of chemical fingerprint features improves the Gibbs energy prediction of biochemical reactions","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Machine Learning in Materials Science","field":"Materials Science","cited_by":18,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"King Abdullah University of Science and Technology","keywords":"Computer science; Software; Feature selection; Fingerprint (computing); Data mining; Artificial intelligence; Gibbs free energy; Machine learning; Selection (genetic algorithm); Energy (signal processing); Feature (linguistics); Mathematics; Statistics","score_opus":0.0071927714951737845,"score_gpt":0.23005514919194137,"score_spread":0.2228623776967676,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2906587988","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.33986178,0.0012176107,0.6487069,0.00037644533,0.000074703945,0.00013330455,0.0015436449,0.006478741,0.0016069043],"genre_scores_gemma":[0.7970214,0.000387075,0.19873345,0.000116418225,0.000046400095,0.00018183168,0.0024527914,0.00038540343,0.00067528075],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9995931,0.00007944787,0.000029129842,0.00013746749,0.00012359156,0.00003736335],"domain_scores_gemma":[0.9984794,0.0007987814,0.00023174702,0.00022411963,0.00020873538,0.000057177953],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010904907,0.0010407834,0.0008406314,0.0011221494,0.0003880402,0.0005745312,0.0007646406,0.0006038921,0.0013115155],"category_scores_gemma":[0.004315198,0.0002921283,0.0010741829,0.0007766911,0.00037644294,0.00094226986,0.0007062861,0.00072363723,0.0006647159],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005539854,0.00052504084,0.032994352,0.00079583394,0.00022555853,0.0002647172,0.000074207084,0.4715496,0.12797663,0.0025478117,0.0052822526,0.35720998],"study_design_scores_gemma":[0.000028527016,0.00007849012,0.0025791647,0.00001564713,0.00004270671,0.00006219178,0.000007052943,0.96031487,0.0343633,0.001604498,0.0008794255,0.000024095325],"about_ca_topic_score_codex":0.0016527734,"about_ca_topic_score_gemma":0.002131591,"teacher_disagreement_score":0.0016527734,"about_ca_system_score_codex":0.00036306013,"about_ca_system_score_gemma":0.0010070897,"threshold_uncertainty_score":0.0057671666},"labels":[],"label_agreement":null},{"id":"W2907347464","doi":"10.1093/bioinformatics/bty1053","title":"The COP9 signalosome influences the epigenetic landscape of <i>Arabidopsis thaliana</i>","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Ubiquitin and proteasome pathways","field":"Biochemistry, Genetics and Molecular Biology","cited_by":19,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Azrieli Foundation; H2020 European Research Council; Tel Aviv University","keywords":"Arabidopsis thaliana; COP9 signalosome; Epigenetics; Biology; Computational biology; Arabidopsis; Genetics; Gene","score_opus":0.013554330975524229,"score_gpt":0.24077501050217973,"score_spread":0.2272206795266555,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2907347464","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97505856,0.0015098788,0.0071397917,0.0002480537,0.000058749312,0.000028367645,0.009260227,0.0011533011,0.00554313],"genre_scores_gemma":[0.98895246,0.00030315085,0.002342668,0.00020692562,0.000008347982,0.000017933511,0.0044771913,0.00025233606,0.0034390122],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.999941,0.0000041657718,0.0000029222788,0.000031205,0.0000108622635,0.000009750539],"domain_scores_gemma":[0.99991167,0.000012134999,0.000026497491,0.000008105174,0.000013705693,0.000027887605],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000064635395,0.00024477692,0.00017183465,0.00019782719,0.00016817485,0.00028002908,0.00019999515,0.00015599545,0.003949573],"category_scores_gemma":[0.00009175357,0.00011801526,0.00020426956,0.00014489568,0.00010577941,0.00015756812,0.00021329106,0.00030829455,0.0010260879],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00010705184,0.000006293466,0.0013066919,0.000047044156,0.000007365122,0.00003292161,0.000013246997,0.0001204922,0.9958462,0.00010310836,0.00033368866,0.0020759094],"study_design_scores_gemma":[0.00004307201,0.00011592902,0.19855969,0.000034556466,0.000082925595,0.00041569976,0.000098051874,0.0073651667,0.7787511,0.00046200163,0.014028269,0.00004352453],"about_ca_topic_score_codex":0.003457011,"about_ca_topic_score_gemma":0.00518459,"teacher_disagreement_score":0.003949573,"about_ca_system_score_codex":0.000565529,"about_ca_system_score_gemma":0.00015985125,"threshold_uncertainty_score":0.013212681},"labels":[],"label_agreement":null},{"id":"W2907957654","doi":"10.1093/bioinformatics/bty1067","title":"UbiHub: a data hub for the explorers of ubiquitination pathways","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Ubiquitin and proteasome pathways","field":"Biochemistry, Genetics and Molecular Biology","cited_by":68,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Windsor; Structural Genomics Consortium; University of Toronto","funders":"Eshelman Institute for Innovation, University of North Carolina at Chapel Hill; Janssen Biotech; Innovative Medicines Initiative; Novartis Pharma; Ontario Genomics Institute; Canada Foundation for Innovation; Wellcome Trust; Ontario Ministry of Research, Innovation and Science; Ontario Genomics; Fundação de Amparo à Pesquisa do Estado de São Paulo; Genome Canada; AbbVie; European Federation of Pharmaceutical Industries and Associations; Structural Genomics Consortium; Merck KGaA; Pfizer; Boehringer Ingelheim","keywords":"Ubiquitin; Proteasome; Drug discovery; Deubiquitinating enzyme; Computational biology; Protein degradation; Biology; Ubiquitin-Protein Ligases; Ubiquitin ligase; Cell biology; Bioinformatics; Biochemistry","score_opus":0.07489918757800679,"score_gpt":0.29358765037220486,"score_spread":0.21868846279419807,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2907957654","genre_codex":"dataset","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0038339526,0.0011931683,0.08910866,0.0011032763,0.00028375146,0.00033481812,0.50347364,0.38461322,0.016055545],"genre_scores_gemma":[0.036066543,0.0011600323,0.14730632,0.0010830498,0.00011454618,0.001102561,0.7502816,0.058279995,0.0046053785],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9983302,0.00027156217,0.000213944,0.00044921102,0.000522035,0.0002130965],"domain_scores_gemma":[0.99564964,0.0012545408,0.00031593005,0.0013666028,0.00075053854,0.00066282536],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0031214207,0.0026118637,0.0017990231,0.0046505453,0.0015172282,0.004528569,0.0044507096,0.0018028558,0.050152954],"category_scores_gemma":[0.010414809,0.0013693093,0.0018297964,0.006043081,0.00092497194,0.0047245263,0.0072471444,0.0032025366,0.040354572],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0018796977,0.00015307746,0.00486627,0.0027832983,0.00028713635,0.00046572086,0.000508259,0.004979333,0.006063293,0.021756344,0.8895003,0.066757284],"study_design_scores_gemma":[0.0005692577,0.000071681025,0.004696151,0.0005117609,0.00013174178,0.00039601975,0.0002783197,0.034354717,0.016351206,0.04288685,0.8995396,0.00021272621],"about_ca_topic_score_codex":0.0088237785,"about_ca_topic_score_gemma":0.010769017,"teacher_disagreement_score":0.050152954,"about_ca_system_score_codex":0.0023209245,"about_ca_system_score_gemma":0.0032812438,"threshold_uncertainty_score":0.16777837},"labels":[],"label_agreement":null},{"id":"W2908677338","doi":"10.1093/bioinformatics/btz012","title":"A web application and service for imputing and visualizing missense variant effect maps","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Rare Diseases","field":"Biochemistry, Genetics and Molecular Biology","cited_by":24,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canadian Institute for Advanced Research; Sinai Health System; Lunenfeld-Tanenbaum Research Institute; University of Toronto","funders":"National Human Genome Research Institute; Canadian Institutes of Health Research; National Institutes of Health; Canada Research Chairs","keywords":"Missense mutation; Imputation (statistics); Computer science; Pipeline (software); Software; Source code; Data mining; Web service; Missing data; Open source; Exome; Computational biology; Biology; Mutation; World Wide Web; Genetics; Machine learning; Exome sequencing; Gene; Programming language","score_opus":0.004101546701540773,"score_gpt":0.23331096607724777,"score_spread":0.229209419375707,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2908677338","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0020871472,0.00037448906,0.18859938,0.0005632587,0.00029601788,0.00033201362,0.12064918,0.6832349,0.0038636778],"genre_scores_gemma":[0.035661723,0.0010081778,0.42028898,0.0020749923,0.0003893021,0.0030571858,0.34670144,0.175331,0.015487266],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99884367,0.0001977116,0.00013197037,0.0002786777,0.00043638737,0.00011162133],"domain_scores_gemma":[0.99617225,0.0019546568,0.00028447938,0.0006721536,0.00054557435,0.00037083757],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0030150597,0.0029038538,0.0019441406,0.004386127,0.001289036,0.002554313,0.0037039926,0.002562178,0.15238757],"category_scores_gemma":[0.010944247,0.0018644638,0.0028305869,0.0038352662,0.0006532628,0.0023050846,0.0052477154,0.003517704,0.08382917],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011974977,0.0002792452,0.005644651,0.0014129375,0.0007405625,0.000996821,0.0003202739,0.0039609224,0.007418091,0.004047229,0.8621089,0.11187303],"study_design_scores_gemma":[0.0029061923,0.00024042398,0.019871881,0.0007619956,0.00052445906,0.0026097714,0.00027260056,0.117268056,0.030162597,0.063753985,0.7608792,0.0007488649],"about_ca_topic_score_codex":0.0060736956,"about_ca_topic_score_gemma":0.008076644,"teacher_disagreement_score":0.15238757,"about_ca_system_score_codex":0.0008080593,"about_ca_system_score_gemma":0.0021830248,"threshold_uncertainty_score":0.5097872},"labels":[],"label_agreement":null},{"id":"W2909472027","doi":"10.1093/bioinformatics/bty1054","title":"DIABLO: an integrative approach for identifying key molecular drivers from multi-omics assays","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1056,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Prevention of Organ Failure; University of British Columbia","funders":"National Institute of Allergy and Infectious Diseases; National Health and Medical Research Council","keywords":"Bioconductor; Omics; Computer science; Computational biology; Benchmark (surveying); Identification (biology); Relevance (law); Data integration; Visualization; Data mining; Data science; Bioinformatics; Biology; Ecology; Cartography; Geography","score_opus":0.015892733093059504,"score_gpt":0.25083534586353734,"score_spread":0.23494261277047784,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2909472027","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.012722272,0.00074308034,0.9638133,0.00044456281,0.00008327961,0.00021877734,0.010392606,0.00970509,0.0018769989],"genre_scores_gemma":[0.10332887,0.0007359275,0.87420034,0.00047826424,0.00012057153,0.0009810205,0.017179802,0.0018180956,0.0011571115],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9984297,0.0005300827,0.0000767912,0.0004985304,0.00038671415,0.0000781725],"domain_scores_gemma":[0.9959786,0.0025263247,0.00036185907,0.0004836219,0.0004070226,0.00024250112],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.006976734,0.0019900799,0.0013417323,0.004109098,0.00063573907,0.0027738896,0.0027824924,0.00075694156,0.005470053],"category_scores_gemma":[0.010899778,0.00083042594,0.0022362226,0.0026956482,0.00082074234,0.0017045651,0.004033125,0.0022345835,0.0015972165],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0017415334,0.00067206187,0.0675237,0.0046834378,0.0045912922,0.00080957206,0.00081517687,0.21106875,0.08997278,0.056329805,0.06772384,0.4940681],"study_design_scores_gemma":[0.00021705066,0.00033437004,0.015769241,0.00023526115,0.0006343159,0.000516492,0.00021709185,0.84996414,0.02255083,0.065851115,0.04348295,0.00022713181],"about_ca_topic_score_codex":0.0025783621,"about_ca_topic_score_gemma":0.0051459563,"teacher_disagreement_score":0.006976734,"about_ca_system_score_codex":0.0010878475,"about_ca_system_score_gemma":0.0025606249,"threshold_uncertainty_score":0.036896944},"labels":[],"label_agreement":null},{"id":"W2911619868","doi":"10.1093/bioinformatics/btz043","title":"SRG extractor: a skinny reference genome approach for reduced-representation sequencing","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Guelph; Université Laval","funders":"Génome Québec; Iowa State University; Genome Canada","keywords":"Reference genome; Genome; Genotyping; Extractor; Computational biology; Computer science; DNA sequencing; Pipeline (software); Biology; Genetics; Gene; Genotype; Operating system","score_opus":0.03827263071399067,"score_gpt":0.2684018089181237,"score_spread":0.23012917820413303,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2911619868","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0062497053,0.0007731555,0.564062,0.0005748191,0.00047183575,0.0003846713,0.12210291,0.29892564,0.006455276],"genre_scores_gemma":[0.018887734,0.00062808616,0.6404223,0.0007827845,0.00012334787,0.0009638809,0.26777926,0.06352097,0.006891602],"study_design_codex":"not_applicable","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9981002,0.0002914155,0.00020646096,0.0006961515,0.00056215003,0.00014362363],"domain_scores_gemma":[0.9966029,0.0011752832,0.00037474176,0.00089965557,0.0008093137,0.00013809315],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0026308494,0.0027794722,0.0015538802,0.0035777413,0.0012390829,0.002368186,0.0033945614,0.0018447023,0.03752571],"category_scores_gemma":[0.012848982,0.0015806789,0.0017016806,0.0037596137,0.0005902931,0.0021695925,0.0030912873,0.0027160523,0.03748483],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0022031364,0.00023730082,0.005848081,0.004814374,0.0006846348,0.001171891,0.0010591456,0.008265356,0.12442319,0.017966256,0.5877514,0.24557522],"study_design_scores_gemma":[0.00041662148,0.00024023284,0.0075232806,0.00071229564,0.0002821818,0.001145972,0.00028256743,0.04114172,0.13180494,0.028809313,0.7872941,0.00034683643],"about_ca_topic_score_codex":0.0027497578,"about_ca_topic_score_gemma":0.0047330097,"teacher_disagreement_score":0.03752571,"about_ca_system_score_codex":0.0007816107,"about_ca_system_score_gemma":0.0018097811,"threshold_uncertainty_score":0.12553602},"labels":[],"label_agreement":null},{"id":"W2912368193","doi":"10.1093/bioinformatics/btz531","title":"Simultaneous prediction of multiple outcomes using revised stacking algorithms","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"HIV/AIDS drug development and treatment","field":"Medicine","cited_by":30,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Victoria; University of Saskatchewan","funders":"Canada Research Chairs","keywords":"Stacking; Computer science; Algorithm; Chemistry","score_opus":0.025746495298406714,"score_gpt":0.2706057821646921,"score_spread":0.24485928686628539,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2912368193","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.062027328,0.0011590767,0.93080765,0.0010727152,0.0002201584,0.0002557587,0.000909159,0.0023907213,0.001157337],"genre_scores_gemma":[0.5929514,0.0007794889,0.3970953,0.00046750205,0.00058444415,0.0005813704,0.0035846625,0.000298194,0.0036575675],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9968088,0.0013684902,0.00024325894,0.00074281154,0.00054838095,0.00028815243],"domain_scores_gemma":[0.98773766,0.007395749,0.0012582458,0.0010753389,0.0019142304,0.0006187807],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.009357234,0.0037019802,0.0035071594,0.003067045,0.0009831286,0.002411059,0.0033208036,0.002127111,0.004251421],"category_scores_gemma":[0.01863537,0.00095531176,0.0030086522,0.0025807729,0.0009767207,0.0020297784,0.0025960084,0.004106796,0.0017935938],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00056217023,0.000315302,0.018322602,0.00013428275,0.00068769575,0.00020945811,0.00010086281,0.6958172,0.0008877989,0.002969513,0.004201828,0.27579144],"study_design_scores_gemma":[0.000015048649,0.000061578365,0.00065023,0.000011236874,0.000046055102,0.000025510517,0.000008521685,0.9946083,0.00039257726,0.003935575,0.00023148676,0.000013883894],"about_ca_topic_score_codex":0.00826777,"about_ca_topic_score_gemma":0.0070829513,"teacher_disagreement_score":0.009357234,"about_ca_system_score_codex":0.0011133204,"about_ca_system_score_gemma":0.0029412964,"threshold_uncertainty_score":0.04948634},"labels":[],"label_agreement":null},{"id":"W2914005557","doi":"10.1093/bioinformatics/btz084","title":"Multispectral tracing in densely labeled mouse brain with nTracer","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Cell Image Analysis Techniques","field":"Biochemistry, Genetics and Molecular Biology","cited_by":27,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"National Institute of Allergy and Infectious Diseases; National Institute of Neurological Disorders and Stroke; National Institute of General Medical Sciences; National Institute of Mental Health; National Institutes of Health; National Science Foundation","keywords":"Tracing; Multispectral image; Computer science; Plug-in; Software; Documentation; Sample (material); Open source; Artificial intelligence; Computer graphics (images); Computer vision; Pattern recognition (psychology); Programming language; Chemistry","score_opus":0.004266281759108959,"score_gpt":0.2284269188327291,"score_spread":0.22416063707362013,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2914005557","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.041982304,0.0013867981,0.8643226,0.00075200887,0.0005094671,0.00036032122,0.015538054,0.058059305,0.017089086],"genre_scores_gemma":[0.06485651,0.0022527,0.83389795,0.0005239121,0.00007822092,0.0015251167,0.022579923,0.037983004,0.03630266],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9992417,0.00005377195,0.000056461504,0.00023178928,0.0003202016,0.00009607486],"domain_scores_gemma":[0.9991817,0.00024414842,0.000112997725,0.00021540764,0.00016419699,0.000081480386],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015117755,0.0009322704,0.0008904795,0.001795544,0.0010579448,0.0017231576,0.0022261199,0.0011791678,0.033614464],"category_scores_gemma":[0.0010648906,0.0012764211,0.001199593,0.000983024,0.0005158795,0.0010402234,0.0010869519,0.0027528678,0.027998434],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002564626,0.000076011456,0.00052866415,0.00063121424,0.000054197404,0.00034618936,0.00027100253,0.0011485474,0.9261272,0.0039079837,0.021729587,0.044922918],"study_design_scores_gemma":[0.00006209575,0.0001179948,0.0033653392,0.00019625264,0.000063795866,0.0012388518,0.000073498086,0.017186219,0.81657517,0.0018022533,0.15920718,0.00011138267],"about_ca_topic_score_codex":0.0020111373,"about_ca_topic_score_gemma":0.0054372502,"teacher_disagreement_score":0.033614464,"about_ca_system_score_codex":0.0005999943,"about_ca_system_score_gemma":0.00076163496,"threshold_uncertainty_score":0.11245155},"labels":[],"label_agreement":null},{"id":"W2914184348","doi":"10.1093/bioinformatics/btz080","title":"ChimeraUGEM: unsupervised gene expression modeling in any given organism","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene Regulatory Network Analysis","field":"Biochemistry, Genetics and Molecular Biology","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Azrieli Foundation; Tel Aviv University; Ministry of Science, Technology and Space","keywords":"Organism; Chlamydomonas reinhardtii; Model organism; Gene; Computational biology; Source code; Computer science; Biology; Gene expression; Coding (social sciences); Software; Coding region; Genetics; Programming language","score_opus":0.010676590549476008,"score_gpt":0.21166432740557284,"score_spread":0.20098773685609683,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2914184348","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.026569389,0.00017464266,0.9544229,0.00025255303,0.0000389306,0.00004971722,0.0039268676,0.0130888475,0.0014762066],"genre_scores_gemma":[0.26492894,0.0004416776,0.7176815,0.0002738589,0.000056274006,0.00063420803,0.010117399,0.0026837932,0.0031823206],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9997873,0.000058369755,0.000008974909,0.00009493548,0.000036311034,0.0000141825385],"domain_scores_gemma":[0.9994623,0.00036404296,0.000041218536,0.00005512216,0.00004824407,0.000029093579],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006983135,0.0008396971,0.0006545219,0.0004948187,0.00046072426,0.000799432,0.0017244361,0.0010147073,0.0028718985],"category_scores_gemma":[0.0019288603,0.00045960286,0.0013322624,0.00063992897,0.00057537126,0.0005342176,0.00084054004,0.0012005692,0.0010966997],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001416551,0.00004739578,0.0026554773,0.00020125357,0.00013050264,0.00011061692,0.00011860597,0.942168,0.011199836,0.0061251316,0.0061970875,0.030904427],"study_design_scores_gemma":[0.000007704226,0.000011636528,0.00023850969,0.000004617933,0.000008059773,0.00002866745,0.000007501523,0.99242157,0.0019466269,0.0030499732,0.0022679518,0.000007233703],"about_ca_topic_score_codex":0.007858149,"about_ca_topic_score_gemma":0.008621186,"teacher_disagreement_score":0.007858149,"about_ca_system_score_codex":0.0007381651,"about_ca_system_score_gemma":0.0014458844,"threshold_uncertainty_score":0.015624821},"labels":[],"label_agreement":null},{"id":"W2914196064","doi":"10.1093/bioinformatics/btz113","title":"Identification of caveolin-1 domain signatures via machine learning and graphlet analysis of single-molecule super-resolution data","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Caveolin-1 and cellular processes","field":"Biochemistry, Genetics and Molecular Biology","cited_by":18,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia; Simon Fraser University","funders":"Natural Sciences and Engineering Research Council of Canada; Canadian Institutes of Health Research; China Scholarship Council; Canada Foundation for Innovation; Prostate Cancer Canada","keywords":"Caveolae; Pattern recognition (psychology); Computational biology; Computer science; Artificial intelligence; Chemistry; Biology; Biochemistry; Membrane","score_opus":0.008597539482588387,"score_gpt":0.2346000855288737,"score_spread":0.22600254604628534,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2914196064","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.41598406,0.00020720651,0.5810057,0.00021967998,0.000016605403,0.000057742065,0.0005454361,0.0012170662,0.0007465537],"genre_scores_gemma":[0.7452297,0.00020507119,0.25295767,0.000044767512,0.000021597076,0.00006645197,0.0009939849,0.00006688724,0.00041385714],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9997743,0.00005627474,0.0000152477505,0.00006798855,0.0000558013,0.000030398824],"domain_scores_gemma":[0.9990088,0.0004858878,0.00018101357,0.0001352574,0.00013921596,0.000049758608],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00061334297,0.00029699333,0.0003596197,0.0019593877,0.00015704539,0.00042154957,0.0003230716,0.00039000684,0.000620797],"category_scores_gemma":[0.0016885318,0.00014383969,0.00037509354,0.0009283959,0.00031818394,0.0005291896,0.00027269617,0.0005322686,0.00029451709],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005988253,0.0002787582,0.031431727,0.00032452372,0.00012577469,0.00026550586,0.00026376845,0.14375152,0.47047907,0.004903778,0.0017091781,0.34586757],"study_design_scores_gemma":[0.000004528393,0.000033286524,0.012746164,0.0000066508123,0.000009285814,0.0001229764,0.00002471859,0.94832724,0.036050253,0.002274849,0.00038676948,0.000013296203],"about_ca_topic_score_codex":0.0010366007,"about_ca_topic_score_gemma":0.0011998992,"teacher_disagreement_score":0.0019593877,"about_ca_system_score_codex":0.0004264005,"about_ca_system_score_gemma":0.00024247893,"threshold_uncertainty_score":0.0032437444},"labels":[],"label_agreement":null},{"id":"W2918710707","doi":"10.1093/bioinformatics/btz155","title":"Enhancing the prediction of disease–gene associations with multimodal deep learning","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":66,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Saskatchewan","funders":"Wuhan Institute of Technology; China Scholarship Council; Natural Sciences and Engineering Research Council of Canada; National Natural Science Foundation of China","keywords":"Computer science; Deep learning; Disease; Artificial intelligence; Computational biology; Machine learning; Biology; Medicine; Internal medicine","score_opus":0.004619670751991929,"score_gpt":0.19699379663438568,"score_spread":0.19237412588239375,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2918710707","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.26264575,0.0033649816,0.7181473,0.0020601472,0.00012960631,0.00010514383,0.0035998975,0.007240976,0.0027062104],"genre_scores_gemma":[0.88645333,0.0005510484,0.10640858,0.00058892363,0.00006386996,0.00011830376,0.0038339684,0.00009607915,0.0018859949],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99966836,0.000090431684,0.000017577404,0.0001015065,0.00007845423,0.000043751344],"domain_scores_gemma":[0.999218,0.00039586882,0.00009587323,0.00008994577,0.00013252819,0.00006770169],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012377948,0.0009686379,0.00063868787,0.0009386373,0.0002554566,0.0005922768,0.0011542982,0.000989091,0.002120539],"category_scores_gemma":[0.0034597914,0.0002846723,0.00069115084,0.00066309515,0.0003987299,0.00083900726,0.0011952525,0.0013004817,0.0006900274],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00081444293,0.00060838147,0.04892707,0.00035681282,0.0004227201,0.00047689787,0.00013665007,0.52740264,0.015473434,0.0033598577,0.012478764,0.38954234],"study_design_scores_gemma":[0.000015012794,0.000029428322,0.001518646,0.00001613599,0.000027183702,0.00004484738,0.000010348712,0.99209666,0.0022922223,0.0034639228,0.0004776315,0.000007954066],"about_ca_topic_score_codex":0.006223067,"about_ca_topic_score_gemma":0.0076492755,"teacher_disagreement_score":0.006223067,"about_ca_system_score_codex":0.0007423784,"about_ca_system_score_gemma":0.0008934912,"threshold_uncertainty_score":0.012373686},"labels":[],"label_agreement":null},{"id":"W2918882492","doi":"10.1093/bioinformatics/btz139","title":"SinNLRR: a robust subspace clustering method for cell type detection by non-negative and low-rank representation","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Single-cell and spatial transcriptomics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":167,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Saskatchewan","funders":"National Natural Science Foundation of China","keywords":"Cluster analysis; Computer science; Data mining; Rank (graph theory); Similarity (geometry); Spectral clustering; Representation (politics); Noise (video); Embedding; Artificial intelligence; Machine learning; Mathematics; Image (mathematics)","score_opus":0.012537531169978787,"score_gpt":0.2482468645510999,"score_spread":0.2357093333811211,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2918882492","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.002017687,0.00016931459,0.9953484,0.00009604943,0.000040261675,0.000040547817,0.00019065602,0.0016435413,0.00045352458],"genre_scores_gemma":[0.06410723,0.00034264554,0.92538625,0.00032459426,0.00011035167,0.00032727068,0.0027632168,0.0011023922,0.00553596],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9989191,0.00025875933,0.00006146987,0.0002748635,0.00039564454,0.00009023017],"domain_scores_gemma":[0.9988802,0.00028351886,0.00014065416,0.00017110324,0.0004390047,0.000085547006],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014608679,0.0011899673,0.0011456781,0.0013928039,0.00058551703,0.0010442188,0.0021724075,0.0013153458,0.004872198],"category_scores_gemma":[0.003198323,0.00044831497,0.0013122457,0.001263646,0.00069616496,0.0010282315,0.0014059125,0.0015607822,0.0037657002],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003493883,0.00012501146,0.001409576,0.00042713468,0.00020061349,0.00017783357,0.00024313323,0.23204394,0.035954144,0.013557164,0.031171635,0.6843404],"study_design_scores_gemma":[0.000021303016,0.0000335287,0.00034490094,0.000016192036,0.000011991028,0.00008586958,0.000029445047,0.9834722,0.0056858235,0.005549616,0.0047127246,0.000036465623],"about_ca_topic_score_codex":0.0058379928,"about_ca_topic_score_gemma":0.008180768,"teacher_disagreement_score":0.0058379928,"about_ca_system_score_codex":0.0006830081,"about_ca_system_score_gemma":0.0016934236,"threshold_uncertainty_score":0.016299129},"labels":[],"label_agreement":null},{"id":"W2920549681","doi":"10.1093/bioinformatics/btz151","title":"VHost-Classifier: virus-host classification using natural language processing","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bacteriophages and microbial interactions","field":"Environmental Science","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Classifier (UML); Computer science; Artificial intelligence; Python (programming language); Virus classification; Identifier; Host (biology); Biology; Programming language; Genetics; Genome","score_opus":0.015616746754287135,"score_gpt":0.2596760937011227,"score_spread":0.24405934694683556,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2920549681","genre_codex":"software","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.085539795,0.0031575372,0.33984938,0.0023950844,0.0016471194,0.0024503453,0.16698457,0.3763849,0.021591173],"genre_scores_gemma":[0.18110825,0.00082519604,0.5324616,0.0015821536,0.0004795843,0.0018758193,0.2605105,0.007755985,0.0134009225],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9976084,0.00035941682,0.00025571816,0.00081452506,0.0006744341,0.0002875314],"domain_scores_gemma":[0.9975611,0.001177748,0.00020057891,0.0002476854,0.0006667168,0.00014620203],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0025382505,0.0022442555,0.0011350639,0.0032805954,0.0011878285,0.0026397728,0.002497444,0.001758711,0.012959035],"category_scores_gemma":[0.0072266785,0.0005891717,0.0017720289,0.0020226047,0.00060107553,0.0023972313,0.0018115884,0.0017562505,0.016281227],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013133968,0.00077501644,0.023180854,0.002998029,0.00044625474,0.00055467756,0.00045280764,0.017179642,0.023625193,0.004400092,0.4967664,0.4283076],"study_design_scores_gemma":[0.00041013982,0.00066999736,0.015409259,0.00054479827,0.00021560355,0.0015498307,0.0006709702,0.6899943,0.06643678,0.025990168,0.19781286,0.00029533135],"about_ca_topic_score_codex":0.005110305,"about_ca_topic_score_gemma":0.007053004,"teacher_disagreement_score":0.012959035,"about_ca_system_score_codex":0.0013686436,"about_ca_system_score_gemma":0.002611902,"threshold_uncertainty_score":0.043352306},"labels":[],"label_agreement":null},{"id":"W2920946515","doi":"10.1093/bioinformatics/btz189","title":"Significance tests for analyzing gene expression data with small sample sizes","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Windsor","funders":"Australian Research Council Centre of Excellence for Mathematical and Statistical Frontiers; Natural Sciences and Engineering Research Council of Canada","keywords":"Sample size determination; Variance (accounting); Monte Carlo method; Statistical hypothesis testing; Computer science; Statistic; Test statistic; Degrees of freedom (physics and chemistry); Sample (material); Distribution (mathematics); Statistics; Mathematics; Expression (computer science); Scripting language; Algorithm","score_opus":0.04233453040245352,"score_gpt":0.281204900667655,"score_spread":0.2388703702652015,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2920946515","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.014933439,0.0012144585,0.9748052,0.00067013816,0.00044290855,0.0007947326,0.0013947233,0.0043905904,0.001353795],"genre_scores_gemma":[0.25501657,0.0004978617,0.7328365,0.0009893366,0.00048932387,0.005695308,0.0020726658,0.0015778398,0.000824562],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9389381,0.044640295,0.0025437528,0.005917861,0.0073408876,0.0006190717],"domain_scores_gemma":[0.5865061,0.387074,0.008260049,0.013549198,0.0034053859,0.0012053124],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.047133233,0.001774388,0.0032166126,0.004016637,0.0014300694,0.002289784,0.0051355115,0.003105192,0.009585674],"category_scores_gemma":[0.2837464,0.0010922165,0.0030651807,0.005043966,0.0056219343,0.0033614424,0.0028895715,0.0071675037,0.0017758126],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0038224747,0.00086284924,0.066203706,0.010219451,0.007996538,0.0040464867,0.0014827013,0.16603443,0.025666235,0.13791916,0.04323056,0.5325154],"study_design_scores_gemma":[0.0010616122,0.0019364395,0.024296874,0.0010658121,0.001045129,0.0019363047,0.00045201552,0.60453415,0.019673435,0.31706935,0.026694637,0.00023432104],"about_ca_topic_score_codex":0.0007281448,"about_ca_topic_score_gemma":0.0005737591,"teacher_disagreement_score":0.047133233,"about_ca_system_score_codex":0.0012833592,"about_ca_system_score_gemma":0.0023728001,"threshold_uncertainty_score":0.24926734},"labels":[],"label_agreement":null},{"id":"W2921591500","doi":"10.1093/bioinformatics/btz158","title":"Dr.VAE: improving drug response prediction via modeling of drug perturbation effects","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Computational Drug Discovery Methods","field":"Computer Science","cited_by":197,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Princess Margaret Cancer Centre; University Health Network; SickKids Foundation; Vector Institute; University of Toronto","funders":"University of Toronto; Canadian Cancer Society Research Institute; Canadian Institutes of Health Research; Hospital for Sick Children; Cancer Research Society","keywords":"Computer science; Autoencoder; Drug; Machine learning; Artificial intelligence; Deep learning; Medicine; Pharmacology","score_opus":0.0097157888356798,"score_gpt":0.24514708915096356,"score_spread":0.23543130031528375,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2921591500","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.055286083,0.0018565501,0.932111,0.0014935479,0.0001291202,0.000099601246,0.0014048873,0.004759558,0.0028596797],"genre_scores_gemma":[0.7099694,0.0009032912,0.28093684,0.0008810094,0.00010360823,0.00021147389,0.0017810368,0.00039358513,0.00481975],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99961084,0.00012759768,0.000019342166,0.00011748328,0.00009021665,0.00003457248],"domain_scores_gemma":[0.99917704,0.0005275815,0.00006951809,0.00008530135,0.00009754149,0.000042939322],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014119113,0.000703185,0.0008092656,0.00044033123,0.0001704848,0.00053942716,0.0011535371,0.000976704,0.002067353],"category_scores_gemma":[0.0035887433,0.0004329725,0.00079127174,0.0002937941,0.0004242423,0.0008710216,0.0008182922,0.0013684742,0.0006412578],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00016327464,0.000091407885,0.0024082921,0.00017073462,0.00015346428,0.00006775531,0.000031093798,0.890002,0.010434182,0.004996878,0.004568146,0.086912766],"study_design_scores_gemma":[0.000007990685,0.000028538889,0.00020060313,0.0000049361797,0.000010563257,0.000026640086,0.0000013281318,0.9948597,0.002404607,0.0018342935,0.0006143334,0.0000065046875],"about_ca_topic_score_codex":0.004246421,"about_ca_topic_score_gemma":0.0064272867,"teacher_disagreement_score":0.004246421,"about_ca_system_score_codex":0.000716681,"about_ca_system_score_gemma":0.0011207926,"threshold_uncertainty_score":0.008443415},"labels":[],"label_agreement":null},{"id":"W2922092653","doi":"10.1093/bioinformatics/btz175","title":"SodaPop: a forward simulation suite for the evolutionary dynamics of asexual populations on protein fitness landscapes","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Evolution and Genetic Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"Natural Sciences and Engineering Research Council of Canada; Université de Montréal","keywords":"Population; Evolutionary dynamics; Fitness landscape; Biology; Context (archaeology); Experimental evolution; Selection (genetic algorithm); Population size; Fixation (population genetics); Protein dynamics; Evolutionary biology; Computer science; Protein structure; Genetics; Artificial intelligence; Gene","score_opus":0.013455842597538683,"score_gpt":0.27168775451530003,"score_spread":0.25823191191776135,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2922092653","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.25274527,0.002277648,0.52397233,0.0025569412,0.0012098795,0.0017002617,0.048970748,0.11850071,0.048066285],"genre_scores_gemma":[0.44622272,0.0022277846,0.45637134,0.0013057562,0.00020028409,0.0056348094,0.056743093,0.018901322,0.012392847],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9996449,0.00013325895,0.00002885634,0.00004599749,0.00009560807,0.00005139188],"domain_scores_gemma":[0.99796426,0.00146423,0.00008455165,0.000101368554,0.00022633755,0.0001591761],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013456085,0.0018923762,0.0013022888,0.0008178689,0.0007063413,0.0011669502,0.0042279726,0.0019440923,0.016111685],"category_scores_gemma":[0.0059305434,0.00088763365,0.0015222381,0.00076412235,0.0006979386,0.0009382028,0.0014974623,0.0025661038,0.0021481449],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00049181585,0.00037359467,0.008153859,0.0018548284,0.0005736371,0.000819928,0.0004116092,0.85853153,0.005088513,0.028827177,0.06422458,0.030648848],"study_design_scores_gemma":[0.00028865738,0.000048912596,0.00055337214,0.00004911757,0.000033207773,0.000062055326,0.000032325428,0.97765756,0.001011966,0.005826685,0.014408425,0.0000277236],"about_ca_topic_score_codex":0.008985845,"about_ca_topic_score_gemma":0.0112636415,"teacher_disagreement_score":0.016111685,"about_ca_system_score_codex":0.0010259388,"about_ca_system_score_gemma":0.0025408028,"threshold_uncertainty_score":0.05389893},"labels":[],"label_agreement":null},{"id":"W2928574665","doi":"10.1093/bioinformatics/btz223","title":"FusionLearn: a biomarker selection algorithm on cross-platform data","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"York University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Computer science; Data mining; Microarray databases; Software; Microarray analysis techniques; Biomarker discovery; R package; Selection (genetic algorithm); Machine learning; Proteomics; Biology; Gene","score_opus":0.03584298378024424,"score_gpt":0.31129712829661516,"score_spread":0.2754541445163709,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2928574665","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.025616571,0.0010652666,0.94764453,0.0006011006,0.00017931873,0.00027144526,0.0023373365,0.0214257,0.00085880473],"genre_scores_gemma":[0.23188783,0.00037591686,0.7500656,0.00084076764,0.00017628666,0.0006445956,0.0120416675,0.0016630603,0.0023042571],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9977337,0.0006548999,0.00019364512,0.0007561431,0.0004916766,0.00016984454],"domain_scores_gemma":[0.9958936,0.0024993839,0.00027293895,0.00045460637,0.0007005245,0.0001789163],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0059417817,0.0020985694,0.0018977915,0.003042289,0.00084358954,0.0013112413,0.0025737344,0.0025855445,0.0033454308],"category_scores_gemma":[0.013205214,0.00079955504,0.0023473585,0.0027261402,0.0007172577,0.0018819654,0.0025304998,0.0024826864,0.002169923],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0018428684,0.00054746814,0.019014739,0.00058762945,0.001313817,0.00071326474,0.0002029095,0.23627473,0.01392708,0.0028472692,0.03926131,0.6834669],"study_design_scores_gemma":[0.00013409994,0.00017871308,0.0015057353,0.000043106666,0.00010658173,0.0001881673,0.000032614953,0.9795725,0.0056324224,0.008837423,0.0037378382,0.000030969226],"about_ca_topic_score_codex":0.0035386698,"about_ca_topic_score_gemma":0.0037125351,"teacher_disagreement_score":0.0059417817,"about_ca_system_score_codex":0.00096509827,"about_ca_system_score_gemma":0.0017195916,"threshold_uncertainty_score":0.03142351},"labels":[],"label_agreement":null},{"id":"W2935257869","doi":"10.1093/bioinformatics/btz229","title":"<i>RTNsurvival</i>: an R/Bioconductor package for regulatory network survival analysis","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":54,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canada's Michael Smith Genome Sciences Centre; BC Cancer Agency","funders":"National Cancer Institute; Cancer Research UK; National Institutes of Health; Conselho Nacional de Desenvolvimento Científico e Tecnológico; Coordenação de Aperfeiçoamento de Pessoal de Nível Superior; Breast Cancer Research Foundation","keywords":"Bioconductor; R package; Computer science; Survival analysis; Network analysis; Computational biology; Biology; Statistics; Mathematics; Genetics; Programming language; Gene","score_opus":0.013043309570168478,"score_gpt":0.24106370461805707,"score_spread":0.2280203950478886,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2935257869","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0033237166,0.0014957516,0.25677362,0.0014850431,0.0012031747,0.00073785876,0.31111336,0.41536698,0.008500484],"genre_scores_gemma":[0.038612593,0.001523055,0.3999478,0.0029381376,0.00050568907,0.010610289,0.32402,0.20615385,0.015688598],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9958507,0.0011156352,0.00038908148,0.001347006,0.0009579184,0.000339734],"domain_scores_gemma":[0.98915184,0.005871377,0.0011009771,0.0016171776,0.001713812,0.0005448348],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0063289604,0.0047370996,0.003974531,0.0043166857,0.0014259273,0.00442685,0.0070422995,0.0021817186,0.12691985],"category_scores_gemma":[0.029347895,0.002297666,0.0036437185,0.004745531,0.0016837168,0.0026874014,0.0042612953,0.0053631845,0.10199551],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006121218,0.000065263164,0.0031172868,0.0048146253,0.0011833099,0.00036801022,0.00041030042,0.007277779,0.0033675046,0.011471392,0.93663126,0.03068108],"study_design_scores_gemma":[0.0010997177,0.00021683614,0.008575611,0.0016069062,0.0012599303,0.0011282741,0.00023992626,0.11602672,0.01625629,0.08760606,0.76551044,0.00047331591],"about_ca_topic_score_codex":0.006198868,"about_ca_topic_score_gemma":0.0082880035,"teacher_disagreement_score":0.12691985,"about_ca_system_score_codex":0.0018000912,"about_ca_system_score_gemma":0.0056364406,"threshold_uncertainty_score":0.42458922},"labels":[],"label_agreement":null},{"id":"W2936760626","doi":"10.1093/bioinformatics/btz278","title":"ORCA: a comprehensive bioinformatics container environment for education and research","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetics, Bioinformatics, and Biomedical Research","field":"Biochemistry, Genetics and Molecular Biology","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Genome British Columbia","funders":"National Human Genome Research Institute; Genome British Columbia; Genome Canada","keywords":"License; Container (type theory); MIT License; Computer science; Suite; Debugging; Software; Source code; Software engineering; Operating system; World Wide Web; Bioinformatics; Biology; Engineering","score_opus":0.03086956639264518,"score_gpt":0.31952365313010106,"score_spread":0.28865408673745585,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2936760626","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0011541666,0.00041146035,0.16213165,0.0010844487,0.00044211847,0.0005135503,0.02810367,0.7787975,0.02736135],"genre_scores_gemma":[0.017764177,0.0011455311,0.37203273,0.0026410294,0.0003407057,0.002485932,0.18752012,0.37358218,0.04248762],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99632484,0.00061218644,0.00039983352,0.0007538968,0.0012623915,0.00064689165],"domain_scores_gemma":[0.99006605,0.00214026,0.0005828217,0.0031468084,0.0020799094,0.0019840882],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.006740146,0.0024109432,0.0021136599,0.0027672667,0.0016103097,0.005776115,0.008302616,0.002189091,0.18169168],"category_scores_gemma":[0.017232146,0.0025959439,0.0027736549,0.0032781574,0.0011007714,0.008815853,0.009825276,0.0059697353,0.20873454],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005908794,0.00013039664,0.00051619316,0.0006620635,0.00006283532,0.00013411888,0.00024947955,0.0011016511,0.004762761,0.008136204,0.9175426,0.066110894],"study_design_scores_gemma":[0.00044875813,0.00008184371,0.0013579389,0.0002671222,0.000045645644,0.00022145419,0.000067376815,0.007993477,0.0068875197,0.015105551,0.9673189,0.00020450521],"about_ca_topic_score_codex":0.0037684883,"about_ca_topic_score_gemma":0.0033167722,"teacher_disagreement_score":0.18169168,"about_ca_system_score_codex":0.0015968233,"about_ca_system_score_gemma":0.00480855,"threshold_uncertainty_score":0.6078192},"labels":[],"label_agreement":null},{"id":"W2938712064","doi":"10.1093/bioinformatics/btz266","title":"Discovering novel mutation signatures by latent Dirichlet allocation with variational Bayes inference","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Cancer Genomics and Diagnostics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Core Research for Evolutional Science and Technology; Institute of Genetics; Japan Society for the Promotion of Science; Waseda University; Ministry of Education, Culture, Sports, Science and Technology","keywords":"Mutation; Latent Dirichlet allocation; Computer science; Cluster analysis; Bayes' theorem; Inference; Mutation rate; Genetics; Computational biology; Biology; Bayesian probability; Artificial intelligence; Topic model; Gene","score_opus":0.0052565859999535594,"score_gpt":0.2151006418825204,"score_spread":0.20984405588256683,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2938712064","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.010215316,0.00032058396,0.9877551,0.00031928695,0.00003783006,0.000075984106,0.00024103525,0.0004797633,0.000555081],"genre_scores_gemma":[0.29006138,0.00063413946,0.7010685,0.0005615644,0.00023230934,0.00059835514,0.0026586906,0.00060659106,0.0035785222],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99679923,0.0014054178,0.00014804718,0.0008661385,0.00048813937,0.00029293195],"domain_scores_gemma":[0.9929717,0.00565851,0.00035070162,0.00030660877,0.00048550387,0.00022693195],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005547988,0.0016671941,0.0024940406,0.0027829993,0.001381201,0.0024303717,0.003721677,0.0026609697,0.0036815947],"category_scores_gemma":[0.013078428,0.0018856936,0.0036258725,0.0017316813,0.0019378149,0.0024691876,0.002500675,0.003722246,0.00088143314],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00027944957,0.0001493458,0.007061665,0.0003756541,0.00031347913,0.000270897,0.00036507484,0.8296967,0.0032823647,0.0546005,0.0058127483,0.09779206],"study_design_scores_gemma":[0.00001276881,0.0000050445756,0.00012222402,0.0000139659505,0.000013995844,0.000022306078,0.000012816561,0.978519,0.0003420024,0.02045075,0.00047248142,0.0000126500045],"about_ca_topic_score_codex":0.013524685,"about_ca_topic_score_gemma":0.013613593,"teacher_disagreement_score":0.013524685,"about_ca_system_score_codex":0.0021148599,"about_ca_system_score_gemma":0.0027680828,"threshold_uncertainty_score":0.029340923},"labels":[],"label_agreement":null},{"id":"W2944510853","doi":"10.1093/bioinformatics/btz308","title":"gpart: human genome partitioning and visualization of high-density SNP data by identifying haplotype blocks","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":49,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Public Health Ontario; University of Toronto; Hospital for Sick Children; Sinai Health System; Lunenfeld-Tanenbaum Research Institute","funders":"Canadian Statistical Sciences Institute; Canadian Institutes of Health Research; National Research Foundation of Korea; Neurosciences Research Foundation","keywords":"Bioconductor; Visualization; Computer science; Cluster analysis; Genome; Tag SNP; dbSNP; Linkage disequilibrium; Data mining; Single-nucleotide polymorphism; R package; Haplotype; Computational biology; SNP genotyping; Reference genome; DNA sequencing; Biology; Genetics; Gene; Artificial intelligence; Computational science","score_opus":0.03210594169097992,"score_gpt":0.3011335103091326,"score_spread":0.2690275686181527,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2944510853","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.012447837,0.0006525217,0.573946,0.000916734,0.0003425013,0.00031780644,0.08433652,0.32168004,0.0053600343],"genre_scores_gemma":[0.097001575,0.0010268291,0.74027663,0.000512901,0.00023439381,0.0019692709,0.08591663,0.06743119,0.0056306743],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9990441,0.00028076334,0.000071000955,0.00024472506,0.00027625845,0.000083114355],"domain_scores_gemma":[0.9982261,0.0009569512,0.00019830182,0.0002734762,0.00024058127,0.00010456031],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001765709,0.001996043,0.001362489,0.0029778967,0.0008794863,0.0025705085,0.002659095,0.0010028133,0.075858176],"category_scores_gemma":[0.007409275,0.001359327,0.0021731455,0.003124844,0.00039339685,0.0015759502,0.0032179768,0.0020231996,0.02042659],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00089688646,0.00012288269,0.0074091284,0.0023408777,0.00083839643,0.0010705198,0.0016062462,0.015167463,0.02297718,0.012784154,0.71833766,0.21644866],"study_design_scores_gemma":[0.0011125869,0.00029669947,0.025224578,0.0007079784,0.00051669264,0.0027441222,0.00051183463,0.4165875,0.04993832,0.0692345,0.4325314,0.00059380115],"about_ca_topic_score_codex":0.0030941868,"about_ca_topic_score_gemma":0.0034932934,"teacher_disagreement_score":0.075858176,"about_ca_system_score_codex":0.0003984487,"about_ca_system_score_gemma":0.0011371905,"threshold_uncertainty_score":0.2537709},"labels":[],"label_agreement":null},{"id":"W2945501761","doi":"10.1093/bioinformatics/btz413","title":"HyAsP, a greedy tool for plasmids identification","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":27,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Contig; Plasmid; Computer science; Genome; Greedy algorithm; Identification (biology); Computational biology; Biology; Genetics; Gene; Algorithm","score_opus":0.01003540163125125,"score_gpt":0.22860359757191284,"score_spread":0.2185681959406616,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2945501761","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.01509468,0.0015978793,0.51892155,0.0005391511,0.00034710063,0.00051306025,0.031589147,0.42780265,0.0035947366],"genre_scores_gemma":[0.04302298,0.0006286583,0.85218364,0.000650949,0.00012758674,0.0010077356,0.064662494,0.035290167,0.002425746],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9970009,0.0007184117,0.0003027522,0.0008974427,0.0008545453,0.00022603627],"domain_scores_gemma":[0.99423957,0.0034930254,0.00051351247,0.00094533886,0.00052215933,0.00028635637],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0042972392,0.0039036125,0.002299457,0.003587267,0.002262556,0.003475921,0.0036576772,0.0019761268,0.017525109],"category_scores_gemma":[0.016044302,0.0023379358,0.0026199191,0.0036447053,0.0011538957,0.0039119446,0.0046235938,0.0027615172,0.021877829],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0033225636,0.00062003225,0.020511296,0.0063872235,0.001324058,0.0013987814,0.0012615966,0.015817678,0.07696138,0.012993469,0.48882702,0.370575],"study_design_scores_gemma":[0.0019066731,0.000931761,0.015363061,0.0008668204,0.0006979746,0.004577113,0.0008575661,0.35517848,0.12041216,0.06371233,0.43473452,0.00076146197],"about_ca_topic_score_codex":0.0015377941,"about_ca_topic_score_gemma":0.0031484698,"teacher_disagreement_score":0.017525109,"about_ca_system_score_codex":0.00071083393,"about_ca_system_score_gemma":0.0026332778,"threshold_uncertainty_score":0.058627307},"labels":[],"label_agreement":null},{"id":"W2946813309","doi":"10.1093/bioinformatics/btz402","title":"YeastSpotter: accurate and parameter-free web segmentation for microscopy images of yeast cells","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Cell Image Analysis Techniques","field":"Biochemistry, Genetics and Molecular Biology","cited_by":109,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Genome Canada; University of Toronto","funders":"Canada Research Chairs","keywords":"Microscopy; Segmentation; Computer science; Artificial intelligence; Computer vision; Yeast; Image segmentation; Biological system; Computer graphics (images); Biology; Optics; Biochemistry; Physics","score_opus":0.006323364051794045,"score_gpt":0.26107417452359144,"score_spread":0.2547508104717974,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2946813309","genre_codex":"software","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.008426951,0.0009012219,0.42220628,0.00031543933,0.00039859337,0.00025101574,0.031113802,0.53309447,0.0032921939],"genre_scores_gemma":[0.055443063,0.0011136583,0.6627922,0.0006950981,0.00017037865,0.0018095439,0.09004126,0.17808233,0.009852516],"study_design_codex":"not_applicable","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9990988,0.00008347542,0.00007612188,0.000297216,0.00033223728,0.00011214842],"domain_scores_gemma":[0.9990237,0.00028434352,0.00009852944,0.00025336194,0.00021060908,0.00012952338],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013547022,0.0019681484,0.0019355236,0.0019316669,0.0011298519,0.002044421,0.0035291873,0.0016339996,0.05145341],"category_scores_gemma":[0.0036323487,0.0022822265,0.0015605437,0.0013059165,0.000547014,0.0017576906,0.0030241725,0.0026023635,0.046845116],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013552849,0.00013629354,0.0023573963,0.0029024163,0.00041194473,0.00062045775,0.0005043942,0.0064427415,0.18630499,0.0050908444,0.62060046,0.17327279],"study_design_scores_gemma":[0.00048720787,0.00019178649,0.007989291,0.00041176908,0.00014878229,0.0021190338,0.00018498303,0.10393728,0.3616786,0.01675056,0.5055666,0.00053417],"about_ca_topic_score_codex":0.0015966394,"about_ca_topic_score_gemma":0.0034337887,"teacher_disagreement_score":0.05145341,"about_ca_system_score_codex":0.0007353661,"about_ca_system_score_gemma":0.0010757288,"threshold_uncertainty_score":0.1721288},"labels":[],"label_agreement":null},{"id":"W2946841580","doi":"10.1093/bioinformatics/bty211","title":"Inference of the human polyadenylation code","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"RNA Research and Splicing","field":"Biochemistry, Genetics and Molecular Biology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ontario Genomics; University of Toronto","funders":"","keywords":"Polyadenylation; Cleavage and polyadenylation specificity factor; Untranslated region; Computational biology; Genetics; Biology; Three prime untranslated region; Gene; Human genome; Inference; Genome; Computer science; Artificial intelligence; Messenger RNA","score_opus":0.022838912652903266,"score_gpt":0.3096751317458032,"score_spread":0.2868362190928999,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2946841580","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.24242969,0.0013050346,0.72675896,0.0017833947,0.00018712877,0.00016146761,0.014102033,0.004384892,0.008887413],"genre_scores_gemma":[0.86409926,0.00049340527,0.11280365,0.00060344185,0.0001565138,0.00019351607,0.015682496,0.0003128827,0.0056548296],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99943537,0.000094377596,0.000019549365,0.00030502293,0.00008714003,0.00005858669],"domain_scores_gemma":[0.9972013,0.0021794715,0.00011984092,0.00016728269,0.0002423246,0.0000898785],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00071392185,0.00059347023,0.00041707937,0.0010363044,0.0003252922,0.00084601005,0.00083364965,0.0013141212,0.005127644],"category_scores_gemma":[0.005587625,0.0004413278,0.0007506687,0.0006995368,0.000630753,0.0007567919,0.00052824244,0.0015111816,0.0013942138],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00075704127,0.00018266543,0.03316171,0.00029678992,0.00010734937,0.000756451,0.00015979321,0.7518614,0.0071927253,0.021436078,0.016772542,0.16731536],"study_design_scores_gemma":[0.00001952964,0.000012239742,0.001306934,0.000020805577,0.000006977254,0.000073027586,0.000009525586,0.97815245,0.0014039675,0.01793503,0.0010519478,0.0000075240923],"about_ca_topic_score_codex":0.009041146,"about_ca_topic_score_gemma":0.007110449,"teacher_disagreement_score":0.009041146,"about_ca_system_score_codex":0.0011417026,"about_ca_system_score_gemma":0.0009882211,"threshold_uncertainty_score":0.017977059},"labels":[],"label_agreement":null},{"id":"W2949023387","doi":"10.1093/bioinformatics/btz456","title":"eFORGE v2.0: updated analysis of cell type-specific signal in epigenomic data","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":153,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"Wellcome Trust; Wellcome","keywords":"Epigenomics; Computer science; SIGNAL (programming language); Type (biology); Programming language; Biology; Genetics; DNA methylation","score_opus":0.011635717343155327,"score_gpt":0.22291294763631644,"score_spread":0.21127723029316112,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2949023387","genre_codex":"software","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.014747058,0.00077710743,0.3042792,0.00062178285,0.0004773523,0.00038310784,0.32476678,0.3472877,0.0066599757],"genre_scores_gemma":[0.042408485,0.0004691196,0.44265875,0.0008851578,0.00017063922,0.0016275421,0.41108057,0.09050931,0.0101904655],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99733007,0.0006476787,0.00028736866,0.00079549226,0.00065939187,0.0002800006],"domain_scores_gemma":[0.995018,0.0029080347,0.0003016954,0.0010188826,0.00054744823,0.00020600247],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008047514,0.0021661867,0.0014995929,0.0039218175,0.0009994635,0.0025604628,0.0029959308,0.001251089,0.056602903],"category_scores_gemma":[0.01914953,0.00206256,0.0022593345,0.0027823204,0.0005448047,0.0023065798,0.003220238,0.002256769,0.026568135],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0023989337,0.00016496271,0.02354348,0.002445082,0.0010994908,0.00090117834,0.00093329133,0.010610702,0.04501934,0.008884274,0.74725467,0.15674451],"study_design_scores_gemma":[0.00077054865,0.00021187076,0.033478156,0.0006597292,0.0006820495,0.0024336611,0.00018482341,0.079075746,0.06983434,0.022549754,0.78942204,0.00069721206],"about_ca_topic_score_codex":0.0036206432,"about_ca_topic_score_gemma":0.007948849,"teacher_disagreement_score":0.056602903,"about_ca_system_score_codex":0.00081362086,"about_ca_system_score_gemma":0.0020578338,"threshold_uncertainty_score":0.18935555},"labels":[],"label_agreement":null},{"id":"W2949176808","doi":"10.1093/bioinformatics/bty449","title":"Transfer learning for biomedical named entity recognition with neural networks","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Biomedical Text Mining and Ontologies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":195,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"National Human Genome Research Institute; National Institutes of Health; Nvidia","keywords":"Computer science; Named-entity recognition; Transfer of learning; Artificial intelligence; Annotation; Deep learning; Conditional random field; Task (project management); Artificial neural network; Field (mathematics); Machine learning; Natural language processing; Word (group theory); Noise (video); State (computer science); Domain (mathematical analysis); Mathematics","score_opus":0.019899863939700585,"score_gpt":0.25561859100615614,"score_spread":0.23571872706645555,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2949176808","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.09474929,0.006156334,0.86162174,0.0022077241,0.0006036463,0.00044251143,0.0026584584,0.024155933,0.007404307],"genre_scores_gemma":[0.7199887,0.0018604076,0.25570902,0.00088986964,0.00042583913,0.0006811883,0.011678108,0.00039530132,0.008371527],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9983621,0.000484906,0.00012121517,0.00051143556,0.00038368715,0.00013671443],"domain_scores_gemma":[0.9957212,0.0023895197,0.00035749053,0.0006655814,0.0007750975,0.00009103172],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003987598,0.0017970223,0.0010270448,0.0024225484,0.0006060783,0.0011241961,0.0026852577,0.0020536869,0.0043917275],"category_scores_gemma":[0.010426974,0.00042943744,0.0009836855,0.002679311,0.001009536,0.0037517084,0.002332265,0.002380607,0.0029152967],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00034129867,0.00030606147,0.0021953485,0.00036996728,0.0001844232,0.00028647124,0.00011725605,0.32673857,0.0055353777,0.004560261,0.013359896,0.64600503],"study_design_scores_gemma":[0.000014617402,0.00005804282,0.00060534687,0.00003047781,0.000024617342,0.000053187454,0.000040764957,0.97544724,0.007507289,0.013257396,0.002941881,0.000019125528],"about_ca_topic_score_codex":0.0057705673,"about_ca_topic_score_gemma":0.0043840325,"teacher_disagreement_score":0.0057705673,"about_ca_system_score_codex":0.0017958849,"about_ca_system_score_gemma":0.0011917595,"threshold_uncertainty_score":0.02108866},"labels":[],"label_agreement":null},{"id":"W2949679909","doi":"10.1093/bioinformatics/btz112","title":"SUBSTRA: Supervised Bayesian Patient Stratification","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Machine Learning in Healthcare","field":"Computer Science","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"BC Cancer Agency; Simon Fraser University","funders":"","keywords":"Computer science; Benchmark (surveying); Machine learning; Bayesian probability; Artificial intelligence; Population stratification; Source code; Data mining; Biology","score_opus":0.013433835080801506,"score_gpt":0.24442750401290844,"score_spread":0.23099366893210693,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2949679909","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.03253236,0.0010860699,0.95383817,0.002252953,0.00015901432,0.00061783387,0.0036823442,0.0035304346,0.0023007914],"genre_scores_gemma":[0.4466196,0.0006878274,0.53490543,0.001955995,0.00063154264,0.0012016011,0.009812184,0.0005415323,0.0036443442],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9963272,0.002402385,0.00016294222,0.0006408969,0.0003632199,0.00010330967],"domain_scores_gemma":[0.993196,0.0044160895,0.000545215,0.0008081485,0.00072258356,0.00031190526],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0065252776,0.0009339827,0.0012919958,0.0010146758,0.00060781697,0.00090659724,0.0016135451,0.0010288483,0.0042764377],"category_scores_gemma":[0.016684601,0.0005680759,0.0011130177,0.0008162355,0.0009989304,0.0008368682,0.0018257934,0.0018286931,0.0025758825],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0021208401,0.0005359141,0.049655158,0.000831633,0.000734556,0.00031880784,0.00047345835,0.1896176,0.006049939,0.026231246,0.055633508,0.6677974],"study_design_scores_gemma":[0.00036343042,0.00028728225,0.005408458,0.00012717862,0.00012898675,0.0004246214,0.000054015607,0.9109127,0.0033268183,0.06845812,0.010445626,0.00006275991],"about_ca_topic_score_codex":0.003134271,"about_ca_topic_score_gemma":0.0053275134,"teacher_disagreement_score":0.0065252776,"about_ca_system_score_codex":0.00069806003,"about_ca_system_score_gemma":0.0025899496,"threshold_uncertainty_score":0.03450936},"labels":[],"label_agreement":null},{"id":"W2949926253","doi":"10.1093/bioinformatics/bty722","title":"Adjutant: an R-based tool to support topic discovery for systematic and literature reviews","year":2018,"lang":"en","type":"review","venue":"Bioinformatics","topic":"Biomedical Text Mining and Ontologies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":18,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"BC Centre for Disease Control; University of British Columbia","funders":"Canadian Institutes of Health Research; Canada Research Chairs; Michael Smith Health Research BC","keywords":"Systematic review; Computer science; Data science; Information retrieval; MEDLINE; Biology","score_opus":0.04961544002561298,"score_gpt":0.3508484078449882,"score_spread":0.30123296781937525,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2949926253","genre_codex":"dataset","genre_gemma":"software","domain_codex":null,"domain_gemma":"methods","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":"methods","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0043477537,0.013094279,0.20039602,0.0037939865,0.001855098,0.0069285682,0.4873695,0.26679373,0.015421022],"genre_scores_gemma":[0.025056327,0.0076540816,0.7150134,0.0017840421,0.00060020504,0.031068781,0.16325451,0.04496195,0.010606734],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.98779374,0.0051862146,0.0027100309,0.0022176637,0.0017004691,0.0003919336],"domain_scores_gemma":[0.8898902,0.08118626,0.011132302,0.0065905494,0.009166508,0.0020341289],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.01793446,0.0036433118,0.0052747773,0.024029307,0.0016716914,0.0065428624,0.0042923386,0.0016030448,0.15821916],"category_scores_gemma":[0.11127992,0.0026981374,0.0077365413,0.021539148,0.0013111158,0.0041082627,0.007095552,0.0027058395,0.06761907],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0017607469,0.00008256655,0.004868162,0.15736997,0.0053734817,0.0013325934,0.0018905612,0.0030273912,0.008014574,0.015229859,0.65602815,0.14502195],"study_design_scores_gemma":[0.0019881204,0.00030076675,0.00749222,0.014674645,0.005549242,0.0019565546,0.00046376613,0.011742223,0.00853753,0.04060745,0.9060903,0.00059707894],"about_ca_topic_score_codex":0.0021377981,"about_ca_topic_score_gemma":0.006434379,"teacher_disagreement_score":0.98206556,"about_ca_system_score_codex":0.0022830965,"about_ca_system_score_gemma":0.015275668,"threshold_uncertainty_score":0.5292958},"labels":[],"label_agreement":null},{"id":"W2950021574","doi":"10.1093/bioinformatics/btz504","title":"Towards reliable named entity recognition in the biomedical domain","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Topic Modeling","field":"Computer Science","cited_by":88,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"National Human Genome Research Institute; Compute Canada; National Institutes of Health; Nvidia","keywords":"CRFS; Conditional random field; Computer science; Artificial intelligence; Dropout (neural networks); Transfer of learning; Named-entity recognition; Natural language processing; Machine learning; Task (project management); Regularization (linguistics); Deep learning; Sequence labeling; Source code; Generalization; Multi-task learning","score_opus":0.021713303600085557,"score_gpt":0.2395539452077731,"score_spread":0.21784064160768754,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2950021574","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.036250085,0.0063629868,0.876975,0.004993779,0.0007837351,0.00030144415,0.015919266,0.05203746,0.0063761743],"genre_scores_gemma":[0.22057045,0.0025695087,0.7038203,0.0014044945,0.0005146472,0.0003308952,0.063585676,0.0018262522,0.005377881],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9939559,0.0023842633,0.0004464016,0.0016652675,0.0012942689,0.00025403078],"domain_scores_gemma":[0.97684413,0.011695658,0.0016138775,0.0050251214,0.0041815387,0.0006396629],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.012952396,0.002213101,0.001525546,0.004336638,0.000815239,0.0029950151,0.0030900503,0.003548447,0.0060580876],"category_scores_gemma":[0.035673484,0.00079220725,0.0014603928,0.0038502007,0.0012622795,0.009502488,0.0047160834,0.0037406995,0.011964508],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011951517,0.0004039822,0.00914653,0.0029715921,0.00038750563,0.00076299737,0.00062363,0.11617828,0.034305066,0.02638712,0.13595565,0.6716825],"study_design_scores_gemma":[0.00008797746,0.00018678725,0.0035900967,0.00038959555,0.00013646789,0.0006065902,0.00038211732,0.8132056,0.06442299,0.049370233,0.06749763,0.00012385815],"about_ca_topic_score_codex":0.0031115648,"about_ca_topic_score_gemma":0.0026875762,"teacher_disagreement_score":0.012952396,"about_ca_system_score_codex":0.00095546566,"about_ca_system_score_gemma":0.0023344816,"threshold_uncertainty_score":0.068499684},"labels":[],"label_agreement":null},{"id":"W2950143580","doi":"10.1093/bioinformatics/btz473","title":"DepthFinder: a tool to determine the optimal read depth for reduced-representation sequencing","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic and phenotypic traits in livestock","field":"Biochemistry, Genetics and Molecular Biology","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Guelph; Université Laval","funders":"Grain Farmers of Ontario; Canadian Field Crop Research Alliance; Genome Canada; Syngenta Canada; Government of Canada; Saskatchewan Pulse Growers","keywords":"Genotyping; DNA sequencing; Computational biology; Identification (biology); Biology; Genome; Selection (genetic algorithm); Computer science; Genetics; Machine learning; DNA; Genotype; Gene; Ecology","score_opus":0.02702048160416613,"score_gpt":0.27424380277383104,"score_spread":0.24722332116966492,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2950143580","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.028753852,0.0013869314,0.86762375,0.0002891924,0.00014420022,0.00046279255,0.010886606,0.088419616,0.0020330062],"genre_scores_gemma":[0.057390608,0.00034358588,0.9289684,0.00019243437,0.000032019558,0.00091512915,0.006430092,0.0046110405,0.0011166484],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9972683,0.000614141,0.0002725541,0.00072514726,0.00093642104,0.00018335956],"domain_scores_gemma":[0.9934376,0.0041788835,0.0010089707,0.0004248405,0.0007724222,0.00017724434],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.006094006,0.0029376026,0.0018092868,0.004274832,0.0010804402,0.0022120494,0.003019311,0.0018859106,0.009660002],"category_scores_gemma":[0.017124142,0.0014812894,0.0014938655,0.0026300587,0.0006560655,0.0022365958,0.0019578824,0.002397374,0.004745092],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0026753652,0.0005419641,0.033057746,0.0047032107,0.00071337324,0.00074606884,0.0015771596,0.07314649,0.18996596,0.014746895,0.07422902,0.6038968],"study_design_scores_gemma":[0.00037420803,0.0006592107,0.011471254,0.0004618905,0.00022640472,0.00077924033,0.00024427587,0.63480955,0.26157606,0.015604178,0.073246576,0.0005471635],"about_ca_topic_score_codex":0.0025593562,"about_ca_topic_score_gemma":0.0033902556,"teacher_disagreement_score":0.009660002,"about_ca_system_score_codex":0.0012334982,"about_ca_system_score_gemma":0.0023286517,"threshold_uncertainty_score":0.03231591},"labels":[],"label_agreement":null},{"id":"W2950511618","doi":"10.1093/bioinformatics/bty387","title":"Indexed variation graphs for efficient and accurate resistome profiling","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":84,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Science and Technology Facilities Council; Department for Business, Energy and Industrial Strategy, UK Government; Public Health Agency of Canada; World Health Organization","keywords":"Resistome; Metagenomics; Profiling (computer programming); Computer science; Workflow; Data mining; Computational biology; Biology; Bioinformatics; Genome; Gene; Genetics; Database","score_opus":0.015061740084855505,"score_gpt":0.24947379069523234,"score_spread":0.23441205061037684,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2950511618","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.022407772,0.00046907825,0.95998687,0.00019810889,0.00006354863,0.00011050496,0.0028366437,0.011713031,0.0022145351],"genre_scores_gemma":[0.2017096,0.00054373563,0.7810051,0.00024628424,0.00008089088,0.00020339727,0.010991396,0.0026498083,0.0025697695],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99894196,0.00020288667,0.000059804694,0.00024986255,0.00044356022,0.00010191922],"domain_scores_gemma":[0.9980883,0.0007161215,0.00024300552,0.0004477239,0.00041137994,0.00009337157],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006054063,0.0009813359,0.00079475576,0.0032123115,0.00064085575,0.0014608379,0.0013421732,0.00081608264,0.0037309423],"category_scores_gemma":[0.0052211457,0.0005196968,0.0011658036,0.003366667,0.0005534785,0.0019797243,0.0013052113,0.0012625747,0.001998035],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005349218,0.00025833157,0.010591112,0.0011382308,0.0002664765,0.00083869946,0.00067933946,0.21242584,0.11947888,0.05509308,0.030883912,0.56781113],"study_design_scores_gemma":[0.00003707545,0.00010293247,0.0026473117,0.00006973732,0.00004949712,0.0004238739,0.00016470003,0.8740317,0.02515978,0.07073129,0.026503487,0.00007856384],"about_ca_topic_score_codex":0.0036688538,"about_ca_topic_score_gemma":0.0056991293,"teacher_disagreement_score":0.0037309423,"about_ca_system_score_codex":0.00083336094,"about_ca_system_score_gemma":0.00085378124,"threshold_uncertainty_score":0.012481272},"labels":[],"label_agreement":null},{"id":"W2950522632","doi":"10.1093/bioinformatics/btz518","title":"Bayesian GWAS with Structured and Non-Local Priors","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Institute on Aging; National Institute of Biomedical Imaging and Bioengineering; Canadian Institutes of Health Research; National Institutes of Health; Genentech; Takeda Pharmaceutical Company; IXICO; National Institute of General Medical Sciences; H. Lundbeck A/S; Servier; Eisai; Northern California Institute for Research and Education; Pfizer; Biogen; BioClinica; Roche; University of Southern California; F. Hoffmann-La Roche; Novartis Pharmaceuticals Corporation; U.S. Department of Defense; Eli Lilly and Company; Bristol-Myers Squibb; Merck; Alzheimer's Association; Foundation for the National Institutes of Health; Alzheimer's Disease Neuroimaging Initiative; Meso Scale Diagnostics","keywords":"Prior probability; Bayesian probability; Computer science; Genome-wide association study; Statistical power; Covariate; Machine learning; Artificial intelligence; Statistics; Mathematics; Single-nucleotide polymorphism; Biology","score_opus":0.0036576116719586166,"score_gpt":0.2119490812847971,"score_spread":0.20829146961283848,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2950522632","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.003929501,0.00030933766,0.99295396,0.0007907162,0.000026873016,0.000049053015,0.00046655102,0.00053006824,0.00094396115],"genre_scores_gemma":[0.16334818,0.0012739879,0.82868814,0.0010071003,0.00025224543,0.0005141558,0.001661028,0.00039585648,0.0028592693],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9928347,0.004877602,0.00028130645,0.0009876514,0.0008517676,0.0001669842],"domain_scores_gemma":[0.97673666,0.017948305,0.0013430582,0.0022540824,0.0013199568,0.0003979052],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01549988,0.00081174914,0.0016131257,0.0014506046,0.0007398366,0.0024685455,0.002481015,0.0018702816,0.0062625776],"category_scores_gemma":[0.055481397,0.0010945528,0.0015098836,0.0031497946,0.002064242,0.0027398695,0.0021528401,0.0031564867,0.002132697],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0004991318,0.00011031245,0.021079443,0.00077941426,0.0006309401,0.0005979939,0.0006334141,0.23772404,0.003945159,0.5241084,0.016863648,0.19302805],"study_design_scores_gemma":[0.00017327053,0.000057790152,0.0036254267,0.00020533729,0.00017229276,0.00052808336,0.000061838335,0.365358,0.0011875039,0.61485404,0.013684312,0.00009202996],"about_ca_topic_score_codex":0.008466853,"about_ca_topic_score_gemma":0.010083939,"teacher_disagreement_score":0.01549988,"about_ca_system_score_codex":0.0013117497,"about_ca_system_score_gemma":0.0023188728,"threshold_uncertainty_score":0.08197218},"labels":[],"label_agreement":null},{"id":"W2950655801","doi":"10.1093/bioinformatics/bty717","title":"C3D: a tool to predict 3D genomic interactions between cis-regulatory elements","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":15,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Princess Margaret Cancer Centre; University of Toronto; Ontario Institute for Cancer Research; University of Waterloo; University Health Network","funders":"Canadian Institutes of Health Research","keywords":"Chromatin; Source code; Computational biology; Computer science; Promoter; License; MIT License; Genome; DNA; R package; Gene; Biology; Data mining; Genetics; Programming language; Gene expression; Operating system","score_opus":0.008380410046805961,"score_gpt":0.24221554847552818,"score_spread":0.23383513842872222,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2950655801","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.022297619,0.0010973318,0.47787222,0.0005690271,0.0003637757,0.0004583398,0.1698088,0.32051843,0.007014462],"genre_scores_gemma":[0.10002136,0.0011925709,0.5409037,0.0008105538,0.00010983826,0.0025751707,0.29034021,0.0587668,0.005279731],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9989722,0.00016046224,0.000057107183,0.00035752222,0.00035345435,0.00009927475],"domain_scores_gemma":[0.99808866,0.0012600684,0.00011899982,0.00021184333,0.00018468776,0.00013577005],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0018537705,0.0044047358,0.002336226,0.0027329128,0.0017024899,0.0025757698,0.00497881,0.0020085196,0.03919894],"category_scores_gemma":[0.006652438,0.0019393427,0.003853648,0.0021167498,0.00089712697,0.0014690554,0.0031193767,0.0030459824,0.014698122],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001748334,0.00046915474,0.02656946,0.0061141187,0.0019014073,0.0015738,0.0009236219,0.14280398,0.042040218,0.024942845,0.6109722,0.13994098],"study_design_scores_gemma":[0.00072758243,0.00020915963,0.007599463,0.00035396795,0.00045424257,0.0011191553,0.00022498223,0.68921196,0.039682195,0.032610465,0.22748034,0.00032657015],"about_ca_topic_score_codex":0.011661945,"about_ca_topic_score_gemma":0.016765717,"teacher_disagreement_score":0.03919894,"about_ca_system_score_codex":0.0011488354,"about_ca_system_score_gemma":0.002986734,"threshold_uncertainty_score":0.1311335},"labels":[],"label_agreement":null},{"id":"W2950705667","doi":"10.1093/bioinformatics/bty549","title":"SATurn: a modular bioinformatics framework for the design of robust maintainable web-based and standalone applications","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetics, Bioinformatics, and Biomedical Research","field":"Biochemistry, Genetics and Molecular Biology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Innovative Medicines Initiative; Ontario Ministry of Research, Innovation and Science; Canada Foundation for Innovation; Kennedy Trust for Rheumatology Research; Wellcome Trust; Ontario Genomics Institute; AbbVie; European Federation of Pharmaceutical Industries and Associations; Boehringer Ingelheim","keywords":"Modular design; Computer science; Saturn; Web application; Open source; Simple (philosophy); World Wide Web; Data science; Software engineering; Software; Programming language","score_opus":0.025869866801164123,"score_gpt":0.2816519004378844,"score_spread":0.2557820336367203,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2950705667","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0041994816,0.00014521123,0.8189436,0.00022211991,0.00006681204,0.00036255707,0.0020671298,0.17109297,0.002900075],"genre_scores_gemma":[0.096539125,0.000527542,0.81848997,0.000593593,0.00009034851,0.0015250792,0.027337013,0.045308344,0.009588902],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9979482,0.00031890935,0.0002056659,0.00050929235,0.0007531592,0.00026476564],"domain_scores_gemma":[0.99761045,0.00065897056,0.00019112304,0.0005621944,0.0005854514,0.00039179865],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005007934,0.0014265201,0.0008900067,0.0013178494,0.0010007822,0.0028158927,0.0039019533,0.0010394254,0.010945104],"category_scores_gemma":[0.006662241,0.0015652612,0.0020531046,0.000953246,0.0014786666,0.0032172133,0.0048100655,0.0030628198,0.009385503],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0025327113,0.00064245774,0.007464404,0.002610923,0.00070431194,0.0013422308,0.0017663216,0.10880746,0.12902443,0.10832269,0.3055912,0.33119088],"study_design_scores_gemma":[0.0005148269,0.0003519438,0.0032007084,0.00030357583,0.00014972902,0.0006378263,0.00026650119,0.55840015,0.07329283,0.061326467,0.3012186,0.00033681712],"about_ca_topic_score_codex":0.0046180603,"about_ca_topic_score_gemma":0.0049042394,"teacher_disagreement_score":0.010945104,"about_ca_system_score_codex":0.0014851827,"about_ca_system_score_gemma":0.0033400564,"threshold_uncertainty_score":0.036615074},"labels":[],"label_agreement":null},{"id":"W2950983802","doi":"10.1093/bioinformatics/btz095","title":"Dhaka: variational autoencoder for unmasking tumor heterogeneity from single cell genomic data","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Single-cell and spatial transcriptomics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":61,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia; BC Cancer Agency","funders":"National Institute of General Medical Sciences; National Institutes of Health; Microsoft Research","keywords":"Autoencoder; Computer science; Computational biology; Dimensionality reduction; Feature (linguistics); Single cell sequencing; Biology; Genomics; Gene; Artificial intelligence; Mutation; Genetics; Genome; Exome sequencing; Deep learning","score_opus":0.03147508797227019,"score_gpt":0.2393536263880751,"score_spread":0.20787853841580492,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2950983802","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.009116348,0.0005798133,0.98635,0.00032296567,0.00007439468,0.000063820815,0.0008040194,0.0020636495,0.00062505255],"genre_scores_gemma":[0.24283378,0.0006339029,0.7423587,0.0006194814,0.00008068138,0.00053043047,0.0059327865,0.0009143126,0.0060959114],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9996722,0.000107176034,0.000018280049,0.00011073087,0.00006376824,0.000027886084],"domain_scores_gemma":[0.9991141,0.00059903634,0.00004254539,0.00009241994,0.00011451533,0.00003745584],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011650904,0.00096057483,0.0007020593,0.00052897603,0.00033984258,0.00076482457,0.0015008415,0.0012328875,0.0023990371],"category_scores_gemma":[0.0039984984,0.0007201678,0.0009367883,0.0004768887,0.0006252436,0.00073727645,0.0013901184,0.0024384765,0.0011822945],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00023557765,0.000070408976,0.0020700684,0.0003561648,0.00028846515,0.0001692403,0.00014987784,0.75002176,0.01950861,0.009914444,0.012219988,0.20499541],"study_design_scores_gemma":[0.00001005565,0.00001312445,0.00021387881,0.000009340791,0.000008689139,0.000023416325,0.000007432313,0.9922387,0.001825838,0.004272451,0.0013680106,0.0000090671865],"about_ca_topic_score_codex":0.008673755,"about_ca_topic_score_gemma":0.013413595,"teacher_disagreement_score":0.008673755,"about_ca_system_score_codex":0.00067244866,"about_ca_system_score_gemma":0.0012644785,"threshold_uncertainty_score":0.017246544},"labels":[],"label_agreement":null},{"id":"W2951158909","doi":"10.1093/bioinformatics/btw777","title":"Scater: pre-processing, quality control, normalization and visualization of single-cell RNA-seq data in R","year":2016,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Single-cell and spatial transcriptomics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":2053,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Health and Medical Research Council; Medical Research Council Canada; Cancer Research UK; Medical Research Council; European Molecular Biology Laboratory","keywords":"Normalization (sociology); Visualization; Computer science; RNA-Seq; Software; Computational biology; Data mining; Biology; Transcriptome; Genetics; Gene expression; Programming language; Gene","score_opus":0.028251786291285042,"score_gpt":0.27560352596503956,"score_spread":0.2473517396737545,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2951158909","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0024365904,0.00064404,0.48822713,0.00067814294,0.00042453533,0.0005747952,0.052910786,0.45019194,0.003912014],"genre_scores_gemma":[0.021556059,0.0010302552,0.68685395,0.0010457796,0.00019872672,0.0049501914,0.062341146,0.21523347,0.0067903306],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9899714,0.0041479217,0.001054319,0.002048617,0.0022664135,0.0005112851],"domain_scores_gemma":[0.97373044,0.015457373,0.002410709,0.004150456,0.0033875713,0.0008634308],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.017276933,0.005536002,0.0039800312,0.003879452,0.0013716043,0.0050437967,0.006485344,0.0016119551,0.07292074],"category_scores_gemma":[0.04427633,0.0033658368,0.004440603,0.00402553,0.0022862563,0.0033507051,0.004754945,0.0064676586,0.10014811],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014295419,0.00012165401,0.0041535166,0.0075848405,0.0015670087,0.0008429956,0.0011955387,0.012501901,0.024015578,0.015666556,0.8387011,0.09221973],"study_design_scores_gemma":[0.001333371,0.00044531983,0.010327916,0.0017437611,0.0009644742,0.0018995363,0.000367484,0.109121345,0.06892585,0.07460879,0.7292371,0.0010250779],"about_ca_topic_score_codex":0.0032481393,"about_ca_topic_score_gemma":0.003884663,"teacher_disagreement_score":0.07292074,"about_ca_system_score_codex":0.001605492,"about_ca_system_score_gemma":0.0055068997,"threshold_uncertainty_score":0.24394417},"labels":[],"label_agreement":null},{"id":"W2951162601","doi":"10.1093/bioinformatics/btz476","title":"Mycorrhiza: genotype assignment using phylogenetic networks","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Mycorrhizal Fungi and Plant Interactions","field":"Agricultural and Biological Sciences","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université Laval; University of British Columbia; McGill University","funders":"","keywords":"Phylogenetic tree; Genotype; Biology; Phylogenetics; Genetics; Computational biology; Computer science; Evolutionary biology; Gene","score_opus":0.014572050027027907,"score_gpt":0.20421164200267822,"score_spread":0.1896395919756503,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2951162601","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.04478374,0.0011630313,0.9433451,0.0006384145,0.00007892458,0.00015443636,0.003729618,0.004130914,0.001975878],"genre_scores_gemma":[0.32597655,0.0008448022,0.66073084,0.00025936362,0.00019330227,0.00027801198,0.008746109,0.0004424626,0.002528551],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99855715,0.0004423476,0.000053617066,0.00063187076,0.0002605658,0.000054499604],"domain_scores_gemma":[0.99760216,0.0013473833,0.0004521112,0.00024550056,0.00023151824,0.000121237535],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00237696,0.0009115579,0.00068324676,0.00186395,0.00063113397,0.0014754295,0.0015266684,0.0010754623,0.0022619457],"category_scores_gemma":[0.008499283,0.00044146265,0.000593211,0.0017436692,0.00061764463,0.0018567408,0.0011491866,0.0014564514,0.0016522871],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0004263545,0.0002889507,0.035923887,0.0011171927,0.000344768,0.00029666064,0.00042929582,0.39788732,0.022039173,0.019227093,0.016945062,0.50507426],"study_design_scores_gemma":[0.000027757962,0.000052320393,0.0047944565,0.00008812751,0.000036786863,0.00023803623,0.000043625325,0.943487,0.0035389394,0.041670732,0.0059908363,0.000031392043],"about_ca_topic_score_codex":0.0023809774,"about_ca_topic_score_gemma":0.0028764163,"teacher_disagreement_score":0.0023809774,"about_ca_system_score_codex":0.00067996373,"about_ca_system_score_gemma":0.0006781753,"threshold_uncertainty_score":0.012570739},"labels":[],"label_agreement":null},{"id":"W2951421219","doi":"10.1093/bioinformatics/btz150","title":"RECAP reveals the true statistical significance of ChIP-seq peak calls","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ottawa Hospital; University of Ottawa","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"False discovery rate; Computer science; Resampling; Statistical hypothesis testing; Chip; Statistical model; Data mining; Statistics; Biology; Mathematics; Algorithm; Machine learning; Genetics","score_opus":0.00652173639743256,"score_gpt":0.22257576152622724,"score_spread":0.21605402512879468,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2951421219","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.08388844,0.00068251695,0.86883354,0.00057152845,0.00019041132,0.00017686984,0.0050665555,0.038773745,0.001816364],"genre_scores_gemma":[0.50747865,0.0003113353,0.47426045,0.0010708435,0.0001349895,0.0005577536,0.0086707305,0.0059749642,0.0015402995],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9884703,0.0036514294,0.0008224374,0.0032342342,0.0033805368,0.00044098063],"domain_scores_gemma":[0.9218411,0.05469372,0.0044334554,0.0117855,0.0064040488,0.00084218493],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.025702726,0.0019959575,0.0021443835,0.0025171954,0.0009243406,0.0024450133,0.003926714,0.0026296473,0.0048562377],"category_scores_gemma":[0.0838396,0.0011061281,0.0017387667,0.0016272306,0.0018907432,0.002106137,0.0022714895,0.0027696919,0.002949557],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0044938605,0.0006112667,0.15048847,0.0040866,0.002529147,0.0018892419,0.0010783095,0.093793735,0.29471624,0.017655829,0.04460096,0.38405642],"study_design_scores_gemma":[0.00024708852,0.0009845524,0.0656149,0.0002908911,0.0007824894,0.0031064325,0.00031300695,0.49537233,0.39453444,0.021559874,0.016758543,0.00043552803],"about_ca_topic_score_codex":0.0009342974,"about_ca_topic_score_gemma":0.0007261311,"teacher_disagreement_score":0.025702726,"about_ca_system_score_codex":0.0010150161,"about_ca_system_score_gemma":0.001434775,"threshold_uncertainty_score":0.13593066},"labels":[],"label_agreement":null},{"id":"W2951662798","doi":"10.1093/bioinformatics/btz433","title":"CoCo: RNA-seq read assignment correction for nested genes and multimapped reads","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"RNA Research and Splicing","field":"Biochemistry, Genetics and Molecular Biology","cited_by":60,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Sherbrooke","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Coco; Transcriptome; RNA-Seq; Python (programming language); Gene; Computational biology; Computer science; R package; Biology; Genetics; Gene expression; Programming language; Artificial intelligence","score_opus":0.01459678843746658,"score_gpt":0.2730451811645093,"score_spread":0.25844839272704273,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2951662798","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.026443718,0.0011176325,0.67320293,0.00049879926,0.0014366871,0.00090788136,0.051528905,0.23828383,0.006579704],"genre_scores_gemma":[0.086319715,0.0003901286,0.73818636,0.0014152888,0.0003210252,0.002957306,0.07838755,0.08185047,0.01017218],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9954318,0.00069235236,0.00034520277,0.0017731555,0.001429147,0.00032838955],"domain_scores_gemma":[0.99218905,0.0027870687,0.0010663414,0.0016956876,0.0019260781,0.00033580122],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0069627967,0.0025467314,0.0014356428,0.0025117525,0.0020711422,0.0027461553,0.0029625248,0.0011738681,0.021422282],"category_scores_gemma":[0.017902482,0.001185609,0.0016673235,0.0023099454,0.0012263317,0.00162702,0.0026944408,0.0024969154,0.010166716],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0024721671,0.00019710088,0.03786741,0.0052274,0.0012362945,0.00084377464,0.0014613556,0.0077011962,0.16628367,0.012036069,0.4759698,0.2887037],"study_design_scores_gemma":[0.000422375,0.00030152695,0.029269636,0.0005675804,0.0005017672,0.0010626084,0.0003979207,0.17111486,0.3946613,0.0120807355,0.38904354,0.00057609065],"about_ca_topic_score_codex":0.007379358,"about_ca_topic_score_gemma":0.015487978,"teacher_disagreement_score":0.021422282,"about_ca_system_score_codex":0.0014543951,"about_ca_system_score_gemma":0.003971834,"threshold_uncertainty_score":0.07166475},"labels":[],"label_agreement":null},{"id":"W2952106966","doi":"10.1093/bioinformatics/btz064","title":"Variational infinite heterogeneous mixture model for semi-supervised clustering of heart enhancers","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"Ontario Ministry of Research and Innovation; Natural Sciences and Engineering Research Council of Canada; Canada Foundation for Innovation","keywords":"Enhancer; Cluster analysis; Computer science; Python (programming language); Inference; Context (archaeology); Mixture model; Dirichlet process; Machine learning; Data mining; Artificial intelligence; Biology; Gene; Genetics; Gene expression","score_opus":0.015514781661221754,"score_gpt":0.2525756993871091,"score_spread":0.23706091772588736,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2952106966","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.008061876,0.00033320743,0.989489,0.0002846383,0.000036301382,0.00007183391,0.00040115786,0.00077807356,0.000543941],"genre_scores_gemma":[0.38948753,0.0005663472,0.58964956,0.0008901794,0.00028557712,0.0011308579,0.006982702,0.0011462009,0.009860965],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99790996,0.0009342788,0.00010121138,0.0005868018,0.0002764479,0.0001912769],"domain_scores_gemma":[0.9939155,0.004395474,0.0003617089,0.00033048843,0.0007794738,0.00021730897],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0053565544,0.0013590417,0.002493636,0.0017086425,0.00093259825,0.0016482484,0.005338149,0.0029089996,0.003951802],"category_scores_gemma":[0.0105589945,0.0014626915,0.0027579623,0.0012613974,0.0023359593,0.0015342232,0.0023894757,0.0035838867,0.0016562097],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00018241964,0.00006221729,0.0015862556,0.00015984611,0.00012774987,0.00008015636,0.00017733422,0.93404305,0.0013546048,0.024157105,0.0032577384,0.03481146],"study_design_scores_gemma":[0.000007891807,0.0000042119755,0.00007670892,0.0000069088615,0.0000044841718,0.00000654101,0.0000040361915,0.99350005,0.00014911625,0.0059890375,0.00024507442,0.0000058458368],"about_ca_topic_score_codex":0.013908905,"about_ca_topic_score_gemma":0.018237224,"teacher_disagreement_score":0.013908905,"about_ca_system_score_codex":0.0024978088,"about_ca_system_score_gemma":0.0021287096,"threshold_uncertainty_score":0.028328478},"labels":[],"label_agreement":null},{"id":"W2952191365","doi":"10.1093/bioinformatics/btz485","title":"SIGN: similarity identification in gene expression","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Vector Institute; Montreal Clinical Research Institute; Institute of Cancer Research; Ontario Institute for Cancer Research; Institut universitaire de cardiologie et de pneumologie de Québec; Université de Montréal; Université Laval; Princess Margaret Cancer Centre; University of Toronto","funders":"Natural Sciences and Engineering Research Council of Canada; Ontario Institute for Cancer Research; Canadian Institutes of Health Research; Government of Ontario; Cancer Research Society","keywords":"Similarity (geometry); Identification (biology); Expression (computer science); Computer science; Computational biology; Breast cancer; Sign (mathematics); Artificial intelligence; Machine learning; Data mining; Bioinformatics; Biology; Cancer; Genetics; Mathematics","score_opus":0.009305137900203311,"score_gpt":0.22990952976610174,"score_spread":0.22060439186589842,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2952191365","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.049996905,0.00092211127,0.88235325,0.0008962503,0.00025684488,0.00044333853,0.028703732,0.03128991,0.0051376238],"genre_scores_gemma":[0.39423528,0.00076406804,0.54516953,0.00063495245,0.00034545592,0.0015677133,0.048907045,0.0041041067,0.00427181],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99593854,0.0012677988,0.00026243535,0.0012065972,0.0011493765,0.00017536216],"domain_scores_gemma":[0.99214,0.004417303,0.0012757315,0.00092175324,0.0008614995,0.0003837933],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0036155563,0.0011928771,0.0012200566,0.0029744175,0.00048184762,0.0017027955,0.0013568139,0.0007939488,0.007122643],"category_scores_gemma":[0.02053775,0.00040670604,0.0010792202,0.0036056307,0.0013027691,0.0015463488,0.0020362486,0.0016762727,0.0049594343],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0026382469,0.00058845745,0.08320823,0.0035001491,0.00079279684,0.0008554668,0.0006992945,0.088990204,0.08319111,0.056360886,0.14038329,0.53879195],"study_design_scores_gemma":[0.00030905698,0.0009169876,0.035743214,0.0002263979,0.000216576,0.0018633873,0.00017674563,0.71732855,0.05076164,0.1280761,0.06413734,0.00024399807],"about_ca_topic_score_codex":0.00081027456,"about_ca_topic_score_gemma":0.0009876267,"teacher_disagreement_score":0.007122643,"about_ca_system_score_codex":0.0006148079,"about_ca_system_score_gemma":0.0013185937,"threshold_uncertainty_score":0.023827612},"labels":[],"label_agreement":null},{"id":"W2952228520","doi":"10.1093/bioinformatics/btz130","title":"LIONS: analysis suite for detecting and quantifying transposable element initiated transcription from RNA-seq","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Chromosomal and Genetic Variations","field":"Agricultural and Biological Sciences","cited_by":33,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canada's Michael Smith Genome Sciences Centre; Terry Fox Research Institute; University of British Columbia","funders":"Natural Sciences and Engineering Research Council of Canada; University of British Columbia","keywords":"RNA-Seq; Transposable element; Suite; Container (type theory); Computer science; Source code; Computational biology; Transcription (linguistics); Transcriptome; Biology; Genetics; Genome; Gene; Gene expression; Programming language","score_opus":0.060007125401452754,"score_gpt":0.2551174693935599,"score_spread":0.19511034399210714,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2952228520","genre_codex":"software","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0131184105,0.0008251815,0.38219222,0.00029574014,0.0002777594,0.0010225123,0.16437012,0.43219933,0.0056987717],"genre_scores_gemma":[0.0423418,0.0006993104,0.6297289,0.0016795433,0.00013221307,0.008106146,0.21332525,0.09494003,0.009046806],"study_design_codex":"not_applicable","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99803096,0.0003635882,0.00021711814,0.0006401767,0.0005600405,0.00018809605],"domain_scores_gemma":[0.997531,0.0013439538,0.00033079818,0.0002671324,0.00038101082,0.0001460942],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0043976195,0.0037705344,0.001963537,0.0025114275,0.001196253,0.0028862485,0.002420383,0.0012254361,0.026376665],"category_scores_gemma":[0.00597147,0.0019653956,0.0021637895,0.0014371305,0.0008302658,0.0015183389,0.0019320172,0.0034455177,0.01732418],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0033351262,0.0003321994,0.01425901,0.009087341,0.0016148486,0.0009020987,0.0016162244,0.010262866,0.29744458,0.012741061,0.53353095,0.11487357],"study_design_scores_gemma":[0.0010175795,0.0009177681,0.031759914,0.00091118854,0.00069898536,0.0012328997,0.00045453367,0.19369912,0.34239143,0.021689638,0.4042369,0.000989979],"about_ca_topic_score_codex":0.0025484287,"about_ca_topic_score_gemma":0.0050546504,"teacher_disagreement_score":0.026376665,"about_ca_system_score_codex":0.0013410188,"about_ca_system_score_gemma":0.0028074286,"threshold_uncertainty_score":0.088238716},"labels":[],"label_agreement":null},{"id":"W2952588534","doi":"10.1093/bioinformatics/btw695","title":"EGAD: ultra-fast functional analysis of gene networks","year":2016,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":78,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canada's Michael Smith Genome Sciences Centre; University of British Columbia","funders":"National Institute of Mental Health; National Institutes of Health","keywords":"Bioconductor; Computer science; Set (abstract data type); R package; Data mining; Software; Gene regulatory network; MIT License; Task (project management); Throughput; Network analysis; Machine learning; Gene; Biology; Programming language; Genetics","score_opus":0.008504445161484603,"score_gpt":0.206415465538498,"score_spread":0.19791102037701339,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2952588534","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.016953023,0.00092355744,0.716079,0.00064377865,0.00030450698,0.00020133535,0.04198247,0.21771568,0.005196589],"genre_scores_gemma":[0.13912329,0.0008732412,0.750414,0.00048178123,0.00014743327,0.0013834122,0.078910604,0.020572292,0.008094008],"study_design_codex":"not_applicable","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9992779,0.00018774951,0.00003489217,0.00018388504,0.00025836568,0.000057199126],"domain_scores_gemma":[0.9982564,0.0009934445,0.00011246775,0.0003038869,0.0002268274,0.00010704712],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0017170005,0.0020435522,0.0013725236,0.0025982228,0.0008582257,0.002028,0.0030579402,0.00096377183,0.032137558],"category_scores_gemma":[0.005695674,0.0008221239,0.001968761,0.0016830948,0.00052508985,0.0019373011,0.0022142048,0.0019758372,0.013226686],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014162263,0.00032064988,0.0092300745,0.0028637203,0.0009924549,0.00081299146,0.00054205296,0.12484489,0.029189972,0.040667452,0.4500872,0.33903235],"study_design_scores_gemma":[0.0004439209,0.0001545672,0.0054422417,0.00016325378,0.00016655053,0.00059076294,0.000108350556,0.7926236,0.017453643,0.07789186,0.104824476,0.00013685512],"about_ca_topic_score_codex":0.0028583196,"about_ca_topic_score_gemma":0.0037171727,"teacher_disagreement_score":0.032137558,"about_ca_system_score_codex":0.00070870254,"about_ca_system_score_gemma":0.0012079737,"threshold_uncertainty_score":0.107510805},"labels":[],"label_agreement":null},{"id":"W2952646324","doi":"10.1093/bioinformatics/btx603","title":"Segway 2.0: Gaussian mixture models and minibatch training","year":2017,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":49,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto; Princess Margaret Cancer Centre; University of British Columbia","funders":"National Human Genome Research Institute; Canadian Institutes of Health Research; National Institutes of Health; Natural Sciences and Engineering Research Council of Canada; University Health Network","keywords":"Computer science; Scripting language; Annotation; Source code; Gaussian; Mixture model; Training set; Data mining; Code (set theory); Artificial intelligence; Machine learning; Pattern recognition (psychology); Set (abstract data type); Programming language","score_opus":0.02396829629731378,"score_gpt":0.25035827344199824,"score_spread":0.22638997714468445,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2952646324","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00875849,0.0008163284,0.6179832,0.0005409288,0.0004783747,0.00039482873,0.016166765,0.3502058,0.0046553197],"genre_scores_gemma":[0.086689636,0.00048390756,0.77563053,0.0014008428,0.0001609412,0.002728336,0.06538908,0.05150453,0.01601218],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9990803,0.00026291018,0.00007338188,0.00032346125,0.00013588277,0.00012408815],"domain_scores_gemma":[0.99809784,0.00092749053,0.000066488334,0.00043441265,0.0003687674,0.00010506803],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0022271527,0.0029983174,0.0015521041,0.0012177583,0.0008986014,0.0022294407,0.007102337,0.003157438,0.072068565],"category_scores_gemma":[0.009617951,0.0024649797,0.002581184,0.0015054883,0.0006906379,0.0026316314,0.0030809976,0.004513373,0.041753996],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014684503,0.00031558596,0.0034218999,0.0009953103,0.00068992196,0.00029150795,0.00031157536,0.183136,0.0042596874,0.009760158,0.48544186,0.30990803],"study_design_scores_gemma":[0.00027223065,0.000103976345,0.0007950676,0.000079443955,0.00008017952,0.00014234982,0.00007117942,0.91501385,0.010673809,0.024606192,0.048065454,0.00009632734],"about_ca_topic_score_codex":0.012204418,"about_ca_topic_score_gemma":0.027217394,"teacher_disagreement_score":0.072068565,"about_ca_system_score_codex":0.0015142701,"about_ca_system_score_gemma":0.0030608738,"threshold_uncertainty_score":0.2410934},"labels":[],"label_agreement":null},{"id":"W2952713926","doi":"10.1093/bioinformatics/bty244","title":"COSSMO: predicting competitive alternative splice site selection using deep learning","year":2018,"lang":"en","type":"article","venue":"Bioinformatics","topic":"RNA Research and Splicing","field":"Biochemistry, Genetics and Molecular Biology","cited_by":66,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Ontario Genomics; University of Toronto","funders":"","keywords":"splice; Computer science; RNA splicing; Sequence (biology); Alternative splicing; Computational biology; Selection (genetic algorithm); Artificial intelligence; Data mining; Gene; Biology; Genetics; RNA","score_opus":0.01461999723287225,"score_gpt":0.2846627027470284,"score_spread":0.2700427055141561,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2952713926","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.47033072,0.0015498964,0.50387603,0.00092715555,0.00017112015,0.00012025967,0.008440958,0.0087363,0.0058475975],"genre_scores_gemma":[0.8647924,0.00037218255,0.113486305,0.00051492336,0.0000737962,0.00018564999,0.014799141,0.00033072373,0.0054448503],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9997961,0.000036737874,0.000009783672,0.000075185155,0.000047974845,0.00003423929],"domain_scores_gemma":[0.9995103,0.00024231055,0.000052121824,0.000050690494,0.00010106036,0.000043592547],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006815013,0.0009800812,0.00052947743,0.0004703545,0.00030506466,0.000516724,0.0012739646,0.00093198713,0.0018646634],"category_scores_gemma":[0.0014560797,0.00034843085,0.0005235047,0.00056166557,0.00039368402,0.00072924996,0.0006184092,0.0012331189,0.00075289066],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005484579,0.00031123546,0.02772311,0.00022266703,0.00018864303,0.00020152601,0.00006541968,0.7972596,0.026633838,0.0066537904,0.017839668,0.122352034],"study_design_scores_gemma":[0.0000129832315,0.000017233415,0.0005719092,0.0000039231827,0.0000066180824,0.000018938363,0.0000040285436,0.9941579,0.0025030093,0.0022336575,0.00046491082,0.0000048750026],"about_ca_topic_score_codex":0.0085039465,"about_ca_topic_score_gemma":0.017013635,"teacher_disagreement_score":0.0085039465,"about_ca_system_score_codex":0.0008113324,"about_ca_system_score_gemma":0.0009747963,"threshold_uncertainty_score":0.016908884},"labels":[],"label_agreement":null},{"id":"W2952985080","doi":"10.1093/bioinformatics/btz445","title":"A scalable method for parameter identification in kinetic models of metabolism using steady-state data","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Microbial Metabolic Engineering and Bioproduction","field":"Biochemistry, Genetics and Molecular Biology","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Identifiability; Computer science; Scalability; Estimation theory; A priori and a posteriori; Biological system; Steady state (chemistry); Identification (biology); Experimental data; Data mining; Algorithm; Mathematics; Statistics; Machine learning; Chemistry; Biology; Database","score_opus":0.033322204304405655,"score_gpt":0.2872780953079232,"score_spread":0.25395589100351756,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2952985080","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000828718,0.0000670653,0.9919517,0.000080113634,0.00002969552,0.00007445198,0.00085062225,0.0057771555,0.00034051784],"genre_scores_gemma":[0.030779384,0.00016265275,0.95991963,0.00013477186,0.00007190877,0.000813201,0.005324976,0.0016823437,0.0011111917],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9977737,0.0006747678,0.00016933653,0.0005471223,0.00073042436,0.00010456417],"domain_scores_gemma":[0.9921848,0.004716213,0.00048195163,0.0014620817,0.00096547185,0.00018945595],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0047722114,0.002415774,0.0017584505,0.002234779,0.0008514033,0.0020155532,0.003154747,0.001895138,0.015041986],"category_scores_gemma":[0.021675436,0.0013770327,0.0024629426,0.0020972437,0.0009978461,0.002483467,0.0031935698,0.0038005782,0.0078073773],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00060763274,0.00045858262,0.0034711312,0.001574764,0.00047936366,0.00061382167,0.00026367969,0.439443,0.027891695,0.05091457,0.037079576,0.4372023],"study_design_scores_gemma":[0.00009516427,0.000034931705,0.00025886938,0.000040500123,0.000027426866,0.00008063542,0.000023799168,0.94304746,0.0038827762,0.044922326,0.007553946,0.000032157557],"about_ca_topic_score_codex":0.0035422693,"about_ca_topic_score_gemma":0.0060177147,"teacher_disagreement_score":0.015041986,"about_ca_system_score_codex":0.0012158736,"about_ca_system_score_gemma":0.0036287403,"threshold_uncertainty_score":0.050320446},"labels":[],"label_agreement":null},{"id":"W2953282636","doi":"10.1093/bioinformatics/btx732","title":"MOSAIC: a chemical-genetic interaction data repository and web resource for exploring chemical modes of action","year":2017,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Computational Drug Discovery Methods","field":"Computer Science","cited_by":18,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"National Institute of General Medical Sciences; National Human Genome Research Institute; RIKEN; Japan Society for the Promotion of Science; Canadian Institutes of Health Research; University of Minnesota; National Institutes of Health; National Science Foundation","keywords":"Computer science; Interface (matter); Resource (disambiguation); Computational biology; Annotation; Mosaic; User interface; Similarity (geometry); Web site; Ontology; Action (physics); Biology; The Internet; Artificial intelligence; World Wide Web","score_opus":0.1828911589620814,"score_gpt":0.36057274565214587,"score_spread":0.17768158669006445,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2953282636","genre_codex":"dataset","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0065667536,0.0010500218,0.047214705,0.00048601997,0.00012957721,0.000541783,0.84869486,0.08944322,0.005873015],"genre_scores_gemma":[0.01171465,0.0007040631,0.063336186,0.00017793408,0.000036373418,0.0008979455,0.9168938,0.004893192,0.0013458756],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99887186,0.00013992866,0.00020560257,0.0002872738,0.0004074314,0.000087743254],"domain_scores_gemma":[0.99705064,0.0007831699,0.00049064466,0.00068215217,0.0004516756,0.0005417365],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0024720903,0.0023101259,0.0021738603,0.0062634614,0.00094809505,0.0020656649,0.004793845,0.0019049222,0.030438185],"category_scores_gemma":[0.006583226,0.0013609417,0.0019118721,0.0073990067,0.0005037469,0.0032545219,0.003715702,0.002095472,0.015967863],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0033591078,0.00045893335,0.008754997,0.010365858,0.0007237736,0.0011304672,0.00034733693,0.013098006,0.06110592,0.018746931,0.77813226,0.103776485],"study_design_scores_gemma":[0.001598784,0.00044309912,0.013165312,0.0007135851,0.0005548864,0.000898715,0.00021522271,0.049929246,0.045137573,0.02793737,0.85898054,0.00042572664],"about_ca_topic_score_codex":0.0057063457,"about_ca_topic_score_gemma":0.006587478,"teacher_disagreement_score":0.030438185,"about_ca_system_score_codex":0.0010945831,"about_ca_system_score_gemma":0.003375977,"threshold_uncertainty_score":0.10182589},"labels":[],"label_agreement":null},{"id":"W2953326935","doi":"10.1093/bioinformatics/btz400","title":"ntEdit: scalable genome sequence polishing","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":132,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canada's Michael Smith Genome Sciences Centre; University of British Columbia; Genome British Columbia; Université Laval; Natural Resources Canada; Ministry of Forests","funders":"National Human Genome Research Institute; National Institutes of Health","keywords":"Sequence (biology); Scalability; Polishing; Computer science; Whole genome sequencing; Genome; Computational biology; Genetics; Biology; Gene; Materials science; Database; Metallurgy","score_opus":0.014467627052544538,"score_gpt":0.2310440146055117,"score_spread":0.21657638755296715,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2953326935","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.030843401,0.0010855368,0.39095196,0.00055931375,0.0005720819,0.0007035224,0.027188344,0.5437394,0.0043564527],"genre_scores_gemma":[0.08277246,0.00067135965,0.7792966,0.0006929279,0.00010976476,0.0013360243,0.08930602,0.04038668,0.0054281354],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9982918,0.00026325972,0.00017408472,0.0005936857,0.0005394948,0.00013770342],"domain_scores_gemma":[0.9971175,0.0010652561,0.00025725082,0.00080046046,0.0005629479,0.0001966297],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002397788,0.0024648434,0.0014990289,0.0013493609,0.0011490379,0.001792068,0.0031682884,0.0011546467,0.0109936185],"category_scores_gemma":[0.008804723,0.0013580492,0.0021455388,0.0017848136,0.0007824562,0.0022220882,0.0024394835,0.0032801132,0.01140517],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.004426415,0.00042855763,0.006770637,0.003443937,0.00076990156,0.0013338835,0.0011317937,0.046430413,0.24819952,0.006317002,0.329496,0.35125196],"study_design_scores_gemma":[0.0014672673,0.0006537031,0.005126001,0.00023582458,0.00024631896,0.0012129437,0.0003063081,0.41589966,0.38173416,0.016478056,0.17615396,0.000485787],"about_ca_topic_score_codex":0.0035155169,"about_ca_topic_score_gemma":0.0046662833,"teacher_disagreement_score":0.0109936185,"about_ca_system_score_codex":0.0009542556,"about_ca_system_score_gemma":0.0015392049,"threshold_uncertainty_score":0.036777377},"labels":[],"label_agreement":null},{"id":"W2953409640","doi":"10.1093/bioinformatics/btz522","title":"Pygenprop: a Python library for programmatic exploration and comparison of organism genome properties","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"Natural Sciences and Engineering Research Council of Canada; Université de Genève; Nuclear Waste Management Organization","keywords":"Python (programming language); Computer science; Documentation; Genome; Source code; Organism; Software; MIT License; Genomics; Programming language; World Wide Web; Biology; Genetics; Gene","score_opus":0.025425893781296366,"score_gpt":0.23177017140923153,"score_spread":0.20634427762793517,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2953409640","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0028139905,0.00031871302,0.2004042,0.00033391683,0.00021712112,0.00023659073,0.06057251,0.72800034,0.0071026348],"genre_scores_gemma":[0.039774805,0.000847358,0.33500466,0.0014017269,0.00014232163,0.0025201996,0.20401214,0.40476793,0.0115288645],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99849737,0.00022971447,0.0001265269,0.00041576862,0.00053358334,0.00019703669],"domain_scores_gemma":[0.99842876,0.00064541085,0.00017214484,0.00030697032,0.00027034758,0.00017646562],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0031569882,0.0027436516,0.0019618594,0.0026188465,0.0016387235,0.0030941367,0.004734088,0.00086589146,0.08908974],"category_scores_gemma":[0.006224778,0.0021313804,0.00281305,0.002176083,0.0012175535,0.004360742,0.004826546,0.0043976507,0.05461715],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009918825,0.0001943569,0.004012258,0.002906936,0.00047760343,0.0005550072,0.0006874325,0.0049092993,0.016454909,0.012863364,0.84301585,0.11293114],"study_design_scores_gemma":[0.0005698927,0.00015838876,0.008717113,0.0005815137,0.00019091599,0.001036243,0.00025169304,0.070370376,0.046802707,0.058891926,0.81195277,0.00047650316],"about_ca_topic_score_codex":0.002896803,"about_ca_topic_score_gemma":0.0031345603,"teacher_disagreement_score":0.08908974,"about_ca_system_score_codex":0.0010896609,"about_ca_system_score_gemma":0.0031034432,"threshold_uncertainty_score":0.29803485},"labels":[],"label_agreement":null},{"id":"W2953737873","doi":"10.1093/bioinformatics/btz534","title":"<i>RTNduals</i>: an R/Bioconductor package for analysis of co-regulation and inference of <i>dual regulons</i>","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":19,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canada's Michael Smith Genome Sciences Centre; BC Cancer Agency","funders":"National Cancer Institute; Conselho Nacional de Desenvolvimento Científico e Tecnológico; Cancer Research UK; Breast Cancer Research Foundation","keywords":"Bioconductor; Regulon; Computer science; Inference; R package; Computational biology; Biology; Data mining; Regulation of gene expression; Gene; Genetics; Artificial intelligence; Programming language","score_opus":0.0168775380279266,"score_gpt":0.27257583472530783,"score_spread":0.25569829669738126,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2953737873","genre_codex":"dataset","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0042009996,0.0022543306,0.29895228,0.0016417755,0.0015070119,0.0007866515,0.39037138,0.2900645,0.010221045],"genre_scores_gemma":[0.03823686,0.0015833266,0.45630112,0.0024147937,0.00049247645,0.008602085,0.29803866,0.18097883,0.013351811],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99388254,0.001944724,0.00057355437,0.0020733608,0.0011392401,0.00038668522],"domain_scores_gemma":[0.9831154,0.010144842,0.0014066778,0.002952789,0.0017797808,0.0006005308],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.009545544,0.005166395,0.0050245337,0.0058443006,0.0016833728,0.0055467677,0.008504867,0.0024834801,0.10615633],"category_scores_gemma":[0.040082917,0.002635034,0.0049634627,0.006696596,0.0017261361,0.0027966178,0.005009317,0.005379003,0.0832917],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006464338,0.0000801882,0.005580122,0.007046993,0.0030058364,0.00038194648,0.00044882757,0.0056732227,0.003523098,0.011911034,0.9297181,0.031984128],"study_design_scores_gemma":[0.0011583611,0.00021232327,0.015619767,0.0017465026,0.0026917115,0.0014899971,0.00026567178,0.07017552,0.013776878,0.08095901,0.81132805,0.00057628355],"about_ca_topic_score_codex":0.0061709504,"about_ca_topic_score_gemma":0.008722637,"teacher_disagreement_score":0.10615633,"about_ca_system_score_codex":0.0013952319,"about_ca_system_score_gemma":0.0055373223,"threshold_uncertainty_score":0.3551283},"labels":[],"label_agreement":null},{"id":"W2953949806","doi":"10.1093/bioinformatics/btz343","title":"Large-scale mammalian genome rearrangements coincide with chromatin interactions","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genome Rearrangement Algorithms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"Agence Nationale de la Recherche","keywords":"Breakpoint; Chromosomal rearrangement; Biology; Genome; Chromatin; Chromosome; Genetics; Gene rearrangement; Computational biology; Gene; Gene duplication; Karyotype","score_opus":0.006863780160027628,"score_gpt":0.22537573313091283,"score_spread":0.21851195297088521,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2953949806","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.25322255,0.002346215,0.52068955,0.0023586012,0.0002740024,0.00041185427,0.09407706,0.11457165,0.012048492],"genre_scores_gemma":[0.5574647,0.00092213193,0.3337788,0.0007843344,0.00012182616,0.0004612854,0.09431409,0.008686785,0.0034659796],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9987337,0.00015691799,0.000070554386,0.0006012339,0.0003293733,0.0001081948],"domain_scores_gemma":[0.9973985,0.0013634391,0.00046287372,0.00033962142,0.00018658856,0.00024906045],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0017967777,0.000737236,0.0008041085,0.001350933,0.0012332158,0.0017438591,0.0012042829,0.0009824005,0.015210304],"category_scores_gemma":[0.007475166,0.0009414859,0.0009899847,0.00225485,0.0008173485,0.001571196,0.0020551893,0.0016388724,0.007707516],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0034257204,0.00031419226,0.16143057,0.0061675757,0.0008996236,0.0012977734,0.0012197158,0.03543381,0.42841178,0.030438276,0.109342545,0.22161838],"study_design_scores_gemma":[0.00057681865,0.00073457,0.28100094,0.00042648026,0.00071676966,0.003856185,0.0006833237,0.20912941,0.27258596,0.09109479,0.13878809,0.00040656762],"about_ca_topic_score_codex":0.0017524977,"about_ca_topic_score_gemma":0.0037543736,"teacher_disagreement_score":0.015210304,"about_ca_system_score_codex":0.0006075531,"about_ca_system_score_gemma":0.0010014361,"threshold_uncertainty_score":0.05088353},"labels":[],"label_agreement":null},{"id":"W2955502047","doi":"10.1093/bioinformatics/btz318","title":"MOLI: multi-omics late integration with deep neural networks for drug response prediction","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":407,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia; Simon Fraser University","funders":"Canadian Institutes of Health Research; Terry Fox Foundation; Canada Foundation for Innovation","keywords":"Drug response; Computer science; Representation (politics); Machine learning; Artificial intelligence; Omics; Precision oncology; Artificial neural network; Computational biology; Data mining; Drug; Precision medicine; Bioinformatics; Biology","score_opus":0.007039244457505459,"score_gpt":0.21430855938702414,"score_spread":0.2072693149295187,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2955502047","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.10855252,0.008728731,0.8443263,0.003481431,0.0003549055,0.00042813574,0.0061997613,0.022094727,0.005833456],"genre_scores_gemma":[0.68581486,0.0016613551,0.28976077,0.0023544226,0.0003307235,0.00073320424,0.012557405,0.00043415406,0.006353132],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.999433,0.00014224596,0.000033231623,0.00014407322,0.0001537544,0.00009368087],"domain_scores_gemma":[0.99930584,0.00031771077,0.00011896668,0.0000732386,0.0001291748,0.000055032382],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0017393784,0.0021510578,0.0015006386,0.0014209943,0.00034072125,0.0010880418,0.0019882084,0.0016019074,0.0021134154],"category_scores_gemma":[0.0030120942,0.00056517764,0.0014666428,0.0012397139,0.00038706337,0.0011293816,0.0015682097,0.0025424904,0.0007691149],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00079020613,0.0006774492,0.014398161,0.0006097226,0.0008572573,0.00035815142,0.00011292489,0.5752792,0.01511381,0.0040181526,0.021048404,0.3667366],"study_design_scores_gemma":[0.00001442192,0.00006682864,0.0005470891,0.000016979178,0.000038521743,0.000027246668,0.00000668726,0.99407035,0.00217333,0.0021611457,0.00086480216,0.000012570868],"about_ca_topic_score_codex":0.006165219,"about_ca_topic_score_gemma":0.007571811,"teacher_disagreement_score":0.006165219,"about_ca_system_score_codex":0.0017224252,"about_ca_system_score_gemma":0.0014312064,"threshold_uncertainty_score":0.012497127},"labels":[],"label_agreement":null},{"id":"W2955780193","doi":"","title":"Method Choice in Gene Set Analysis Has Important Consequences for Analysis Outcome.","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetics, Bioinformatics, and Biomedical Research","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Royal University Hospital; University of Saskatchewan","funders":"","keywords":"Outcome (game theory); Set (abstract data type); Computer science; Genetic analysis; Computational biology; Gene; Genetics; Biology; Mathematical economics; Mathematics; Programming language","score_opus":0.05818490484697021,"score_gpt":0.36843060612477324,"score_spread":0.31024570127780304,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2955780193","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"methods","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":"methods","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.13311815,0.011083996,0.82821566,0.014567435,0.0014459736,0.00071435294,0.0024295284,0.001790042,0.0066348296],"genre_scores_gemma":[0.52312684,0.0036176404,0.46106347,0.004530213,0.0006126979,0.0012561007,0.001997281,0.002242138,0.0015535435],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.85952663,0.113969356,0.007266974,0.007638869,0.010080557,0.0015175423],"domain_scores_gemma":[0.5522031,0.42324814,0.003088539,0.013533298,0.006200553,0.0017263521],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.14937802,0.0012197561,0.0021282062,0.0027318085,0.0022051674,0.0057421993,0.0028415543,0.002710014,0.0023595528],"category_scores_gemma":[0.3072792,0.0006861694,0.0019976203,0.0035353622,0.0031206687,0.003392133,0.0028750189,0.0045711906,0.0016223482],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0075237895,0.0012523044,0.20191744,0.0067363298,0.009333667,0.0022184534,0.0031520885,0.023506856,0.07361672,0.058496635,0.02914931,0.5830965],"study_design_scores_gemma":[0.0021536958,0.0020835274,0.09544454,0.002527309,0.0029896016,0.00490656,0.0022592645,0.16511258,0.09327152,0.58279127,0.045537252,0.0009228687],"about_ca_topic_score_codex":0.0014101691,"about_ca_topic_score_gemma":0.0029637236,"teacher_disagreement_score":0.850622,"about_ca_system_score_codex":0.0011074761,"about_ca_system_score_gemma":0.0031048567,"threshold_uncertainty_score":0.7899959},"labels":[],"label_agreement":null},{"id":"W2956622428","doi":"10.1093/bioinformatics/btz393","title":"MCS2: minimal coordinated supports for fast enumeration of minimal cut sets in metabolic networks","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Microbial Metabolic Engineering and Bioproduction","field":"Biochemistry, Genetics and Molecular Biology","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Computer science; Correctness; Set (abstract data type); Enumeration; Metabolic network; Computation; Constraint (computer-aided design); Novelty; Dual (grammatical number); Algorithm; Mathematical optimization; Theoretical computer science; Discrete mathematics; Mathematics; Programming language; Bioinformatics","score_opus":0.00520616366990512,"score_gpt":0.21934020487152653,"score_spread":0.2141340412016214,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2956622428","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.02199838,0.00017252783,0.97047204,0.00027639713,0.000057634927,0.00015907231,0.0010268204,0.0027958467,0.0030412283],"genre_scores_gemma":[0.196598,0.0001287296,0.7968778,0.00015182771,0.00007160215,0.0004798098,0.0030271402,0.0008170547,0.0018480833],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9991473,0.00020285464,0.00004519083,0.0001431518,0.00036240258,0.000099115416],"domain_scores_gemma":[0.99694747,0.0017285629,0.00018262635,0.0004725182,0.00047636777,0.00019245638],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000841129,0.0017825646,0.0012192281,0.0018242559,0.00067541335,0.0017218951,0.0023179695,0.0013249385,0.009025827],"category_scores_gemma":[0.008806429,0.000644429,0.0016209664,0.0015877949,0.00093631184,0.0023179452,0.0027813707,0.0022680634,0.0017414885],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00060500414,0.00017561716,0.0035394302,0.00059568975,0.00013194086,0.00033195436,0.0003563353,0.6800835,0.008897931,0.080972694,0.0176744,0.2066354],"study_design_scores_gemma":[0.00004383844,0.000025418609,0.00012056578,0.000019363792,0.000008671941,0.000034990775,0.00003047026,0.95794547,0.0017796297,0.03776438,0.0022166583,0.000010574661],"about_ca_topic_score_codex":0.0045738337,"about_ca_topic_score_gemma":0.0062558833,"teacher_disagreement_score":0.009025827,"about_ca_system_score_codex":0.0009735799,"about_ca_system_score_gemma":0.0018155844,"threshold_uncertainty_score":0.030194461},"labels":[],"label_agreement":null},{"id":"W2960228950","doi":"10.1093/bioinformatics/btz335","title":"Enhancing the drug discovery process: Bayesian inference for the analysis and comparison of dose–response experiments","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Statistical Methods in Clinical Trials","field":"Mathematics","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal; Institute for Research in Immunology and Cancer","funders":"Genome Canada","keywords":"Computer science; Bayesian probability; Inference; Data mining; Bayesian inference; Machine learning; Artificial intelligence","score_opus":0.23821785965506612,"score_gpt":0.530511155671486,"score_spread":0.29229329601641985,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2960228950","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0016638668,0.00021865027,0.99686414,0.0002601161,0.00001684857,0.00012065897,0.00012242264,0.00043898402,0.0002943712],"genre_scores_gemma":[0.075197004,0.00043093276,0.9218145,0.0005494348,0.000113437876,0.000804733,0.00044195887,0.00025625495,0.00039166328],"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9614468,0.030076703,0.0012589913,0.0021258504,0.0047456506,0.00034593564],"domain_scores_gemma":[0.8836495,0.10094577,0.004799567,0.0050773895,0.0046783295,0.000849378],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.055530496,0.0016888509,0.0024127054,0.0032100559,0.00073571975,0.0030078043,0.0028551926,0.0027540305,0.0042972104],"category_scores_gemma":[0.1569102,0.0012842048,0.0026515624,0.002458247,0.001963223,0.002997991,0.0028406424,0.0045468756,0.0013775128],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013711831,0.00057119527,0.0071848696,0.0013738757,0.001245962,0.000269965,0.00031222941,0.5862555,0.008063361,0.0718238,0.006881962,0.31464612],"study_design_scores_gemma":[0.00014799897,0.00014802057,0.00087697164,0.00009584976,0.00011969822,0.00007552059,0.000011227682,0.9209541,0.002742581,0.07199001,0.0027959747,0.000042035947],"about_ca_topic_score_codex":0.0028928337,"about_ca_topic_score_gemma":0.0029849568,"teacher_disagreement_score":0.055530496,"about_ca_system_score_codex":0.0022303446,"about_ca_system_score_gemma":0.004675132,"threshold_uncertainty_score":0.29367685},"labels":[],"label_agreement":null},{"id":"W2961295341","doi":"10.1093/bioinformatics/btz439","title":"ISMB/ECCB 2019 Proceedings","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Single-cell and spatial transcriptomics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":3,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"","keywords":"Computer science","score_opus":0.006793132206369344,"score_gpt":0.2020652547703505,"score_spread":0.19527212256398116,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2961295341","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.011456425,0.008753636,0.21047297,0.03952384,0.033779014,0.0010757959,0.058001004,0.037262026,0.5996753],"genre_scores_gemma":[0.023159696,0.0041983332,0.08689095,0.0040102783,0.0033931923,0.0005070736,0.07578204,0.01010883,0.7919495],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9986425,0.0003508867,0.00006231371,0.0002726869,0.00052725646,0.00014435493],"domain_scores_gemma":[0.9954046,0.0007200618,0.00011044549,0.0010031038,0.0016478796,0.001113962],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0050458405,0.0016376801,0.0013667438,0.0021689332,0.0012064851,0.004946263,0.001719062,0.0024121122,0.26011175],"category_scores_gemma":[0.005311394,0.0005334098,0.00086121535,0.0016815363,0.0008979668,0.001669805,0.0026770437,0.0021762373,0.29997492],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00035748523,0.00009151853,0.00023938957,0.00023667347,0.000043846332,0.000108644366,0.000030324522,0.0006843243,0.0077826595,0.0068876767,0.8305622,0.15297525],"study_design_scores_gemma":[0.00004242835,0.000056718203,0.0010961357,0.00007493218,0.0000201372,0.00010422409,0.000027813758,0.0032454317,0.002915146,0.0061448193,0.98625284,0.000019383977],"about_ca_topic_score_codex":0.0035238923,"about_ca_topic_score_gemma":0.0030408895,"teacher_disagreement_score":0.26011175,"about_ca_system_score_codex":0.0012220303,"about_ca_system_score_gemma":0.0025648542,"threshold_uncertainty_score":0.87016046},"labels":[],"label_agreement":null},{"id":"W2962028921","doi":"10.1093/bioinformatics/btz355","title":"Collaborative intra-tumor heterogeneity detection","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Cancer Genomics and Diagnostics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":25,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"BC Cancer Agency; Simon Fraser University","funders":"","keywords":"Computer science; Inference; Phylogenetic tree; Deconvolution; Source code; Process (computing); Data mining; Computational biology; Machine learning; Artificial intelligence; Biology; Algorithm; Genetics; Gene","score_opus":0.004623206389557855,"score_gpt":0.22304652164240288,"score_spread":0.21842331525284503,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2962028921","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.09111304,0.0024543982,0.8989091,0.0007573547,0.00012393933,0.00018469743,0.0013505531,0.0027775448,0.0023293977],"genre_scores_gemma":[0.7849638,0.0006503361,0.20448184,0.00048066626,0.00035628036,0.0001754151,0.0041642673,0.00029319545,0.0044343555],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9967758,0.0006215404,0.00013755124,0.001440195,0.00070248893,0.00032234268],"domain_scores_gemma":[0.99453664,0.0029471088,0.0006051318,0.0009237706,0.0006165878,0.00037074878],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0031677387,0.0013799065,0.0024052477,0.0023292678,0.0008315964,0.0016111984,0.0031222592,0.0019523444,0.0012150216],"category_scores_gemma":[0.007328323,0.0007128413,0.0014317108,0.0017375321,0.00078259036,0.0013135618,0.0023266443,0.0012892602,0.00087928574],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0012366168,0.00037214305,0.040585097,0.0005955007,0.00091830187,0.0021187274,0.0006154707,0.37131363,0.0364132,0.00519262,0.019801404,0.5208373],"study_design_scores_gemma":[0.00004338129,0.00010761463,0.004570236,0.00002294902,0.00014358548,0.0010118597,0.000099024626,0.96967006,0.011316153,0.008951764,0.004018829,0.000044560446],"about_ca_topic_score_codex":0.00546622,"about_ca_topic_score_gemma":0.0058937585,"teacher_disagreement_score":0.00546622,"about_ca_system_score_codex":0.0007462281,"about_ca_system_score_gemma":0.0011809225,"threshold_uncertainty_score":0.01675278},"labels":[],"label_agreement":null},{"id":"W2962655813","doi":"10.1093/bioinformatics/btz337","title":"Prediction of mRNA subcellular localization using deep recurrent neural networks","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Machine Learning in Bioinformatics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":86,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal; Montreal Clinical Research Institute; McGill University","funders":"Institut de Valorisation des Données","keywords":"Computational biology; Sequence (biology); Computer science; Identification (biology); RNA; Subcellular localization; Variety (cybernetics); Messenger RNA; Biology; Set (abstract data type); Deep learning; RNA-binding protein; Artificial neural network; Artificial intelligence; Gene; Genetics","score_opus":0.012123227150818776,"score_gpt":0.23152418127151167,"score_spread":0.21940095412069288,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2962655813","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.2756405,0.00252247,0.7018111,0.00074769143,0.00011839814,0.000058733865,0.003960232,0.011136943,0.0040040263],"genre_scores_gemma":[0.89763504,0.00076452066,0.08994395,0.0001760483,0.0000484029,0.00006790537,0.006307368,0.00019444557,0.0048622317],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.999853,0.000022094693,0.0000071994136,0.00006048161,0.000030522013,0.000026700462],"domain_scores_gemma":[0.9995516,0.00019343925,0.00007420644,0.00003228538,0.000121783225,0.00002669524],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003727928,0.0008370242,0.00046962698,0.0006033487,0.00017278247,0.00055526634,0.0007384542,0.00058557926,0.001440438],"category_scores_gemma":[0.0011642672,0.00029757209,0.000432448,0.0004753293,0.0002692344,0.00051829877,0.00036907487,0.0008988288,0.0009705163],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00050990126,0.00015397325,0.0073031806,0.00029781333,0.00010638578,0.00034060105,0.000045645334,0.7865116,0.049992546,0.004663385,0.007656909,0.14241803],"study_design_scores_gemma":[0.000005376842,0.000012660193,0.00037082704,0.0000054348143,0.0000063808243,0.000014722877,0.0000037079642,0.9928202,0.005260788,0.0012626004,0.00023401789,0.0000033265405],"about_ca_topic_score_codex":0.0060318955,"about_ca_topic_score_gemma":0.008377201,"teacher_disagreement_score":0.0060318955,"about_ca_system_score_codex":0.0008321742,"about_ca_system_score_gemma":0.00057854457,"threshold_uncertainty_score":0.011993587},"labels":[],"label_agreement":null},{"id":"W2964417596","doi":"10.1093/bioinformatics/btz602","title":"Non-parametric individual treatment effect estimation for survival data with random forests","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Advanced Causal Inference Techniques","field":"Mathematics","cited_by":41,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"HEC Montréal","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Covariate; Random forest; Censoring (clinical trials); Computer science; Estimation; Statistics; Parametric statistics; Population; Random effects model; Survival analysis; Average treatment effect; Machine learning; Artificial intelligence; Data mining; Mathematics; Medicine; Propensity score matching; Engineering","score_opus":0.112018737931895,"score_gpt":0.39302598033528546,"score_spread":0.2810072424033905,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2964417596","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0065254094,0.001891606,0.9879666,0.0004409289,0.0001497679,0.00018718917,0.0013333464,0.001097976,0.0004072178],"genre_scores_gemma":[0.31990448,0.0020892718,0.66523504,0.00085180526,0.00079696975,0.001615054,0.0069177877,0.00055483077,0.002034852],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9929309,0.005046937,0.00024503635,0.0009784971,0.0005850829,0.0002134935],"domain_scores_gemma":[0.9711048,0.024001926,0.0015108396,0.0019743142,0.0011304416,0.00027764603],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.020549217,0.0012794421,0.0019463189,0.0013585201,0.000532346,0.0010796223,0.0023570354,0.0016985274,0.0039327224],"category_scores_gemma":[0.03832948,0.00064061,0.0027544287,0.0016647379,0.0011119938,0.0012126826,0.0010807326,0.003105403,0.0015083561],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001262202,0.00027896502,0.019156015,0.0020070544,0.0011127638,0.00043644564,0.00029177972,0.490398,0.004299866,0.032203734,0.025904922,0.42264828],"study_design_scores_gemma":[0.00017801463,0.00016966296,0.0051178914,0.00030703176,0.00023879923,0.00035855122,0.000039367977,0.89176965,0.0024168487,0.08986528,0.009462648,0.000076307355],"about_ca_topic_score_codex":0.0031243656,"about_ca_topic_score_gemma":0.004007761,"teacher_disagreement_score":0.020549217,"about_ca_system_score_codex":0.00073890336,"about_ca_system_score_gemma":0.0013669691,"threshold_uncertainty_score":0.10867596},"labels":[],"label_agreement":null},{"id":"W2964526879","doi":"10.1093/bioinformatics/btz573","title":"TFEA.ChIP: a tool kit for transcription factor binding site enrichment analysis capitalizing on ChIP-seq datasets","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":55,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Child and Family Research Institute; University of British Columbia","funders":"Ministerio de Ciencia e Innovación; Fundación Caja Madrid; Comunidad de Madrid","keywords":"Chip; Transcription factor; Computer science; Computational biology; Software; Factor (programming language); Data mining; Embedded system; Biology; Gene; Genetics; Operating system; Programming language; Telecommunications","score_opus":0.009978782737267027,"score_gpt":0.23460678342288,"score_spread":0.22462800068561298,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2964526879","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0044419738,0.0010778423,0.4407016,0.00018829311,0.0003267427,0.0006446776,0.24392357,0.30560923,0.0030860826],"genre_scores_gemma":[0.019269763,0.00075715996,0.5345486,0.0008599165,0.0001504449,0.0070304857,0.3655061,0.06641904,0.0054586087],"study_design_codex":"not_applicable","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9981529,0.0003599659,0.000184133,0.0006371315,0.00045539075,0.00021049635],"domain_scores_gemma":[0.9977774,0.001115083,0.00022125595,0.00044403522,0.00029131238,0.00015088891],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003685877,0.004111218,0.0030889488,0.0039928243,0.0012873968,0.0020589973,0.004450344,0.0012022342,0.0565092],"category_scores_gemma":[0.006287982,0.002591285,0.0030644056,0.0023347815,0.0006711071,0.0013272371,0.0028847682,0.0035205483,0.04770667],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0015165401,0.00020426336,0.006138945,0.0072067315,0.0018476596,0.000634389,0.00043706203,0.009790645,0.103177756,0.0056427843,0.7460164,0.11738678],"study_design_scores_gemma":[0.0013983694,0.0005210764,0.018324813,0.0007189352,0.00090213475,0.0019146391,0.00019695539,0.09069236,0.15146077,0.027115183,0.7058813,0.00087342976],"about_ca_topic_score_codex":0.0019262193,"about_ca_topic_score_gemma":0.0043047396,"teacher_disagreement_score":0.0565092,"about_ca_system_score_codex":0.0006920989,"about_ca_system_score_gemma":0.0019399538,"threshold_uncertainty_score":0.18904209},"labels":[],"label_agreement":null},{"id":"W2964875256","doi":"10.1093/bioinformatics/btz616","title":"MultiGuideScan: a multi-processing tool for designing CRISPR guide RNA libraries","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"CRISPR and Genetic Engineering","field":"Biochemistry, Genetics and Molecular Biology","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Saskatchewan","funders":"National Key Research and Development Program of China; National Natural Science Foundation of China","keywords":"CRISPR; Cas9; RNA; Guide RNA; Computational biology; Genome; Computer science; Genome editing; Biology; Gene; Genetics","score_opus":0.015178518619465653,"score_gpt":0.29648346503472606,"score_spread":0.2813049464152604,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2964875256","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.01025654,0.0007828907,0.71174747,0.0001994805,0.00016957223,0.0005503861,0.014064292,0.2568637,0.005365613],"genre_scores_gemma":[0.034096204,0.00078756746,0.8914107,0.00039048961,0.00003910645,0.0018060361,0.029046262,0.03279529,0.009628396],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9988531,0.0001411163,0.00011432721,0.0003033485,0.000473546,0.000114564486],"domain_scores_gemma":[0.99900204,0.00048141903,0.00014121408,0.00010774131,0.00016857313,0.00009892915],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0018467789,0.0026691584,0.0017105361,0.0022564866,0.0010188081,0.0014785039,0.0030118423,0.0011791246,0.036614306],"category_scores_gemma":[0.002628283,0.0020309833,0.0015291345,0.0015506801,0.00050411245,0.0012036179,0.0017094811,0.0022282435,0.018410448],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001969522,0.00038749355,0.0030930014,0.004346463,0.0005461302,0.0017984558,0.00075272477,0.018281648,0.3322967,0.010689622,0.1955217,0.43031654],"study_design_scores_gemma":[0.00056133285,0.0004500043,0.0022553622,0.00027486365,0.00022708878,0.0016575215,0.000105619285,0.09749181,0.5934257,0.007058787,0.29612517,0.0003667987],"about_ca_topic_score_codex":0.0011000795,"about_ca_topic_score_gemma":0.001984845,"teacher_disagreement_score":0.036614306,"about_ca_system_score_codex":0.00077760045,"about_ca_system_score_gemma":0.0014978125,"threshold_uncertainty_score":0.12248701},"labels":[],"label_agreement":null},{"id":"W2965351431","doi":"10.1093/bioinformatics/btz593","title":"DEEPCON: protein contact prediction using dilated convolutional neural networks with dropout","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Protein Structure and Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":51,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Ryerson University; University of Missouri; Nvidia; University of Missouri-St. Louis; National Science Foundation","keywords":"Computer science; Convolutional neural network; Dropout (neural networks); Artificial intelligence; Artificial neural network; Residual; Deep learning; Range (aeronautics); Machine learning; Pattern recognition (psychology); Algorithm","score_opus":0.005098687638019777,"score_gpt":0.2017933459060781,"score_spread":0.19669465826805832,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2965351431","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.3914162,0.007154425,0.46884272,0.0022961847,0.0004645541,0.000375886,0.01905712,0.094967745,0.015425149],"genre_scores_gemma":[0.77679247,0.0014881351,0.1620679,0.0011083054,0.00013235121,0.0003851987,0.0418298,0.0013694185,0.01482648],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9994881,0.00007722206,0.000021209979,0.00017367661,0.00016331891,0.00007634788],"domain_scores_gemma":[0.99934083,0.00018497802,0.000092956754,0.00016738071,0.00014818677,0.000065710585],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012231726,0.0018666205,0.00095087045,0.0007773401,0.00065352215,0.00092217006,0.002426972,0.0015148376,0.0035343983],"category_scores_gemma":[0.002317243,0.0005938327,0.000715575,0.000889938,0.00060328917,0.0019143277,0.0013551436,0.0014877936,0.0020643352],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014893416,0.0006762865,0.019802094,0.0010128154,0.0005610972,0.0006319818,0.00016334823,0.51273745,0.03342262,0.014482912,0.15574275,0.25927728],"study_design_scores_gemma":[0.000043764754,0.00006588389,0.0010886536,0.000016961401,0.000019778707,0.00006179434,0.000010853785,0.9827923,0.009347367,0.0030127119,0.0035238045,0.000016079575],"about_ca_topic_score_codex":0.01260954,"about_ca_topic_score_gemma":0.018763028,"teacher_disagreement_score":0.01260954,"about_ca_system_score_codex":0.0017265199,"about_ca_system_score_gemma":0.001598977,"threshold_uncertainty_score":0.025072277},"labels":[],"label_agreement":null},{"id":"W2970462530","doi":"10.1093/bioinformatics/btz671","title":"Automatic local resolution-based sharpening of cryo-EM maps","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Advanced Electron Microscopy Techniques and Applications","field":"Biochemistry, Genetics and Molecular Biology","cited_by":166,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"Ministerio de Economía y Competitividad; European Commission; Comunidad de Madrid","keywords":"Sharpening; Computer science; Visualization; Artificial intelligence; Workflow; Initialization; Resolution (logic); Flexibility (engineering); Interpretability; Cryo-electron microscopy; Algorithm; Data mining; Pattern recognition (psychology); Mathematics; Physics","score_opus":0.005560192124800248,"score_gpt":0.2713040208489383,"score_spread":0.26574382872413804,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2970462530","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.009863468,0.0005669979,0.97950405,0.0001444463,0.000068613204,0.000075988384,0.00022202711,0.00843554,0.0011189167],"genre_scores_gemma":[0.064777344,0.0005522855,0.9288135,0.00016555266,0.000052948842,0.0001295928,0.0008963094,0.002497512,0.0021149071],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9988141,0.00018580887,0.00007276769,0.00032768562,0.0005006272,0.000099043355],"domain_scores_gemma":[0.9981382,0.0005759927,0.00023763574,0.0005044903,0.0004355212,0.00010816061],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0023096795,0.0017078771,0.0011628922,0.0015986733,0.00072159234,0.0018751089,0.0026117058,0.001832123,0.0060851336],"category_scores_gemma":[0.005381951,0.00087164884,0.0014921558,0.0007770623,0.0010124688,0.0020713618,0.0027549672,0.0020622863,0.005121508],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0004901413,0.00019485879,0.0019553313,0.0018389022,0.0002526809,0.00036632633,0.00069607835,0.099567235,0.28880295,0.01864883,0.020727552,0.5664591],"study_design_scores_gemma":[0.000050904433,0.00013351552,0.0012477727,0.00013252073,0.00007782665,0.0006608682,0.00009666655,0.7401582,0.22284907,0.015083176,0.019383706,0.00012586311],"about_ca_topic_score_codex":0.0009169229,"about_ca_topic_score_gemma":0.0011805592,"teacher_disagreement_score":0.0060851336,"about_ca_system_score_codex":0.0005956409,"about_ca_system_score_gemma":0.0010350982,"threshold_uncertainty_score":0.020356774},"labels":[],"label_agreement":null},{"id":"W2970852745","doi":"10.1093/bioinformatics/btz645","title":"DeepCOP: deep learning-based approach to predict gene regulating effects of small molecules","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Machine Learning in Bioinformatics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":33,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"Canadian Institutes of Health Research","keywords":"Deep learning; Computational biology; Computer science; Drug discovery; Artificial intelligence; Genomics; Gene expression; Gene; Gene regulatory network; Machine learning; Biology; Bioinformatics; Genetics; Genome","score_opus":0.00474345889461414,"score_gpt":0.20703925714135216,"score_spread":0.20229579824673802,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2970852745","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.10387715,0.0030615819,0.8682047,0.0010440173,0.00019704213,0.00021778671,0.007837832,0.011279941,0.0042798743],"genre_scores_gemma":[0.6972639,0.001838551,0.27672088,0.0009872323,0.00012340858,0.0007083309,0.014940536,0.00047018714,0.0069470275],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99979717,0.000032752996,0.0000105985555,0.00007343104,0.000055104825,0.00003104839],"domain_scores_gemma":[0.99968696,0.0001485823,0.000040761784,0.00002806707,0.00006576082,0.000029923836],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00055984914,0.0010500106,0.00079212384,0.0007302253,0.00024325511,0.00064168265,0.0011569727,0.0008691674,0.002556434],"category_scores_gemma":[0.0009765862,0.00030486955,0.00081655756,0.0005990087,0.0003905672,0.000506529,0.0007488436,0.001458877,0.00058774714],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005636454,0.00045933426,0.00832048,0.0006582007,0.00037548257,0.0003079632,0.00005549928,0.67511374,0.053174622,0.008022488,0.016840609,0.23610796],"study_design_scores_gemma":[0.000013833305,0.000046844307,0.00030605085,0.000006446891,0.000014818724,0.00001717429,0.0000032726514,0.9911213,0.0046393964,0.002663697,0.0011615327,0.000005618988],"about_ca_topic_score_codex":0.0049499064,"about_ca_topic_score_gemma":0.0067081386,"teacher_disagreement_score":0.0049499064,"about_ca_system_score_codex":0.0009550114,"about_ca_system_score_gemma":0.0011615665,"threshold_uncertainty_score":0.009842157},"labels":[],"label_agreement":null},{"id":"W2971034312","doi":"10.1093/bioinformatics/btz681","title":"breakpointR: an R/Bioconductor package to localize strand state changes in Strand-seq data","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Single-cell and spatial transcriptomics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":60,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia; BC Cancer Agency","funders":"","keywords":"Bioconductor; Computer science; DNA sequencing; Directionality; R package; Computational biology; DNA; Biology; Data mining; Genetics; Computational science; Gene","score_opus":0.02961237499436232,"score_gpt":0.26459468354524496,"score_spread":0.23498230855088265,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2971034312","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0031446847,0.002161398,0.22955789,0.0011707278,0.00086537783,0.0005907721,0.31363255,0.44214255,0.006734083],"genre_scores_gemma":[0.024366217,0.0017388371,0.37266636,0.0021733225,0.0002886298,0.0062156264,0.36183956,0.2191708,0.011540674],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.995171,0.0011857362,0.00047072087,0.0016734883,0.0010868917,0.00041204705],"domain_scores_gemma":[0.9915753,0.0037293378,0.0010063763,0.0014851707,0.0017247746,0.00047910016],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007807156,0.0061940555,0.0044677863,0.0066603827,0.0017544357,0.0047751293,0.0092839645,0.0034891742,0.08836258],"category_scores_gemma":[0.02493602,0.002918456,0.0036017373,0.0060090586,0.001317661,0.0031110088,0.0052848947,0.0058550406,0.11514362],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00076858327,0.00009022302,0.0037857704,0.0061830035,0.0015837117,0.00041122784,0.00039966314,0.0035969976,0.0070584244,0.007892442,0.9372205,0.031009432],"study_design_scores_gemma":[0.000993326,0.00020792919,0.010665867,0.0015428998,0.0013161533,0.0012331527,0.00026209216,0.065188885,0.021782309,0.06421989,0.8320663,0.0005212387],"about_ca_topic_score_codex":0.0053554266,"about_ca_topic_score_gemma":0.0072855484,"teacher_disagreement_score":0.08836258,"about_ca_system_score_codex":0.0015079977,"about_ca_system_score_gemma":0.005142811,"threshold_uncertainty_score":0.29560226},"labels":[],"label_agreement":null},{"id":"W2971052755","doi":"10.1093/bioinformatics/btz673","title":"A Gaussian process-based definition reveals new and bona fide genetic interactions compared to a multiplicative model in the Gram-negative <i>Escherichia coli</i>","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Saskatchewan; University of Regina","funders":"Natural Sciences and Engineering Research Council of Canada; Canadian Institutes of Health Research","keywords":"Multiplicative function; Epistasis; Computer science; Computational biology; Gene; Biology; Machine learning; Data mining; Genetics; Artificial intelligence; Mathematics","score_opus":0.01868175590371989,"score_gpt":0.26243915728513884,"score_spread":0.24375740138141894,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2971052755","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.60025567,0.0003782094,0.38859776,0.00070511905,0.000059837115,0.00009122705,0.0045638913,0.0019215111,0.0034268168],"genre_scores_gemma":[0.92992646,0.00015073364,0.06514925,0.0001875431,0.000019680821,0.00011725967,0.0035520408,0.00018646503,0.0007105951],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99904233,0.0002556082,0.000051960178,0.00037817738,0.00018520922,0.000086742184],"domain_scores_gemma":[0.9966127,0.002229181,0.00039306277,0.00029883397,0.00030659148,0.00015960596],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0020003181,0.0005643247,0.000656947,0.0012138116,0.00049077976,0.0012589002,0.0010801624,0.00092358486,0.0021820925],"category_scores_gemma":[0.008860595,0.00019777004,0.0009174249,0.0015888966,0.0010167495,0.0013487181,0.0013622021,0.0013778061,0.0003810299],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011099018,0.000331471,0.10634296,0.0012992257,0.00038890407,0.0009314589,0.00049782236,0.60951364,0.07721081,0.11024534,0.011548839,0.08057967],"study_design_scores_gemma":[0.000047373884,0.00011106792,0.02316367,0.00004647771,0.000079662634,0.0003486264,0.000094233175,0.90284944,0.009371849,0.061269514,0.0025548474,0.000063168096],"about_ca_topic_score_codex":0.0050947787,"about_ca_topic_score_gemma":0.00511023,"teacher_disagreement_score":0.0050947787,"about_ca_system_score_codex":0.0008487348,"about_ca_system_score_gemma":0.0010000579,"threshold_uncertainty_score":0.010578811},"labels":[],"label_agreement":null},{"id":"W2971405760","doi":"10.1093/bioinformatics/btz699","title":"Protein–protein interaction site prediction through combining local and global features with deep neural networks","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":296,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Saskatchewan","funders":"Higher Education Discipline Innovation Project; National Natural Science Foundation of China; Fundamental Research Funds for the Central Universities; Central South University; Virginia Commonwealth University","keywords":"Computer science; Convolutional neural network; Sequence (biology); Artificial intelligence; Deep learning; Artificial neural network; Source code; Code (set theory); Protein sequencing; Sliding window protocol; Machine learning; Peptide sequence; Window (computing); Biology","score_opus":0.0047272565283904555,"score_gpt":0.20812378998873063,"score_spread":0.20339653346034017,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2971405760","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.18996358,0.009304972,0.74921566,0.0021028216,0.0003572399,0.00022124614,0.013410763,0.02396571,0.011458047],"genre_scores_gemma":[0.7400098,0.0034641689,0.21050367,0.0011468913,0.00026913214,0.00036152985,0.033787236,0.00086125487,0.009596449],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9995492,0.000054375123,0.000023608296,0.000178896,0.0001315584,0.00006233413],"domain_scores_gemma":[0.99961865,0.000101959566,0.00007216679,0.000049499,0.00009640733,0.000061255276],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005781298,0.0016966527,0.0012001415,0.0013043414,0.00039481907,0.0008674353,0.0015705021,0.0010781541,0.0041884845],"category_scores_gemma":[0.0014474468,0.0005650438,0.001112911,0.0012218026,0.00047128845,0.0016270115,0.0014686729,0.001969265,0.0023097927],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009281622,0.0006558078,0.016561866,0.0013872759,0.0006550044,0.0007394903,0.00010570433,0.409487,0.061164226,0.010547102,0.06434027,0.43342808],"study_design_scores_gemma":[0.000023565131,0.000038048685,0.0011676216,0.000022156442,0.000046422836,0.000058317313,0.000009761027,0.98436534,0.005607471,0.006543742,0.00209981,0.000017728147],"about_ca_topic_score_codex":0.0051002637,"about_ca_topic_score_gemma":0.0074162986,"teacher_disagreement_score":0.0051002637,"about_ca_system_score_codex":0.0008027167,"about_ca_system_score_gemma":0.0008903141,"threshold_uncertainty_score":0.01401186},"labels":[],"label_agreement":null},{"id":"W2974257601","doi":"10.1093/bioinformatics/btz709","title":"Subtype-specific transcriptional regulators in breast tumors subjected to genetic and epigenetic alterations","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Human Genome Research Institute; Kreftforeningen; National Institute of General Medical Sciences; Canadian Institute for Advanced Research; National Institutes of Health; National Science Foundation","keywords":"Epigenetics; Biology; DNA methylation; Breast cancer; Transcriptional regulation; Transcription factor; Genetics; Computational biology; Gene; ENCODE; Regulation of gene expression; Gene expression; Cancer","score_opus":0.006417791607965352,"score_gpt":0.19851005491516588,"score_spread":0.19209226330720053,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2974257601","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9064704,0.0015117436,0.016673848,0.0001674343,0.000018024588,0.0000542655,0.07147644,0.0016974377,0.0019304301],"genre_scores_gemma":[0.8517173,0.00086592755,0.028597971,0.00019748775,0.000016306994,0.00013875547,0.11673587,0.00036144216,0.0013689542],"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9996767,0.000028636023,0.000019449693,0.00018400126,0.000056400928,0.000034877685],"domain_scores_gemma":[0.9996164,0.00018191565,0.00007843556,0.000042672586,0.000043962915,0.000036567282],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00042582452,0.00025129603,0.00049685506,0.0008144966,0.0002732755,0.0005819616,0.00035091653,0.00027185353,0.002523644],"category_scores_gemma":[0.00093686103,0.00015159123,0.00057116675,0.0014387402,0.00016393485,0.0002491945,0.00046980046,0.00038458474,0.00081619],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0022403817,0.0002546402,0.5136538,0.0018481044,0.0006138605,0.00040828018,0.0004493036,0.057390116,0.33221915,0.002480378,0.0081998315,0.08024217],"study_design_scores_gemma":[0.00010697701,0.00041023386,0.6638676,0.00012961117,0.00066071673,0.0015158263,0.00047581832,0.18147603,0.11601484,0.008247857,0.026995966,0.00009866348],"about_ca_topic_score_codex":0.0028729008,"about_ca_topic_score_gemma":0.0050861095,"teacher_disagreement_score":0.0028729008,"about_ca_system_score_codex":0.00043980015,"about_ca_system_score_gemma":0.00054565305,"threshold_uncertainty_score":0.008442402},"labels":[],"label_agreement":null},{"id":"W2974534360","doi":"10.1093/bioinformatics/btv095","title":"NAIL, a software toolset for inferring, analyzing and visualizing regulatory networks","year":2015,"lang":"en","type":"erratum","venue":"Bioinformatics","topic":"Information and Cyber Security","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"BC Cancer Agency","funders":"","keywords":"Computer science; Software; Visualization; Data mining; Programming language","score_opus":0.024421380029254532,"score_gpt":0.27085366004537587,"score_spread":0.24643228001612133,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2974534360","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0028564024,0.0022250458,0.7123292,0.008139551,0.048316684,0.00020586386,0.069536135,0.13334538,0.023045795],"genre_scores_gemma":[0.016765984,0.004446121,0.6950988,0.0048676487,0.0019065948,0.00076579035,0.08675569,0.038206905,0.15118642],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99823713,0.00021554742,0.0002443409,0.00028120299,0.00094730046,0.00007448452],"domain_scores_gemma":[0.9926783,0.0045868503,0.00026738638,0.00088885514,0.0014248921,0.00015369855],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0023879306,0.002448909,0.0013418639,0.0048682718,0.0015337514,0.003106999,0.003762392,0.0025827498,0.10538534],"category_scores_gemma":[0.02540666,0.0013544274,0.0012803851,0.003956681,0.00094591506,0.0036207254,0.0021934425,0.00302906,0.029978443],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00017324633,0.00003355585,0.0006891493,0.0009181553,0.00004423673,0.00060218835,0.0001626954,0.004731655,0.002542659,0.012692725,0.8802461,0.09716372],"study_design_scores_gemma":[0.000105639425,0.000029796909,0.0005520102,0.00028088424,0.000048901777,0.0006561657,0.00008106943,0.01607468,0.0038284392,0.020149639,0.9581118,0.000080980644],"about_ca_topic_score_codex":0.008985772,"about_ca_topic_score_gemma":0.024260078,"teacher_disagreement_score":0.10538534,"about_ca_system_score_codex":0.0015607302,"about_ca_system_score_gemma":0.0031278965,"threshold_uncertainty_score":0.35254908},"labels":[],"label_agreement":null},{"id":"W2978380288","doi":"10.1093/bioinformatics/btz743","title":"Fast and flexible coarse-grained prediction of protein folding routes using ensemble modeling and evolutionary sequence variation","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Protein Structure and Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"McGill University","funders":"","keywords":"Computer science; Leverage (statistics); Protein folding; Folding (DSP implementation); Data mining; Machine learning; Chemistry","score_opus":0.015492938426154098,"score_gpt":0.23175904906454817,"score_spread":0.21626611063839407,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2978380288","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.20775342,0.0002439377,0.7783643,0.00023401943,0.00005111593,0.00008404461,0.0014537825,0.00942323,0.002392168],"genre_scores_gemma":[0.6869321,0.00019098227,0.30670014,0.00010120836,0.000037710433,0.0002149181,0.003698017,0.0010357125,0.0010891784],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9997284,0.00008650029,0.000012754559,0.00006871712,0.0000706969,0.000032979202],"domain_scores_gemma":[0.9988331,0.00059084466,0.000102974154,0.0002094827,0.00013989881,0.00012369646],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009655172,0.000843217,0.0012501469,0.00093027216,0.0007337487,0.0008147353,0.0013085789,0.0009770204,0.0024587696],"category_scores_gemma":[0.0029236542,0.00050734705,0.0011392148,0.00076839543,0.00042006353,0.0010526151,0.0010149315,0.0011176122,0.00057711866],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001395302,0.00008397604,0.0042246277,0.000056746303,0.00008180816,0.0001101161,0.000051246752,0.96081823,0.004200863,0.0039638546,0.002021482,0.024247494],"study_design_scores_gemma":[0.000006004697,0.0000054249986,0.00016848357,0.0000016114284,0.0000026021062,0.000005510472,0.000002458255,0.99814856,0.00035593112,0.0011693059,0.00013062445,0.000003469748],"about_ca_topic_score_codex":0.011224063,"about_ca_topic_score_gemma":0.015120265,"teacher_disagreement_score":0.011224063,"about_ca_system_score_codex":0.00075616076,"about_ca_system_score_gemma":0.0016579878,"threshold_uncertainty_score":0.02231747},"labels":[],"label_agreement":null},{"id":"W2979095428","doi":"10.1093/bioinformatics/btz716","title":"VikNGS: a C++ variant integration kit for next generation sequencing association analysis","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto; SickKids Foundation; Hospital for Sick Children","funders":"Wellcome Trust; Genome Canada","keywords":"Computer science; Spurious relationship; Data mining; Covariate; Software; R package; Graphical user interface; Genetic association; Visualization; Association test; Sample size determination; Computational biology; Genotype; Statistics; Biology; Machine learning; Operating system; Genetics; Single-nucleotide polymorphism; Computational science; Mathematics","score_opus":0.03261885538734776,"score_gpt":0.26305040613424846,"score_spread":0.2304315507469007,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2979095428","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0050787805,0.0010060244,0.4613841,0.0007028215,0.0010150593,0.0010792749,0.15791923,0.36471683,0.00709791],"genre_scores_gemma":[0.02084266,0.0008259353,0.6840246,0.001473791,0.00028177557,0.0050808564,0.17052379,0.10561804,0.011328578],"study_design_codex":"not_applicable","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9960978,0.0012873317,0.00040918047,0.0009922459,0.00087625,0.00033721564],"domain_scores_gemma":[0.992774,0.0044209543,0.00068748713,0.0010017942,0.00074481976,0.00037100114],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008040396,0.0035386395,0.002920498,0.003433142,0.0012954823,0.0034526752,0.005397295,0.0016828632,0.11208579],"category_scores_gemma":[0.023922505,0.0032960235,0.0029282463,0.003305533,0.0011101955,0.0019116785,0.0041287616,0.0043667173,0.06844548],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0019819692,0.00022642387,0.009502209,0.004410639,0.0017850291,0.0014523447,0.0007973512,0.014047808,0.01633139,0.019182893,0.8139679,0.11631416],"study_design_scores_gemma":[0.002442728,0.00039456974,0.0129283,0.0012042028,0.00093171635,0.0024375224,0.00020217977,0.095014855,0.022849621,0.06925678,0.7916056,0.00073197327],"about_ca_topic_score_codex":0.0033022545,"about_ca_topic_score_gemma":0.006139494,"teacher_disagreement_score":0.11208579,"about_ca_system_score_codex":0.00093799306,"about_ca_system_score_gemma":0.003741878,"threshold_uncertainty_score":0.3749643},"labels":[],"label_agreement":null},{"id":"W2979820532","doi":"10.1093/bioinformatics/btz744","title":"Soft windowing application to improve analysis of high-throughput phenotyping data","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Biomedical Text Mining and Ontologies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":12,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Lunenfeld-Tanenbaum Research Institute; Mount Sinai Hospital; University of Manitoba; Toronto Centre for Phenogenomics; Hospital for Sick Children","funders":"Deutsches Zentrum für Diabetesforschung; National Human Genome Research Institute; National Institutes of Health; Agence Nationale de la Recherche; European Molecular Biology Laboratory","keywords":"Computer science; Throughput; Software; Data mining; Computational biology; Programming language; Biology; Operating system","score_opus":0.015526743440320546,"score_gpt":0.27634203815419134,"score_spread":0.26081529471387077,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2979820532","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.017747931,0.0004322646,0.9293454,0.0004084,0.00020924232,0.00019253597,0.0031763199,0.047881708,0.00060617103],"genre_scores_gemma":[0.05533906,0.00023793723,0.929635,0.00028616446,0.00009937912,0.00063481525,0.004413248,0.008281447,0.0010728928],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9953477,0.0018352509,0.00040307152,0.0011212268,0.0010521631,0.00024061523],"domain_scores_gemma":[0.9762981,0.017035669,0.0014515428,0.0028860716,0.0017783844,0.00055034197],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.012584967,0.0019518046,0.0016347427,0.0026007623,0.0009244692,0.0025657082,0.0025919233,0.0014126457,0.011828153],"category_scores_gemma":[0.047586642,0.0013848639,0.0026003874,0.002650555,0.0008741265,0.0022981807,0.003200082,0.0031468198,0.0041145436],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0028864383,0.00046975267,0.034217622,0.0029385397,0.0022331153,0.0015015965,0.0017879733,0.09751954,0.1640367,0.018211104,0.10711815,0.5670795],"study_design_scores_gemma":[0.000398484,0.00036859262,0.018825663,0.0002123629,0.0003759962,0.0008149876,0.00019352429,0.79595256,0.09165774,0.03743505,0.05344804,0.00031703085],"about_ca_topic_score_codex":0.002725528,"about_ca_topic_score_gemma":0.0034733897,"teacher_disagreement_score":0.012584967,"about_ca_system_score_codex":0.0007490545,"about_ca_system_score_gemma":0.00216099,"threshold_uncertainty_score":0.06655645},"labels":[],"label_agreement":null},{"id":"W2982906277","doi":"10.1093/bioinformatics/btz824","title":"<tt>MDiNE</tt> : a model to estimate differential co-occurrence networks in microbiome studies","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gut microbiota and health","field":"Biochemistry, Genetics and Molecular Biology","cited_by":60,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; HEC Montréal; Jewish General Hospital","funders":"Canadian Institutes of Health Research","keywords":"Covariate; Microbiome; Computer science; Multinomial distribution; Mixture model; Gaussian; Data mining; Statistics; Artificial intelligence; Machine learning; Mathematics; Biology; Bioinformatics","score_opus":0.02227945325592585,"score_gpt":0.33256682864261705,"score_spread":0.3102873753866912,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2982906277","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.019984316,0.00036568812,0.9715561,0.0012951127,0.000099382014,0.000108040826,0.0037816719,0.0012041176,0.0016055529],"genre_scores_gemma":[0.483824,0.000869914,0.49034533,0.0009547869,0.00032235694,0.0015372587,0.00910502,0.00084274076,0.012198513],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9988133,0.0006568157,0.000044285134,0.00029993482,0.00011327051,0.0000724538],"domain_scores_gemma":[0.9940426,0.004479785,0.00045406682,0.0004411272,0.00038919205,0.0001933057],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002940712,0.0006778493,0.0010045053,0.0010909027,0.0006539084,0.001214918,0.0026142,0.0020578268,0.008545858],"category_scores_gemma":[0.016252482,0.00058773876,0.0012615391,0.0014819716,0.0010585795,0.001454355,0.0017239453,0.0018422375,0.0018234643],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00019698666,0.00005261202,0.010029046,0.00022321405,0.00015475543,0.00030332367,0.00016960692,0.85933363,0.0015820032,0.07136496,0.013200262,0.0433896],"study_design_scores_gemma":[0.000019195273,0.000018700293,0.00082027406,0.000024046438,0.000018356373,0.00007724137,0.0000122721185,0.96492916,0.0003041931,0.030293448,0.0034669812,0.00001610706],"about_ca_topic_score_codex":0.012612817,"about_ca_topic_score_gemma":0.014266168,"teacher_disagreement_score":0.012612817,"about_ca_system_score_codex":0.0012005526,"about_ca_system_score_gemma":0.0011830854,"threshold_uncertainty_score":0.028588712},"labels":[],"label_agreement":null},{"id":"W2987116017","doi":"10.1093/bioinformatics/btz834","title":"DiscoRhythm: an easy-to-use web application and R package for discovering rhythmicity","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Circadian rhythm and melatonin","field":"Neuroscience","cited_by":67,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Centre for Addiction and Mental Health","funders":"Canadian Institutes of Health Research; National Institute of Mental Health; Krembil Foundation","keywords":"Computer science; R package; World Wide Web; Software package; Web application; Software; Data science; Programming language","score_opus":0.023196733752914162,"score_gpt":0.2569678781145153,"score_spread":0.23377114436160115,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2987116017","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0050119236,0.0024197763,0.22113016,0.0009847364,0.0008267829,0.000498763,0.23207264,0.5279571,0.009098135],"genre_scores_gemma":[0.03285529,0.0019597681,0.38103476,0.0020186487,0.00047187615,0.0039950265,0.29311717,0.27205515,0.012492377],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99798137,0.00050421816,0.00017325906,0.00066346844,0.0005202372,0.0001574459],"domain_scores_gemma":[0.99574524,0.0024506436,0.000527027,0.0005775989,0.00043059877,0.00026883886],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0034682506,0.003170884,0.0025735046,0.0033493773,0.0011679683,0.0028320246,0.005357807,0.0016907123,0.067888275],"category_scores_gemma":[0.015696362,0.0017963346,0.002797348,0.0026375316,0.0011192715,0.0022190202,0.004421102,0.0031795925,0.0687687],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006930268,0.00006189753,0.0041356757,0.0038204584,0.00059695716,0.0003957594,0.00046713892,0.0041055838,0.008024354,0.005624288,0.9059483,0.06612669],"study_design_scores_gemma":[0.00073112553,0.00020644284,0.009951533,0.0009164922,0.00048713054,0.0011073599,0.00018661587,0.04080062,0.016445842,0.044680495,0.8840823,0.00040393128],"about_ca_topic_score_codex":0.0027983377,"about_ca_topic_score_gemma":0.003941584,"teacher_disagreement_score":0.067888275,"about_ca_system_score_codex":0.0008541556,"about_ca_system_score_gemma":0.0028653995,"threshold_uncertainty_score":0.2271089},"labels":[],"label_agreement":null},{"id":"W2989634691","doi":"10.1093/bioinformatics/btaa500","title":"iSOM-GSN: an integrative approach for transforming multi-omic data into gene similarity networks via self-organizing maps","year":2020,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":24,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Windsor","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Computer science; Dimensionality reduction; Convolutional neural network; Graph; Euclidean distance; Visualization; Grid; Similarity (geometry); Artificial intelligence; Data mining; Pattern recognition (psychology); Theoretical computer science; Mathematics; Image (mathematics)","score_opus":0.029755114328808348,"score_gpt":0.2581663139303418,"score_spread":0.22841119960153344,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2989634691","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0086831385,0.00017428766,0.98391634,0.00014767556,0.00007199775,0.00011191714,0.00084176025,0.0050916057,0.0009612668],"genre_scores_gemma":[0.11591653,0.00022241178,0.8773028,0.00020789383,0.000058901445,0.0005105038,0.0030611863,0.00067127636,0.0020485537],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9995759,0.00011424879,0.00002582164,0.00012550241,0.0001319431,0.000026546457],"domain_scores_gemma":[0.9994924,0.00020805556,0.000048739097,0.00011099662,0.00010919607,0.000030609743],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013243607,0.0010248423,0.00068658945,0.0017136143,0.0004576406,0.0008667343,0.001964408,0.0008814827,0.0029574006],"category_scores_gemma":[0.003026068,0.0005573236,0.0019955812,0.0014247654,0.00057829067,0.0009034709,0.0019799958,0.0012066896,0.0009500441],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00022292607,0.00018973036,0.0053636874,0.000619796,0.000518759,0.00032942725,0.0004276398,0.44840592,0.012442881,0.032486554,0.014448978,0.48454368],"study_design_scores_gemma":[0.000012120535,0.000024172057,0.00057492894,0.000015119765,0.000016852378,0.00003597177,0.00002586688,0.9762974,0.0017870837,0.01872123,0.0024764545,0.000012647177],"about_ca_topic_score_codex":0.0055963434,"about_ca_topic_score_gemma":0.0087665655,"teacher_disagreement_score":0.0055963434,"about_ca_system_score_codex":0.00081035873,"about_ca_system_score_gemma":0.0010209482,"threshold_uncertainty_score":0.0111275315},"labels":[],"label_agreement":null},{"id":"W2990665925","doi":"10.1093/bioinformatics/btz875","title":"Easy-HLA: a validated web application suite to reveal the full details of HLA typing","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"vaccines and immunoinformatics approaches","field":"Biochemistry, Genetics and Molecular Biology","cited_by":25,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University and Génome Québec Innovation Centre","funders":"Institut National de la Santé et de la Recherche Médicale; Agence Nationale de la Recherche; National Institute of Allergy and Infectious Diseases; European Commission","keywords":"Suite; Human leukocyte antigen; Typing; Computer science; Web application; World Wide Web; Biology; Genetics; Antigen; Speech recognition; Geography","score_opus":0.010953251528008443,"score_gpt":0.22960791220949608,"score_spread":0.21865466068148764,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2990665925","genre_codex":"software","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.01802644,0.0011143085,0.20580657,0.00061714626,0.0003357299,0.00078732194,0.07500913,0.69275457,0.005548787],"genre_scores_gemma":[0.13168955,0.00158538,0.32604975,0.002947085,0.0004746509,0.004101698,0.35666603,0.16577335,0.010712558],"study_design_codex":"not_applicable","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9970124,0.00075980276,0.00041273056,0.0006079552,0.0009901316,0.00021691869],"domain_scores_gemma":[0.9907844,0.0056001954,0.0007116747,0.001218076,0.0010991594,0.00058658404],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00563387,0.0029738264,0.001348621,0.003177355,0.00059768587,0.0024480473,0.003134867,0.0014157756,0.027627986],"category_scores_gemma":[0.017370332,0.0014379853,0.0019242337,0.0016100378,0.0005668476,0.002676448,0.0046660258,0.0020175849,0.022408146],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0052385116,0.0010362326,0.034158945,0.004283311,0.0018184393,0.00210006,0.0007922123,0.015719969,0.02789756,0.0071630613,0.579353,0.32043865],"study_design_scores_gemma":[0.0032468983,0.00086746935,0.047003724,0.0014034014,0.0006470867,0.0033948969,0.00033082813,0.25644505,0.066804774,0.032579374,0.5862289,0.0010475599],"about_ca_topic_score_codex":0.0019779399,"about_ca_topic_score_gemma":0.0016705133,"teacher_disagreement_score":0.027627986,"about_ca_system_score_codex":0.00044212345,"about_ca_system_score_gemma":0.00142169,"threshold_uncertainty_score":0.09242481},"labels":[],"label_agreement":null},{"id":"W2991026871","doi":"10.1093/bioinformatics/btz881","title":"SimRVSequences: an R package to simulate genetic sequence data for pedigrees","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canada's Michael Smith Genome Sciences Centre; Simon Fraser University","funders":"Natural Sciences and Engineering Research Council of Canada; Canadian Statistical Sciences Institute; Canadian Institutes of Health Research","keywords":"Pedigree chart; R package; Sequence (biology); Computer science; Sample (material); Data mining; Computational biology; Genetics; Biology; Programming language; Gene","score_opus":0.0579715405356776,"score_gpt":0.33166235434410263,"score_spread":0.273690813808425,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2991026871","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.012006266,0.0010997988,0.63271195,0.0010975742,0.000598551,0.0004302713,0.13180473,0.21275131,0.0074994364],"genre_scores_gemma":[0.0972453,0.0014022341,0.56890666,0.0019372286,0.00032794135,0.003579728,0.15906386,0.15859637,0.008940717],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99773186,0.0011994134,0.00017224171,0.00043715135,0.0003126059,0.0001467918],"domain_scores_gemma":[0.9932822,0.004589778,0.00044937854,0.00081520027,0.0005714076,0.00029200752],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005152772,0.0024437227,0.0022369323,0.0016085368,0.00074482633,0.002196678,0.004516918,0.0015644656,0.061927542],"category_scores_gemma":[0.027298577,0.0021530197,0.0030594713,0.0016773288,0.00078229845,0.0016701961,0.0025292642,0.0034118108,0.040262423],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013319674,0.0001652525,0.02167472,0.0031046,0.0031042919,0.00093390234,0.00087028724,0.16184737,0.004227828,0.03228918,0.6742881,0.096162535],"study_design_scores_gemma":[0.0016591459,0.00024518304,0.005279059,0.000690186,0.0007912078,0.0010556517,0.00016924176,0.46753916,0.0061031333,0.10500076,0.41114673,0.00032057104],"about_ca_topic_score_codex":0.0067917006,"about_ca_topic_score_gemma":0.007161283,"teacher_disagreement_score":0.061927542,"about_ca_system_score_codex":0.0008653636,"about_ca_system_score_gemma":0.0028364072,"threshold_uncertainty_score":0.20716828},"labels":[],"label_agreement":null},{"id":"W2992956341","doi":"10.1093/bioinformatics/btz902","title":"Fusion-Bloom: fusion detection in assembled transcriptomes","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":12,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia; Canada's Michael Smith Genome Sciences Centre","funders":"National Human Genome Research Institute; Natural Sciences and Engineering Research Council of Canada; National Institutes of Health","keywords":"Fusion; Computer science; Bloom; Bloom filter; Computational biology; Transcriptome; Artificial intelligence; Biology; Algorithm; Genetics; Gene expression; Gene","score_opus":0.00669476493309633,"score_gpt":0.21643492601708136,"score_spread":0.20974016108398502,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2992956341","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.12276727,0.0015255861,0.55146766,0.00072985125,0.00059003447,0.0004391568,0.07169244,0.24551547,0.005272526],"genre_scores_gemma":[0.22616519,0.00061171455,0.6227125,0.00058835454,0.00015289015,0.00068636826,0.12749088,0.018780265,0.0028118468],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9972663,0.00037794185,0.00024945196,0.0011562611,0.00072201353,0.00022810196],"domain_scores_gemma":[0.996283,0.0019012995,0.00058916165,0.000598924,0.00041554047,0.00021199962],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004525679,0.0020935265,0.0013719342,0.0035710451,0.0014089403,0.002861081,0.001694697,0.0016399629,0.007240286],"category_scores_gemma":[0.011567561,0.001190555,0.0019687854,0.002292244,0.0008829049,0.00231445,0.0034792954,0.001971599,0.0051394915],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.005344401,0.0004171568,0.08138907,0.0042662523,0.0018985749,0.0023255525,0.002025695,0.03561351,0.3584667,0.02060185,0.17385583,0.3137954],"study_design_scores_gemma":[0.00047429634,0.00065120537,0.039083324,0.0004487589,0.000482583,0.0028680007,0.00057632173,0.44023892,0.34302723,0.04504599,0.12657909,0.0005243597],"about_ca_topic_score_codex":0.0016362625,"about_ca_topic_score_gemma":0.0028148405,"teacher_disagreement_score":0.007240286,"about_ca_system_score_codex":0.001284975,"about_ca_system_score_gemma":0.0018398849,"threshold_uncertainty_score":0.024221182},"labels":[],"label_agreement":null},{"id":"W2996887063","doi":"10.1093/bioinformatics/btz918","title":"MLDSP-GUI: an alignment-free standalone tool with an interactive graphical user interface for DNA sequence comparison and analysis","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Fractal and DNA sequence analysis","field":"Biochemistry, Genetics and Molecular Biology","cited_by":30,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo; Western University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Graphical user interface; Executable; Computer science; Graphical user interface testing; Software; User interface; Open source; Source code; Data mining; Programming language; Python (programming language); User interface design","score_opus":0.013248105958841972,"score_gpt":0.2835636738212264,"score_spread":0.2703155678623844,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2996887063","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.004605366,0.00036589662,0.13664137,0.00034243354,0.00029743926,0.00034795483,0.29059038,0.5613472,0.005461974],"genre_scores_gemma":[0.026326185,0.00039912167,0.22155644,0.0008222615,0.00013429247,0.0023509257,0.64160454,0.09664858,0.010157686],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99858046,0.00021137806,0.0001256097,0.00048277114,0.0004457347,0.00015407546],"domain_scores_gemma":[0.997575,0.0011976945,0.00020817392,0.00036569792,0.00041771898,0.00023563902],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002230752,0.0031637962,0.0022176176,0.0034256736,0.0008865024,0.0021057904,0.0040100743,0.001446345,0.1207807],"category_scores_gemma":[0.006549474,0.0011513397,0.0016453022,0.0023157254,0.0004827676,0.0025963208,0.0029351125,0.0022606386,0.0972199],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008815591,0.0001399761,0.0027237574,0.0018937959,0.00023859652,0.00026983672,0.00019056782,0.0015654506,0.012791605,0.0018988933,0.90269595,0.074710034],"study_design_scores_gemma":[0.0010549308,0.00025576993,0.011493414,0.00064450724,0.0002023409,0.0011167695,0.00018638687,0.049689766,0.044820093,0.015902342,0.8741862,0.0004473765],"about_ca_topic_score_codex":0.002197508,"about_ca_topic_score_gemma":0.002845996,"teacher_disagreement_score":0.1207807,"about_ca_system_score_codex":0.0005971142,"about_ca_system_score_gemma":0.0014673176,"threshold_uncertainty_score":0.40405172},"labels":[],"label_agreement":null},{"id":"W2997453389","doi":"10.1093/bioinformatics/btz951","title":"Multi-omics reveals functional genomic and metabolic mechanisms of milk production and quality in dairy cows","year":2019,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Ruminant Nutrition and Digestive Physiology","field":"Agricultural and Biological Sciences","cited_by":55,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"National Natural Science Foundation of China","keywords":"Metabolomics; Biology; Biotechnology; Ruminant; Rumen; Omics; Metagenomics; Microbiome; Forage; Metabolome; Food science; Crop; Biochemistry; Agronomy; Bioinformatics; Fermentation; Gene","score_opus":0.043010523610600586,"score_gpt":0.24695242706680479,"score_spread":0.2039419034562042,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2997453389","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.96269196,0.008100561,0.016108397,0.00071280514,0.000039369927,0.000052109994,0.010885128,0.00018394236,0.0012258061],"genre_scores_gemma":[0.9568836,0.0031360327,0.026807383,0.00063953805,0.00004633028,0.00010306699,0.011464628,0.00007158576,0.00084783416],"study_design_codex":"bench_or_experimental","study_design_gemma":"observational","domain_scores_codex":[0.9995467,0.000086436004,0.000031340263,0.00015973255,0.00011112424,0.00006454345],"domain_scores_gemma":[0.99972004,0.00005671687,0.00009691299,0.000026382637,0.00007303651,0.000026850674],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008849389,0.00055980077,0.0007369294,0.0010117413,0.00045739944,0.0013268946,0.000346403,0.00068873016,0.00086767465],"category_scores_gemma":[0.00044620308,0.00025497942,0.00084203196,0.0014245252,0.00027451303,0.00065487664,0.000740795,0.00046864044,0.0001970788],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009455605,0.000105762156,0.0964348,0.00073703314,0.00055881526,0.00014013534,0.00024775811,0.002328661,0.86758316,0.0003706032,0.0006441359,0.029903455],"study_design_scores_gemma":[0.00004410348,0.00048812776,0.8193729,0.00019576466,0.0010199151,0.00038045237,0.0010199087,0.022094106,0.14245912,0.0030900415,0.00971226,0.0001232255],"about_ca_topic_score_codex":0.0022368045,"about_ca_topic_score_gemma":0.0025732976,"teacher_disagreement_score":0.0022368045,"about_ca_system_score_codex":0.0006067554,"about_ca_system_score_gemma":0.00056216016,"threshold_uncertainty_score":0.0046800375},"labels":[],"label_agreement":null},{"id":"W2998529923","doi":"10.1093/bioinformatics/btaa039","title":"<tt>incaRNAfbinv 2.0</tt>: a webserver and software with motif control for fragment-based design of RNAs","year":2020,"lang":"en","type":"article","venue":"Bioinformatics","topic":"RNA and protein synthesis mechanisms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; Université du Québec à Montréal","funders":"College of Natural Resources and Sciences, Humboldt State University; Centre National de la Recherche Scientifique; Ministry of Science and Technology","keywords":"Computer science; Web server; Fragment (logic); Software; Motif (music); Computational biology; Operating system; World Wide Web; Biology; Programming language; The Internet; Art","score_opus":0.01839855060877201,"score_gpt":0.21210891677918414,"score_spread":0.19371036617041212,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2998529923","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0041757324,0.0004942853,0.19044209,0.0003684732,0.00029809235,0.00024171628,0.05131031,0.7391404,0.013528928],"genre_scores_gemma":[0.05466777,0.00082520203,0.32517627,0.001658412,0.00015971875,0.003524744,0.18983684,0.39198187,0.032169163],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99919087,0.00012439636,0.00006163472,0.00020587015,0.0002877322,0.00012952555],"domain_scores_gemma":[0.9989643,0.0004966906,0.00011164653,0.00016583488,0.0001509385,0.000110607514],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0018344772,0.0027109152,0.0016199466,0.0013862639,0.0008102479,0.0018650114,0.004412551,0.0015739128,0.15157281],"category_scores_gemma":[0.0032218364,0.0017827944,0.0016848592,0.0010251723,0.00057892874,0.0016080246,0.0021836918,0.0029458923,0.07477859],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013225093,0.00018567215,0.0017462217,0.0019773003,0.00033658487,0.0005252983,0.0002463017,0.005041544,0.03438607,0.01151516,0.83416826,0.10854914],"study_design_scores_gemma":[0.001039242,0.00035196877,0.0037378957,0.00055377686,0.00017657116,0.0011721198,0.000074177806,0.08373362,0.110799395,0.033193674,0.76470464,0.00046293356],"about_ca_topic_score_codex":0.0021521621,"about_ca_topic_score_gemma":0.0028958102,"teacher_disagreement_score":0.15157281,"about_ca_system_score_codex":0.00092802895,"about_ca_system_score_gemma":0.0014055462,"threshold_uncertainty_score":0.5070616},"labels":[],"label_agreement":null},{"id":"W2999299543","doi":"10.1093/bioinformatics/btaa024","title":"Nearest-neighbor Projected-Distance Regression (NPDR) for detecting network interactions with adjustments for multiple tests and confounding","year":2020,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Toronto Metropolitan University","funders":"National Institute of General Medical Sciences; National Institutes of Health","keywords":"Covariate; Feature selection; Computer science; Spurious relationship; Regression; Feature (linguistics); Artificial intelligence; Inference; Machine learning; k-nearest neighbors algorithm; Confounding; Multiple comparisons problem; Data mining; Statistics; Mathematics","score_opus":0.03123330608107756,"score_gpt":0.3010418191188315,"score_spread":0.26980851303775394,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2999299543","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.013212259,0.00028982776,0.98114794,0.00023706953,0.000055164146,0.000095882715,0.0011217317,0.0032731667,0.0005667989],"genre_scores_gemma":[0.26323333,0.00025475898,0.7277516,0.0003228557,0.00010858512,0.0006370859,0.0045976033,0.0012010952,0.0018931676],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9958974,0.0023580384,0.00019976583,0.0007733757,0.0006669055,0.00010448584],"domain_scores_gemma":[0.9782115,0.01627872,0.0011376223,0.0026002368,0.0015439943,0.00022794037],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00838653,0.0009147143,0.0011959694,0.0013907513,0.00060487166,0.000879429,0.0021714196,0.0010209909,0.0054428205],"category_scores_gemma":[0.051098667,0.0005236142,0.0014309414,0.0016784238,0.0006890686,0.0012289375,0.0018038995,0.0020450046,0.0019248604],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00092471245,0.00025947622,0.03979022,0.00084331754,0.00087908143,0.0006125888,0.00027068576,0.2858313,0.004758979,0.023773756,0.03946932,0.6025865],"study_design_scores_gemma":[0.00009098348,0.00007421211,0.0031847674,0.000039627,0.0000526325,0.00017045547,0.000021477294,0.96927017,0.0018168658,0.021261701,0.0039840434,0.000033081888],"about_ca_topic_score_codex":0.0042765494,"about_ca_topic_score_gemma":0.004490437,"teacher_disagreement_score":0.00838653,"about_ca_system_score_codex":0.0005579588,"about_ca_system_score_gemma":0.001314113,"threshold_uncertainty_score":0.04435277},"labels":[],"label_agreement":null},{"id":"W3001892728","doi":"10.1093/bioinformatics/btaa001","title":"GSOAP: a tool for visualization of gene set over-representation analysis","year":2020,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto; Krembil Foundation","funders":"Natural Sciences and Engineering Research Council of Canada; Canada Foundation for Innovation; Cummings Foundation","keywords":"Visualization; Representation (politics); Set (abstract data type); Computer science; Computational biology; Artificial intelligence; Biology; Programming language","score_opus":0.01917918027975292,"score_gpt":0.2852849429940011,"score_spread":0.2661057627142482,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3001892728","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.008483329,0.001242435,0.42433068,0.0015443239,0.000540108,0.00027346553,0.10225535,0.45353475,0.007795552],"genre_scores_gemma":[0.10843348,0.0022886589,0.67203254,0.0014235659,0.00035446513,0.0020411399,0.13129954,0.07350179,0.008624801],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9991098,0.0001835696,0.00007781177,0.00020337991,0.00033064059,0.0000948498],"domain_scores_gemma":[0.9970266,0.0017004034,0.00024092344,0.00032262524,0.0005005956,0.00020884037],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0017841918,0.0023849064,0.0012125815,0.00514085,0.0008862148,0.0027505832,0.0019129699,0.0010355636,0.055260316],"category_scores_gemma":[0.007915567,0.0006943012,0.0017761304,0.0042897533,0.0005440893,0.0021393357,0.0035209218,0.0026608128,0.012138173],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010132238,0.00013560301,0.0063993423,0.0038883332,0.00054197514,0.0009453409,0.0016747733,0.009910045,0.026452744,0.021787005,0.72634804,0.20090361],"study_design_scores_gemma":[0.0005069544,0.00018933034,0.012644898,0.0010915807,0.0003391319,0.0018997325,0.0007968601,0.1364668,0.04681966,0.101603284,0.69719344,0.00044821503],"about_ca_topic_score_codex":0.0039142305,"about_ca_topic_score_gemma":0.0035818238,"teacher_disagreement_score":0.055260316,"about_ca_system_score_codex":0.00060875586,"about_ca_system_score_gemma":0.0016177315,"threshold_uncertainty_score":0.18486416},"labels":[],"label_agreement":null},{"id":"W3002103667","doi":"10.1093/bioinformatics/btaa043","title":"Transfer index, NetUniFrac and some useful shortest path-based distances for community analysis in sequence similarity networks","year":2020,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Montréal","funders":"Natural Sciences and Engineering Research Council of Canada; Fonds Québécois de la Recherche sur la Nature et les Technologies","keywords":"Phylogenetic tree; Phylogenetic network; Similarity (geometry); Horizontal gene transfer; UniFrac; Biology; Evolutionary biology; Phylogenetics; Tree (set theory); Genome; Computational biology; Artificial intelligence; Gene; Genetics; Computer science; Mathematics","score_opus":0.03580043274894369,"score_gpt":0.25190005433598767,"score_spread":0.216099621587044,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3002103667","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.11707697,0.0028298134,0.873815,0.0004717509,0.00010613388,0.0001703879,0.0013559832,0.00049978617,0.0036741558],"genre_scores_gemma":[0.49176607,0.0014736131,0.5018331,0.000108140994,0.00018116833,0.0003470198,0.002296065,0.00012223913,0.0018725877],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9985507,0.00038506679,0.00011987383,0.00029849674,0.0005546554,0.00009113782],"domain_scores_gemma":[0.9938595,0.0030495157,0.0014285501,0.0003605352,0.0009041739,0.00039761444],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016166584,0.001058921,0.0007591513,0.0067907493,0.0008700788,0.0014095817,0.0014184782,0.001232436,0.0017021577],"category_scores_gemma":[0.008851575,0.00022501672,0.00082757114,0.005889086,0.001308905,0.0037555776,0.0012505793,0.0014468004,0.0004977256],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0004730155,0.0003595332,0.029625768,0.0012327656,0.00030364215,0.00083195604,0.0010710713,0.34128097,0.023335846,0.19210689,0.006603435,0.40277514],"study_design_scores_gemma":[0.00002059005,0.00018807994,0.0074488427,0.00007397275,0.000046857218,0.00072433625,0.00023632482,0.83745974,0.0058260174,0.14134675,0.0065445006,0.00008395827],"about_ca_topic_score_codex":0.0024647897,"about_ca_topic_score_gemma":0.0020943147,"teacher_disagreement_score":0.0067907493,"about_ca_system_score_codex":0.0016366988,"about_ca_system_score_gemma":0.00087720866,"threshold_uncertainty_score":0.011875153},"labels":[],"label_agreement":null},{"id":"W3003794061","doi":"10.1093/bioinformatics/btaa049","title":"CoMeBack: DNA methylation array data analysis for co-methylated regions","year":2020,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Epigenetics and DNA Methylation","field":"Biochemistry, Genetics and Molecular Biology","cited_by":75,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"BC Children's Hospital; Pacific Centre for Reproductive Medicine; Simon Fraser University; University of British Columbia","funders":"Canadian Institutes of Health Research","keywords":"DNA methylation; Methylation; DNA; Computational biology; Methylated DNA immunoprecipitation; Genetics; Biology; Computer science; Gene; Gene expression","score_opus":0.07130694085583668,"score_gpt":0.32100795630313245,"score_spread":0.24970101544729578,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3003794061","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.012310854,0.0010233594,0.39062947,0.0007995555,0.00040908167,0.00055522623,0.25597805,0.3345027,0.003791682],"genre_scores_gemma":[0.11320424,0.0006486247,0.49568996,0.0021134666,0.00024024521,0.0035986179,0.33238727,0.045475047,0.0066425405],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99625176,0.0009079061,0.00030744533,0.0012693324,0.0010826752,0.00018085433],"domain_scores_gemma":[0.9938577,0.0034883972,0.0006819474,0.0011294561,0.00059435517,0.00024806743],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005758674,0.0020835595,0.0016649036,0.0035979352,0.00095955306,0.0018124309,0.003113716,0.001777047,0.031488683],"category_scores_gemma":[0.027185962,0.0014989161,0.0017719622,0.0028095574,0.0007235366,0.0019895155,0.0036130631,0.002107718,0.013845296],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0045556366,0.00029856738,0.047512922,0.0066661057,0.0025576502,0.001319249,0.0010755588,0.021414643,0.05939213,0.018346587,0.60803,0.22883096],"study_design_scores_gemma":[0.0019676222,0.00086408743,0.0651345,0.0011918211,0.00097355875,0.0022602086,0.00036141236,0.26189557,0.14057548,0.089811295,0.43408737,0.00087710115],"about_ca_topic_score_codex":0.003577893,"about_ca_topic_score_gemma":0.0065636123,"teacher_disagreement_score":0.031488683,"about_ca_system_score_codex":0.0008651019,"about_ca_system_score_gemma":0.0017538036,"threshold_uncertainty_score":0.10534012},"labels":[],"label_agreement":null},{"id":"W3006429100","doi":"10.1093/bioinformatics/btaa104","title":"Michela<scp>Ngl</scp>o: sculpting protein views on web pages without coding","year":2020,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetics, Bioinformatics, and Biomedical Research","field":"Biochemistry, Genetics and Molecular Biology","cited_by":28,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Eshelman Institute for Innovation, University of North Carolina at Chapel Hill; National Institute for Health and Care Research; Novartis Pharma; Canada Foundation for Innovation; Ontario Ministry of Economic Development and Innovation; Ministero dello Sviluppo Economico; National Institute on Handicapped Research; Kennedy Trust for Rheumatology Research; Genome Canada; AbbVie; European Federation of Pharmaceutical Industries and Associations; Merck KGaA; Innovative Medicines Initiative; Wellcome; Wellcome Trust; Pfizer; Boehringer Ingelheim","keywords":"Coding (social sciences); Computer science; World Wide Web; Computational biology; Biology; Mathematics","score_opus":0.04376914577511182,"score_gpt":0.28878770850128405,"score_spread":0.24501856272617223,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3006429100","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0035420393,0.000461332,0.18363632,0.0032193097,0.0018852443,0.00055248075,0.04636645,0.6803162,0.08002054],"genre_scores_gemma":[0.050270867,0.0012139173,0.16810727,0.0049664066,0.0010579284,0.0016524695,0.1245283,0.50430197,0.14390098],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9977787,0.00033494766,0.00015933327,0.00031369095,0.0010976072,0.00031562464],"domain_scores_gemma":[0.993414,0.0014014003,0.00033462807,0.002445994,0.0011401472,0.0012638394],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00332894,0.0017357863,0.0012822077,0.0018059433,0.002066283,0.0039964877,0.0044640633,0.0018382182,0.28391],"category_scores_gemma":[0.012834818,0.0013287149,0.0015371492,0.0021204452,0.002002325,0.0047387164,0.008148935,0.0027611062,0.25456572],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002084089,0.000047201116,0.00039225048,0.00056111143,0.000020989539,0.00025754608,0.00020103747,0.00039393787,0.0062669646,0.005010835,0.9288148,0.05782481],"study_design_scores_gemma":[0.000104920306,0.000048566988,0.0015329539,0.00032777546,0.000017769937,0.00062298094,0.000088247754,0.0069100363,0.017667454,0.010610698,0.9619339,0.00013465124],"about_ca_topic_score_codex":0.0046337983,"about_ca_topic_score_gemma":0.0061400877,"teacher_disagreement_score":0.28391,"about_ca_system_score_codex":0.0015311794,"about_ca_system_score_gemma":0.002590207,"threshold_uncertainty_score":0.9497736},"labels":[],"label_agreement":null},{"id":"W3008641372","doi":"10.1093/bioinformatics/btaa136","title":"PSORTm: a bacterial and archaeal protein subcellular localization prediction tool for metagenomics data","year":2020,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Microbial Community Ecology and Physiology","field":"Environmental Science","cited_by":18,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Dalhousie University; Simon Fraser University","funders":"Natural Sciences and Engineering Research Council of Canada; Simon Fraser University; Genome British Columbia; Genome Canada","keywords":"Metagenomics; In silico; Computational biology; Identification (biology); Biology; Human Microbiome Project; Microbiome; Source code; Normalization (sociology); Biological classification; Computer science; Bioinformatics; Genetics; Ecology; Gene; Evolutionary biology","score_opus":0.0358246009567333,"score_gpt":0.22478674216198435,"score_spread":0.18896214120525107,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3008641372","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.065723024,0.0013820478,0.6197974,0.00063102704,0.00014782046,0.00040148906,0.041130833,0.2695969,0.0011894362],"genre_scores_gemma":[0.16939694,0.0005522332,0.76948756,0.000273259,0.00007447977,0.00092296273,0.051975735,0.006576877,0.0007399541],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.998728,0.00030850002,0.00014684843,0.00040787045,0.0003472178,0.00006168011],"domain_scores_gemma":[0.9977253,0.0011461305,0.0004497977,0.000307765,0.0002110859,0.00015986233],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003989723,0.0025731248,0.0013997337,0.0035505602,0.0007988072,0.0015544122,0.0018087261,0.0013217158,0.0031997703],"category_scores_gemma":[0.008643083,0.00084274955,0.00234743,0.0022702862,0.0004391883,0.001679179,0.0017932763,0.0016738764,0.0028453602],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0061391843,0.00088074716,0.06436895,0.006745257,0.002688922,0.0026511124,0.0014232451,0.07189058,0.24160793,0.007301292,0.119695924,0.47460696],"study_design_scores_gemma":[0.000435991,0.0007258407,0.021802343,0.00044751941,0.0004239777,0.0013767922,0.00025799434,0.8058559,0.09870395,0.011853359,0.05786335,0.00025302818],"about_ca_topic_score_codex":0.0010265171,"about_ca_topic_score_gemma":0.0013907909,"teacher_disagreement_score":0.003989723,"about_ca_system_score_codex":0.00057900604,"about_ca_system_score_gemma":0.0011082201,"threshold_uncertainty_score":0.021099925},"labels":[],"label_agreement":null},{"id":"W3012898726","doi":"10.1093/bioinformatics/btaa206","title":"Annotation of tandem mass spectrometry data using stochastic neural networks in shotgun proteomics","year":2020,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Advanced Proteomics Techniques and Applications","field":"Chemistry","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Artificial Intelligence in Medicine (Canada)","funders":"","keywords":"Computer science; Benchmark (surveying); Fragmentation (computing); Annotation; Context (archaeology); Shotgun; Artificial intelligence; Tandem mass spectrometry; Pattern recognition (psychology); Mass spectrometry; Chemistry; Biology; Gene","score_opus":0.04870397349116574,"score_gpt":0.29562667999238523,"score_spread":0.2469227065012195,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3012898726","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.14684077,0.00132905,0.82649416,0.0006844499,0.000119609205,0.0001160188,0.004064786,0.018704863,0.0016463387],"genre_scores_gemma":[0.38636085,0.00067295646,0.59855884,0.000352281,0.0000654814,0.00024292507,0.010875903,0.0008286554,0.0020420577],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99798054,0.00066463277,0.00011323146,0.0006487349,0.0004755022,0.00011729379],"domain_scores_gemma":[0.9974847,0.001327695,0.00026959062,0.0003337007,0.00047794718,0.00010639413],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004154878,0.0012601803,0.00084477325,0.0018912661,0.00075079355,0.0013678194,0.0019440361,0.0014939841,0.0018216078],"category_scores_gemma":[0.008181968,0.00046154734,0.0007886356,0.0019755494,0.0007741215,0.0019137064,0.0016446649,0.0011988054,0.0010797997],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001377457,0.00042619815,0.018565143,0.0015831513,0.00051664794,0.0006301219,0.00036669453,0.49091786,0.10823833,0.011222302,0.017085321,0.3490708],"study_design_scores_gemma":[0.000021413935,0.00005199933,0.001323588,0.00003223128,0.00003178682,0.000092725524,0.000045016688,0.9589395,0.026537925,0.010185433,0.0027092039,0.000029102792],"about_ca_topic_score_codex":0.0028582325,"about_ca_topic_score_gemma":0.0037630335,"teacher_disagreement_score":0.004154878,"about_ca_system_score_codex":0.0010753677,"about_ca_system_score_gemma":0.0012002136,"threshold_uncertainty_score":0.021973372},"labels":[],"label_agreement":null},{"id":"W3015062804","doi":"10.1093/bioinformatics/btaa232","title":"CircMiner: accurate and rapid detection of circular RNA through splice-aware pseudo-alignment scheme","year":2020,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Circular RNAs in diseases","field":"Biochemistry, Genetics and Molecular Biology","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia; Simon Fraser University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Scheme (mathematics); splice; Circular RNA; Computer science; RNA; Computational biology; Algorithm; Mathematics; Biology; Genetics; Gene","score_opus":0.019972313764952867,"score_gpt":0.2463254117506706,"score_spread":0.22635309798571773,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3015062804","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.09036268,0.0017012106,0.8580097,0.00040745153,0.00013772679,0.00023284846,0.0036179526,0.04318169,0.002348741],"genre_scores_gemma":[0.16302249,0.000431078,0.82568544,0.00056048215,0.000040317278,0.0005163615,0.006530939,0.0012909542,0.0019219201],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99867284,0.00027564247,0.00011711128,0.0004292672,0.00044448097,0.000060599366],"domain_scores_gemma":[0.998319,0.00081989996,0.0003621586,0.00020535554,0.00024208036,0.00005156396],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0019985721,0.00096943235,0.00074185577,0.0013048048,0.0005188076,0.0010615749,0.0014050377,0.001352517,0.0019028744],"category_scores_gemma":[0.0054329485,0.00059751415,0.00073084317,0.00075986277,0.00064886746,0.0011169156,0.0008289519,0.0011955143,0.0019338774],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0017528539,0.0002714705,0.011040145,0.0020243393,0.00030424062,0.00093831966,0.00048175472,0.044283614,0.60984427,0.008682964,0.02359567,0.29678035],"study_design_scores_gemma":[0.000084805804,0.00034079916,0.0024445278,0.00006020191,0.000037315287,0.0010280125,0.000050074,0.624278,0.35244817,0.0044021695,0.014709731,0.000116244366],"about_ca_topic_score_codex":0.00066738,"about_ca_topic_score_gemma":0.001141309,"teacher_disagreement_score":0.0019985721,"about_ca_system_score_codex":0.00055055966,"about_ca_system_score_gemma":0.0006987433,"threshold_uncertainty_score":0.010569572},"labels":[],"label_agreement":null},{"id":"W3015080146","doi":"10.1093/bioinformatics/btaa228","title":"MaveQuest: a web resource for planning experimental tests of human variant effects","year":2020,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Rare Diseases","field":"Biochemistry, Genetics and Molecular Biology","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Sinai Health System; Lunenfeld-Tanenbaum Research Institute; University of Toronto","funders":"National Institutes of Health; National Human Genome Research Institute; Canada Excellence Research Chairs, Government of Canada","keywords":"Computer science; Resource (disambiguation); Web resource; World Wide Web; Computational biology; Biology","score_opus":0.01608964652797489,"score_gpt":0.2720287293379269,"score_spread":0.255939082809952,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3015080146","genre_codex":"dataset","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.007992463,0.004140133,0.17462538,0.0016766523,0.00084953866,0.00077436806,0.5515771,0.23948415,0.018880328],"genre_scores_gemma":[0.04076327,0.0028772252,0.2380463,0.0031857425,0.00041502135,0.0052239797,0.6086946,0.091846146,0.008947643],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99726295,0.0008337518,0.00028230605,0.0006011721,0.00084272266,0.00017697204],"domain_scores_gemma":[0.9916963,0.006111239,0.00040449973,0.0009065377,0.00043997754,0.00044143002],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0072139697,0.0032276532,0.003604377,0.003290922,0.0016963119,0.0048455405,0.006016106,0.003393563,0.12215351],"category_scores_gemma":[0.01914916,0.0025899815,0.0031766377,0.0028592723,0.0008779679,0.0025581634,0.00376592,0.0048370864,0.0681923],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001786098,0.0002023479,0.00665968,0.004912732,0.00097122265,0.0012383029,0.00034785646,0.007010485,0.021147477,0.013974434,0.8870041,0.054745317],"study_design_scores_gemma":[0.0025291806,0.0002996895,0.00817813,0.00086159224,0.0006425675,0.0025987702,0.0001525328,0.035343327,0.029271122,0.072733395,0.8468803,0.0005093143],"about_ca_topic_score_codex":0.001927984,"about_ca_topic_score_gemma":0.0035255083,"teacher_disagreement_score":0.12215351,"about_ca_system_score_codex":0.0008388099,"about_ca_system_score_gemma":0.0030092387,"threshold_uncertainty_score":0.40864426},"labels":[],"label_agreement":null},{"id":"W3016108274","doi":"10.1093/bioinformatics/btaa243","title":"Pooled variable scaling for cluster analysis","year":2020,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"Natural Sciences and Engineering Research Council of Canada; KU Leuven","keywords":"Cluster (spacecraft); Scaling; Variable (mathematics); Computer science; Statistics; Mathematics; Programming language","score_opus":0.03153041412441573,"score_gpt":0.292185891279953,"score_spread":0.2606554771555373,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3016108274","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00082110113,0.00029715206,0.99720097,0.00014054833,0.00010433774,0.000115397655,0.00024434764,0.0005445188,0.00053174276],"genre_scores_gemma":[0.05454522,0.00066311983,0.9387827,0.00032140993,0.000537876,0.0015669253,0.0015916264,0.0010391827,0.0009519353],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9795725,0.010763131,0.0012022494,0.0043320763,0.0037094415,0.0004205557],"domain_scores_gemma":[0.9654409,0.017145945,0.002493835,0.008475514,0.005898864,0.00054489454],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.018092316,0.0023935998,0.0025712654,0.004422941,0.0018002661,0.0035941745,0.0034239858,0.0018848402,0.0073255175],"category_scores_gemma":[0.08534949,0.0009827511,0.0037943348,0.006996623,0.0036762937,0.003470336,0.005187569,0.004698406,0.003973896],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00038775703,0.00014358453,0.0047033713,0.0023348841,0.001325622,0.00029162833,0.0012567294,0.103095524,0.00791371,0.3416163,0.034617737,0.50231326],"study_design_scores_gemma":[0.000097886325,0.00026307275,0.0034440928,0.00062351464,0.00028385976,0.00029222635,0.00024936008,0.4082718,0.0069558285,0.5313033,0.048036937,0.00017811071],"about_ca_topic_score_codex":0.0018947598,"about_ca_topic_score_gemma":0.0012754344,"teacher_disagreement_score":0.018092316,"about_ca_system_score_codex":0.001750095,"about_ca_system_score_gemma":0.0033158609,"threshold_uncertainty_score":0.0956825},"labels":[],"label_agreement":null},{"id":"W3016821318","doi":"10.1093/bioinformatics/btaa253","title":"ntJoin: Fast and lightweight assembly-guided scaffolding using minimizer graphs","year":2020,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":44,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canada's Michael Smith Genome Sciences Centre","funders":"National Human Genome Research Institute; National Institutes of Health; Genome British Columbia; Genome Canada","keywords":"Sequence assembly; Computer science; Python (programming language); Reference genome; Synteny; Genome; Graph; Programming language; Theoretical computer science; Biology; Genetics; Gene","score_opus":0.03114288380858979,"score_gpt":0.24813345082368043,"score_spread":0.21699056701509065,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3016821318","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.006296914,0.00039017358,0.86574084,0.00017840839,0.00023716397,0.00017504861,0.004558897,0.119818315,0.0026042212],"genre_scores_gemma":[0.042839605,0.0004090328,0.8910504,0.00017800629,0.00007634716,0.0006294396,0.025670262,0.035038136,0.004108811],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9981316,0.00030269488,0.00017693947,0.0005005638,0.0007662137,0.000122055455],"domain_scores_gemma":[0.99679947,0.0014598935,0.00027437467,0.000735809,0.00051609625,0.00021437435],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0021954984,0.0027270035,0.0017381923,0.0018922293,0.0016879275,0.0024532685,0.004139867,0.0015151551,0.022444285],"category_scores_gemma":[0.009555066,0.0016917364,0.002576043,0.0018490709,0.0009222298,0.002312515,0.0034465052,0.0030302943,0.016250286],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0017052362,0.0002952573,0.0038667007,0.004752384,0.00070384145,0.001361288,0.001743738,0.10450478,0.12577829,0.035803106,0.24396014,0.47552526],"study_design_scores_gemma":[0.0006708394,0.0003426182,0.0015075976,0.00042948214,0.00021250005,0.0009967502,0.00032786775,0.62336004,0.09005646,0.06896695,0.21274939,0.00037943968],"about_ca_topic_score_codex":0.0026722294,"about_ca_topic_score_gemma":0.005240405,"teacher_disagreement_score":0.022444285,"about_ca_system_score_codex":0.000784025,"about_ca_system_score_gemma":0.0019352648,"threshold_uncertainty_score":0.07508361},"labels":[],"label_agreement":null},{"id":"W3016851473","doi":"10.1093/bioinformatics/btaa248","title":"Codon optimization: a mathematical programing approach","year":2020,"lang":"en","type":"article","venue":"Bioinformatics","topic":"RNA and protein synthesis mechanisms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":76,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Codon usage bias; Heuristics; Computer science; Genetic code; Solver; Mathematical optimization; Computational biology; Mathematics; Biology; Genetics; Amino acid; Programming language; Gene; Genome","score_opus":0.022471143699864988,"score_gpt":0.23043188396903885,"score_spread":0.20796074026917385,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3016851473","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0021207281,0.00060337206,0.98590463,0.0004285229,0.000049621925,0.00007375046,0.00012465687,0.00012475591,0.010569859],"genre_scores_gemma":[0.13396493,0.002791227,0.8461976,0.00047024668,0.00026557714,0.0009742478,0.00045757522,0.0004163414,0.014462367],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.999141,0.00036837312,0.000039401388,0.00014302575,0.00023257163,0.00007553455],"domain_scores_gemma":[0.99753076,0.0019436022,0.000166632,0.000060121252,0.00024834156,0.000050596565],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0021746124,0.0020500508,0.0010981561,0.0011477704,0.0006290612,0.0018099742,0.0012753851,0.0014508256,0.008375715],"category_scores_gemma":[0.0039395704,0.00070568227,0.0014904921,0.0015526674,0.0014756866,0.0012924746,0.0012451474,0.0025843703,0.0014844178],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000019528505,0.000053502263,0.00024274005,0.00024065012,0.000026795495,0.00009002675,0.00006764765,0.83301735,0.00076665834,0.14207515,0.0024810338,0.020918896],"study_design_scores_gemma":[0.000012604201,0.000024273533,0.000038114642,0.000040592564,0.000012829248,0.000027112108,0.000016580285,0.93453556,0.00040096394,0.0587785,0.006106401,0.000006388723],"about_ca_topic_score_codex":0.0033200446,"about_ca_topic_score_gemma":0.0024137064,"teacher_disagreement_score":0.008375715,"about_ca_system_score_codex":0.001448645,"about_ca_system_score_gemma":0.002026088,"threshold_uncertainty_score":0.028019547},"labels":[],"label_agreement":null},{"id":"W3017020108","doi":"10.1093/bioinformatics/btaa756","title":"HLA predictions from the bronchoalveolar lavage fluid and blood samples of eight COVID-19 patients at the pandemic onset","year":2020,"lang":"en","type":"article","venue":"Bioinformatics","topic":"SARS-CoV-2 and COVID-19 Research","field":"Medicine","cited_by":33,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Genome British Columbia","funders":"National Human Genome Research Institute; National Institutes of Health; Genome British Columbia; Genome Canada","keywords":"Bronchoalveolar lavage; Human leukocyte antigen; Immunology; Allele; Pandemic; Biology; Population; Coronavirus disease 2019 (COVID-19); Allele frequency; Transcriptome; Disease; Gene; Medicine; Antigen; Genetics; Internal medicine; Infectious disease (medical specialty); Gene expression; Lung","score_opus":0.05237294114448221,"score_gpt":0.3052391455710094,"score_spread":0.2528662044265272,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3017020108","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.94422245,0.0011825442,0.016182087,0.0012041061,0.00022633222,0.000084581436,0.03406148,0.0005054402,0.0023310059],"genre_scores_gemma":[0.9353332,0.00028656557,0.008702513,0.00034263733,0.0000703867,0.00006810763,0.054228596,0.00006610528,0.00090191467],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9994159,0.0001338647,0.0000703704,0.00021298228,0.000060093378,0.00010679182],"domain_scores_gemma":[0.9992637,0.00031719962,0.0001070968,0.000085501044,0.0001128954,0.00011353188],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008481745,0.00047102955,0.00044340378,0.0011512553,0.0004904201,0.00087414193,0.000353069,0.0005424336,0.0051025623],"category_scores_gemma":[0.0027937365,0.0001735107,0.0008548961,0.0010364668,0.00016587027,0.0004145482,0.00053514383,0.0006675266,0.0020857381],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001494274,0.0002773381,0.88159364,0.0002072653,0.00046170483,0.0014186791,0.00031015254,0.006248463,0.031601056,0.0005326469,0.015635354,0.06021939],"study_design_scores_gemma":[0.00036103933,0.00070342445,0.77304465,0.00017888645,0.00088966487,0.0052667707,0.002405722,0.14900704,0.017767904,0.0058494764,0.044405233,0.00012035025],"about_ca_topic_score_codex":0.0029489263,"about_ca_topic_score_gemma":0.003679189,"teacher_disagreement_score":0.0051025623,"about_ca_system_score_codex":0.00026157784,"about_ca_system_score_gemma":0.00068933784,"threshold_uncertainty_score":0.017069817},"labels":[],"label_agreement":null},{"id":"W3017871916","doi":"10.1093/bioinformatics/btaa246","title":"SparkINFERNO: a scalable high-throughput pipeline for inferring molecular mechanisms of non-coding genetic variants","year":2020,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":14,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Weston Brain Institute; National Institute on Aging; Alzheimer's Association","keywords":"Genome-wide association study; Computer science; Inference; Pipeline (software); Scalability; SPARK (programming language); Computational biology; Expression quantitative trait loci; Genomics; Data mining; Biology; Gene; Genome; Artificial intelligence; Genetics; Database; Operating system; Single-nucleotide polymorphism; Programming language; Genotype","score_opus":0.017940404636638648,"score_gpt":0.253130555931134,"score_spread":0.23519015129449533,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3017871916","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0129683325,0.0016191064,0.62683684,0.0012158195,0.0004581417,0.0005719594,0.085635185,0.2653049,0.0053897407],"genre_scores_gemma":[0.12931968,0.0014713008,0.5910842,0.0013725703,0.00033448666,0.0019076143,0.23431592,0.035358462,0.004835808],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9980811,0.0002706439,0.00016179886,0.00074729644,0.00054249994,0.00019665781],"domain_scores_gemma":[0.99608153,0.0020141392,0.0003123705,0.00062434643,0.0005229556,0.00044464553],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0045133885,0.0024203819,0.0026400203,0.0023668502,0.001425114,0.003328351,0.0048372536,0.0013420607,0.016589848],"category_scores_gemma":[0.011812692,0.001548862,0.0043339147,0.0022352545,0.0010564785,0.0023666045,0.0038546035,0.0027422232,0.008435351],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0043095006,0.00035935527,0.018608619,0.0045761093,0.0025463619,0.002086207,0.0010148535,0.07915838,0.040647242,0.034003343,0.6221972,0.19049287],"study_design_scores_gemma":[0.002645037,0.00023187304,0.010093062,0.00025508396,0.0005502701,0.0010887252,0.0002722124,0.6068595,0.02807847,0.16034861,0.18911476,0.0004623277],"about_ca_topic_score_codex":0.007713218,"about_ca_topic_score_gemma":0.009724269,"teacher_disagreement_score":0.016589848,"about_ca_system_score_codex":0.0011886331,"about_ca_system_score_gemma":0.005361541,"threshold_uncertainty_score":0.0554986},"labels":[],"label_agreement":null},{"id":"W3023043090","doi":"10.1093/bioinformatics/btaa289","title":"pepFunk: a tool for peptide-centric functional analysis of metaproteomic human gut microbiome studies","year":2020,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Advanced Proteomics Techniques and Applications","field":"Chemistry","cited_by":19,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canadian Institute for Advanced Research; University of Ottawa","funders":"Natural Sciences and Engineering Research Council of Canada; Ontario Genomics Institute; Genome Canada","keywords":"Workflow; KEGG; Computer science; Metagenomics; Computational biology; Source code; In silico; Metaproteomics; Biology; Database; Gene; Biochemistry; Transcriptome; Programming language","score_opus":0.0484184661842432,"score_gpt":0.3062297102970739,"score_spread":0.25781124411283074,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3023043090","genre_codex":"software","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.02018773,0.0019936594,0.33790198,0.00078159623,0.00056428224,0.00059758016,0.1457572,0.4876585,0.0045574782],"genre_scores_gemma":[0.10151488,0.0017575213,0.6564498,0.0012347071,0.00020154077,0.003306662,0.17026025,0.06065996,0.0046147425],"study_design_codex":"not_applicable","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99842227,0.00029457975,0.00017348478,0.0005775886,0.0003680239,0.00016405372],"domain_scores_gemma":[0.99794894,0.0010388858,0.00025712646,0.000282186,0.000269827,0.0002031374],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005104468,0.0035574685,0.0022816467,0.00466146,0.001479075,0.0028434314,0.0030329595,0.0013505194,0.02304715],"category_scores_gemma":[0.006815399,0.00144675,0.0030524342,0.0028097162,0.00095133344,0.0028016916,0.004920355,0.0026820581,0.012105715],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0066250884,0.00067210034,0.026071457,0.01387316,0.0027944595,0.0037633106,0.0025075462,0.025407104,0.1033031,0.01843633,0.5622989,0.2342474],"study_design_scores_gemma":[0.002437475,0.000767202,0.042780984,0.0021387904,0.0009195674,0.004232534,0.00087165006,0.17664595,0.13348481,0.080759145,0.55372995,0.0012319122],"about_ca_topic_score_codex":0.0013667289,"about_ca_topic_score_gemma":0.0015748484,"teacher_disagreement_score":0.02304715,"about_ca_system_score_codex":0.0009109279,"about_ca_system_score_gemma":0.002294961,"threshold_uncertainty_score":0.077100396},"labels":[],"label_agreement":null},{"id":"W3026082741","doi":"10.1093/bioinformatics/btaa529","title":"BioPAX-Parser: parsing and enrichment analysis of BioPAX pathways","year":2020,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":33,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto; University Health Network","funders":"","keywords":"Parsing; Computer science; MIT License; License; Software; Pathway analysis; Programming language; Artificial intelligence; Biology; Operating system; Gene","score_opus":0.017125912921960085,"score_gpt":0.21962636162523322,"score_spread":0.20250044870327313,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3026082741","genre_codex":"software","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.006186116,0.0006939326,0.33674482,0.00058260956,0.0002547588,0.0002916916,0.16868868,0.4794288,0.0071285325],"genre_scores_gemma":[0.039281324,0.0011364213,0.45477545,0.0007120084,0.00010196403,0.001206448,0.44107047,0.054772485,0.0069434866],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99900514,0.0001953265,0.00014052856,0.00032156083,0.00025631808,0.00008123286],"domain_scores_gemma":[0.9977068,0.0011524888,0.00018258688,0.0003800621,0.0004585742,0.00011950873],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002191035,0.002970753,0.00088944414,0.004080728,0.0011400882,0.0029634277,0.0019226804,0.0011372664,0.030856838],"category_scores_gemma":[0.0048770467,0.0011718683,0.0023046064,0.0037303749,0.00058791216,0.002211338,0.0026375167,0.0021547559,0.019961994],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0018014695,0.00032936686,0.0063316794,0.005594326,0.000576355,0.0014911639,0.0008816874,0.008550903,0.044539202,0.0233569,0.7598089,0.14673808],"study_design_scores_gemma":[0.00044427364,0.00019528011,0.013327365,0.00085522217,0.0003282877,0.0016652017,0.00046877077,0.09710224,0.10063624,0.0860792,0.69851685,0.00038100148],"about_ca_topic_score_codex":0.0021014982,"about_ca_topic_score_gemma":0.002051256,"teacher_disagreement_score":0.030856838,"about_ca_system_score_codex":0.0007601047,"about_ca_system_score_gemma":0.0017043235,"threshold_uncertainty_score":0.10322642},"labels":[],"label_agreement":null},{"id":"W3028383715","doi":"10.1093/bioinformatics/btaa542","title":"TaxoNN: ensemble of neural networks on stratified microbiome data for disease prediction","year":2020,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gut microbiota and health","field":"Biochemistry, Genetics and Molecular Biology","cited_by":79,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University Health Network; Princess Margaret Cancer Centre; Hospital for Sick Children; Public Health Ontario; University of Toronto","funders":"Natural Sciences and Engineering Research Council of Canada; Crohn's and Colitis Canada; Leona M. and Harry B. Helmsley Charitable Trust","keywords":"Microbiome; Computer science; Artificial intelligence; Machine learning; Convolutional neural network; Artificial neural network; Ensemble learning; Pattern recognition (psychology); Data mining; Bioinformatics; Biology","score_opus":0.04297981616172603,"score_gpt":0.2717342480828832,"score_spread":0.2287544319211572,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3028383715","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.39103302,0.0040635886,0.5859409,0.0013410177,0.0005929147,0.0003405615,0.0037307478,0.009367837,0.0035894988],"genre_scores_gemma":[0.87567514,0.0005748798,0.11633836,0.00033897502,0.00014269074,0.00022551401,0.0042218445,0.00010178624,0.0023807888],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9994293,0.0002162086,0.00003844407,0.00016709186,0.00007765338,0.00007119489],"domain_scores_gemma":[0.9985807,0.000677398,0.0001289316,0.00016258411,0.00036088168,0.00008946651],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002913961,0.0017259973,0.0011698105,0.0013012901,0.0004564721,0.0007180349,0.0013167096,0.0010310306,0.0012757533],"category_scores_gemma":[0.0045968737,0.0004784279,0.0009857223,0.00092036795,0.00028523227,0.0010357418,0.0010054264,0.0014066248,0.0005113571],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00058748544,0.00025936303,0.02629631,0.00011498353,0.0005514182,0.00017210166,0.00006336975,0.77052146,0.0023516207,0.0008813529,0.004722091,0.19347838],"study_design_scores_gemma":[0.000006510663,0.00004008371,0.00060206983,0.0000073355227,0.000013480928,0.000011635115,0.000004999135,0.9982266,0.0003566275,0.000557239,0.00016857035,0.0000048822076],"about_ca_topic_score_codex":0.017177692,"about_ca_topic_score_gemma":0.016627885,"teacher_disagreement_score":0.017177692,"about_ca_system_score_codex":0.0009446412,"about_ca_system_score_gemma":0.0011069572,"threshold_uncertainty_score":0.03415543},"labels":[],"label_agreement":null},{"id":"W3036062177","doi":"10.1093/bioinformatics/btaa577","title":"MDIPA: a microRNA–drug interaction prediction approach based on non-negative matrix factorization","year":2020,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Computational Drug Discovery Methods","field":"Computer Science","cited_by":25,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Saskatchewan","funders":"","keywords":"Computer science; microRNA; Similarity (geometry); Computational biology; Matrix decomposition; Identification (biology); Data mining; Artificial intelligence; Machine learning; Biology; Gene; Genetics","score_opus":0.027623798108706295,"score_gpt":0.283240223459039,"score_spread":0.2556164253503327,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3036062177","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.016402746,0.0015801325,0.96916884,0.0006558577,0.0001197444,0.00032462482,0.0024830932,0.0073313387,0.0019337399],"genre_scores_gemma":[0.19907741,0.0011369992,0.7906714,0.0004916326,0.00015285594,0.00064889505,0.0052454467,0.00030452234,0.002270904],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9996356,0.0001034409,0.000021797707,0.000087410204,0.00011959609,0.000032058557],"domain_scores_gemma":[0.9993407,0.0003767819,0.0000857039,0.00003135353,0.000114614406,0.00005088715],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008794282,0.0012294685,0.0010583315,0.0010437533,0.00041023837,0.0005436774,0.0010777736,0.0008854188,0.003938068],"category_scores_gemma":[0.0023149918,0.0004011679,0.0011860608,0.00060015335,0.00026545115,0.0005732559,0.0007968378,0.0011580597,0.0012181741],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007381211,0.00042143938,0.006349645,0.001315177,0.00048377397,0.0006834605,0.0001092198,0.49367735,0.031379938,0.007273984,0.040956017,0.41661188],"study_design_scores_gemma":[0.000043034885,0.00008410078,0.00034893243,0.000012290959,0.000024669927,0.00011885849,0.000008168178,0.9918894,0.0022542153,0.002336526,0.0028619855,0.000017809052],"about_ca_topic_score_codex":0.004277004,"about_ca_topic_score_gemma":0.006066368,"teacher_disagreement_score":0.004277004,"about_ca_system_score_codex":0.00048413072,"about_ca_system_score_gemma":0.0013869433,"threshold_uncertainty_score":0.013174176},"labels":[],"label_agreement":null},{"id":"W3037207689","doi":"10.1093/bioinformatics/btaa596","title":"DeepAntigen: a novel method for neoantigen prioritization via 3D genome and deep sparse learning","year":2020,"lang":"en","type":"article","venue":"Bioinformatics","topic":"vaccines and immunoinformatics approaches","field":"Biochemistry, Genetics and Molecular Biology","cited_by":32,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; Computational biology; Prioritization; Genome; Web server; ENCODE; Immunogenicity; Source code; Deep learning; Artificial intelligence; Machine learning; Biology; The Internet; Genetics; Antigen; Gene","score_opus":0.02133889510537646,"score_gpt":0.24811301357242543,"score_spread":0.22677411846704898,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3037207689","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.01540409,0.000237481,0.98041683,0.00023791486,0.0000527846,0.000045677905,0.0002786647,0.0021651986,0.0011613831],"genre_scores_gemma":[0.35891277,0.0005030537,0.6302879,0.00079664856,0.000110614514,0.0002466834,0.0023990346,0.0004987461,0.0062445938],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9997799,0.00003134315,0.000011864408,0.00005849476,0.00007775606,0.000040659033],"domain_scores_gemma":[0.9997398,0.00007608476,0.000041695937,0.00003425402,0.000075138276,0.000033052063],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00042739345,0.0009350201,0.00085114705,0.00087528926,0.0003477578,0.00063033745,0.0017033039,0.0011226465,0.0026589783],"category_scores_gemma":[0.0010301969,0.00051218417,0.0010171133,0.0007118715,0.00042552422,0.001054167,0.0012210528,0.0013948174,0.0006417365],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00016636425,0.00013048742,0.0034012683,0.00018304396,0.0001275284,0.00026282173,0.000102523554,0.57049036,0.021523878,0.010314267,0.012261453,0.381036],"study_design_scores_gemma":[0.000006514182,0.00001115208,0.00008497889,0.000003180465,0.0000055167434,0.000024884212,0.000005420942,0.99534214,0.0014475125,0.0023728465,0.0006915885,0.0000042405695],"about_ca_topic_score_codex":0.008503411,"about_ca_topic_score_gemma":0.010924109,"teacher_disagreement_score":0.008503411,"about_ca_system_score_codex":0.0007722497,"about_ca_system_score_gemma":0.0013242181,"threshold_uncertainty_score":0.016907811},"labels":[],"label_agreement":null},{"id":"W3037253559","doi":"10.1093/bioinformatics/btaa580","title":"ProbeRating: a recommender system to infer binding profiles for nucleic acid-binding proteins","year":2020,"lang":"en","type":"article","venue":"Bioinformatics","topic":"RNA Research and Splicing","field":"Biochemistry, Genetics and Molecular Biology","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"Natural Sciences and Engineering Research Council of Canada; Genome Canada","keywords":"Nucleic acid; Computer science; Recommender system; Computational biology; DNA-binding protein; Chemistry; Biochemistry; World Wide Web; Biology; Transcription factor; Gene","score_opus":0.033628283545524384,"score_gpt":0.2824179649258074,"score_spread":0.248789681380283,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3037253559","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.25350067,0.009098378,0.66858876,0.0024097718,0.00059049035,0.00053022814,0.026600003,0.0316955,0.006986208],"genre_scores_gemma":[0.55831045,0.0020036027,0.3860814,0.0018035143,0.00030769288,0.00039467405,0.041659776,0.00048923667,0.008949688],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9993293,0.00014159565,0.000046677393,0.00029563604,0.00013306731,0.000053650754],"domain_scores_gemma":[0.9985367,0.00089160766,0.0001099564,0.00015151208,0.0002313167,0.00007890491],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008220804,0.0014132027,0.001032495,0.0015723963,0.00042825454,0.0005700486,0.0014497753,0.0017989547,0.003403332],"category_scores_gemma":[0.0035905715,0.00044183165,0.00091003487,0.0011066219,0.0002330514,0.0012852724,0.0006173951,0.0011860201,0.00197333],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0017622076,0.0010765598,0.045063574,0.0015373152,0.00089926523,0.00080589624,0.00035415823,0.20308761,0.039133694,0.0036201004,0.08477074,0.6178889],"study_design_scores_gemma":[0.000051758077,0.00019778026,0.0026635067,0.000031120944,0.00007674469,0.00018545325,0.000046974157,0.9829437,0.0052956566,0.0023209837,0.006142961,0.000043395874],"about_ca_topic_score_codex":0.01560358,"about_ca_topic_score_gemma":0.027913807,"teacher_disagreement_score":0.01560358,"about_ca_system_score_codex":0.0007610219,"about_ca_system_score_gemma":0.000940075,"threshold_uncertainty_score":0.03102547},"labels":[],"label_agreement":null},{"id":"W3041248822","doi":"10.1093/bioinformatics/btaa588","title":"TreeSAPP: the Tree-based Sensitive and Accurate Phylogenetic Profiler","year":2020,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":14,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Genome British Columbia; University of British Columbia","funders":"Natural Sciences and Engineering Research Council of Canada; National Energy Research Scientific Computing Center; Joint Genome Institute; Compute Canada","keywords":"Python (programming language); Phylogenetic tree; Metagenomics; Computer science; Tree (set theory); Rank (graph theory); Taxonomic rank; Software; Genome; Biology; Data mining; Ecology; Gene; Genetics; Mathematics","score_opus":0.018115108358277635,"score_gpt":0.22407693987934757,"score_spread":0.20596183152106995,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3041248822","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.013918619,0.00088104635,0.41193128,0.0005352306,0.00048838987,0.00041982034,0.07920962,0.4858736,0.0067423386],"genre_scores_gemma":[0.07465811,0.001151583,0.6411883,0.0010650017,0.0002285561,0.0017245248,0.14973453,0.120513804,0.0097355675],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99753726,0.00037335916,0.0001882673,0.0007703916,0.0009095503,0.0002211295],"domain_scores_gemma":[0.99693584,0.001160304,0.00047721385,0.0004544825,0.0006938664,0.00027834444],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003112865,0.002566259,0.0018850825,0.0024091534,0.0013570465,0.0034997314,0.0030795245,0.0013081428,0.023707172],"category_scores_gemma":[0.009299323,0.0018853571,0.0019293709,0.002469014,0.00095958274,0.0038934501,0.0029223782,0.0036859156,0.024654446],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0027315966,0.0002867242,0.01282888,0.0032750054,0.0008281414,0.0010911059,0.0022791598,0.0103107765,0.07826337,0.017005958,0.62733454,0.24376464],"study_design_scores_gemma":[0.00051668653,0.00048765415,0.015190829,0.00070291106,0.0004181694,0.0020646686,0.00071125885,0.24336942,0.1127063,0.04960749,0.573427,0.0007975865],"about_ca_topic_score_codex":0.0032058933,"about_ca_topic_score_gemma":0.0038294627,"teacher_disagreement_score":0.023707172,"about_ca_system_score_codex":0.0007380135,"about_ca_system_score_gemma":0.002312325,"threshold_uncertainty_score":0.07930845},"labels":[],"label_agreement":null},{"id":"W3042330596","doi":"10.1093/bioinformatics/btaa447","title":"EvoLSTM: context-dependent models of sequence evolution using a sequence-to-sequence LSTM","year":2020,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"Compute Canada; Genome Canada","keywords":"Sequence (biology); Context (archaeology); Computer science; Alignment-free sequence analysis; Inference; Mutation; Code (set theory); Probabilistic logic; Sequence learning; Artificial intelligence; Computational biology; Sequence alignment; Biology; Genetics; Peptide sequence; Programming language","score_opus":0.0826937897695777,"score_gpt":0.2807156309608377,"score_spread":0.19802184119125998,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3042330596","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.110448726,0.0012716645,0.8471849,0.0011046017,0.00043612928,0.0001584657,0.0070372806,0.026935624,0.0054225973],"genre_scores_gemma":[0.7191204,0.0007015055,0.26012433,0.000561527,0.00010526774,0.0004692326,0.01010354,0.0017743936,0.007039815],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99979573,0.000041736304,0.000013013684,0.00008506066,0.00004063456,0.000023859275],"domain_scores_gemma":[0.999589,0.00021680605,0.00003184444,0.000049886366,0.00007837972,0.00003410986],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00069397467,0.00096026866,0.0006730382,0.0004325461,0.00035823204,0.0008232539,0.0025592076,0.0013695301,0.0060418323],"category_scores_gemma":[0.0026622366,0.0005893739,0.00075787987,0.00058701937,0.00047475603,0.001910436,0.0011355003,0.0019069496,0.0018343235],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00026460504,0.00009168982,0.0020638248,0.0002632579,0.00014975756,0.00022729942,0.00010789697,0.88845956,0.009847216,0.008092942,0.010540942,0.07989096],"study_design_scores_gemma":[0.000008886595,0.000015536041,0.000081283644,0.0000058983005,0.000006208768,0.000016791577,0.0000035792339,0.9943509,0.001164621,0.0036685101,0.0006730935,0.000004651088],"about_ca_topic_score_codex":0.0063364115,"about_ca_topic_score_gemma":0.010459571,"teacher_disagreement_score":0.0063364115,"about_ca_system_score_codex":0.0009229572,"about_ca_system_score_gemma":0.001022832,"threshold_uncertainty_score":0.020211935},"labels":[],"label_agreement":null},{"id":"W3042374971","doi":"10.1093/bioinformatics/btaa488","title":"Factorized embeddings learns rich and biologically meaningful embedding spaces using factorized tensor decomposition","year":2020,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Tensor decomposition and applications","field":"Mathematics","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal; Mila - Quebec Artificial Intelligence Institute; Institute for Research in Immunology and Cancer","funders":"Canadian Institutes of Health Research; Institut de Valorisation des Données; Canada First Research Excellence Fund; Canadian Institute for Advanced Research","keywords":"Representation (politics); Computer science; Tensor (intrinsic definition); Dimensionality reduction; Embedding; Curse of dimensionality; Rank (graph theory); Sample (material); Artificial intelligence; Multifactor dimensionality reduction; Matrix decomposition; Theoretical computer science; Pattern recognition (psychology); Gene; Mathematics; Biology","score_opus":0.08759766551448768,"score_gpt":0.35351002339032034,"score_spread":0.2659123578758327,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3042374971","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.038631156,0.0005893686,0.95418376,0.00056031113,0.00008634449,0.00006623008,0.000864756,0.0034209806,0.0015971196],"genre_scores_gemma":[0.516851,0.00083821284,0.46449333,0.0005652644,0.00018358362,0.0003038553,0.0072995727,0.0011056726,0.008359422],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99933535,0.00020733777,0.000034983408,0.00024509564,0.00010197092,0.00007514699],"domain_scores_gemma":[0.99803823,0.00089944166,0.00017798367,0.0004340486,0.0003311072,0.00011916483],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012929911,0.0018933051,0.0010111328,0.00073129777,0.00040736742,0.0011444499,0.0010702168,0.0013687393,0.0033217806],"category_scores_gemma":[0.0063970257,0.0006716836,0.001403846,0.0008424195,0.0011212277,0.002317414,0.0014999941,0.0028751802,0.0017771147],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00028558957,0.00018981488,0.003595238,0.00028177528,0.00019139395,0.00015160056,0.00024476706,0.6937393,0.010194982,0.028159276,0.017825495,0.24514085],"study_design_scores_gemma":[0.000013548524,0.000032188927,0.00018907753,0.000016665479,0.000010045438,0.000028440541,0.000016606871,0.9771047,0.0011410292,0.020350488,0.0010846466,0.0000125511115],"about_ca_topic_score_codex":0.0060960306,"about_ca_topic_score_gemma":0.0063784155,"teacher_disagreement_score":0.0060960306,"about_ca_system_score_codex":0.000929834,"about_ca_system_score_gemma":0.0011223605,"threshold_uncertainty_score":0.012121081},"labels":[],"label_agreement":null},{"id":"W3042645205","doi":"10.1093/bioinformatics/btaa442","title":"AITL: Adversarial Inductive Transfer Learning with input and output space adaptation for pharmacogenomics","year":2020,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Machine Learning in Bioinformatics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":60,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia; Simon Fraser University","funders":"Canadian Institutes of Health Research; Terry Fox Foundation; Canada Foundation for Innovation","keywords":"Pharmacogenomics; Adaptation (eye); Adversarial system; Computer science; Transfer of learning; Space (punctuation); Artificial intelligence; Machine learning; Biology; Bioinformatics; Neuroscience","score_opus":0.01805784941033034,"score_gpt":0.23809088593746033,"score_spread":0.22003303652713,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3042645205","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.01441342,0.0009772173,0.9797505,0.00081301853,0.00009619995,0.00007666569,0.00021031527,0.0019387729,0.0017238278],"genre_scores_gemma":[0.8491237,0.0007539916,0.14028224,0.0015720815,0.00021781241,0.00045894997,0.0012966846,0.00028919923,0.006005332],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99881506,0.0005475108,0.000055639517,0.00024757,0.0002310306,0.00010324999],"domain_scores_gemma":[0.996051,0.0029248116,0.00024550178,0.00029054412,0.00034724385,0.00014087942],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0026766988,0.0015629245,0.0010596965,0.0006998372,0.00038160174,0.000872336,0.0023672758,0.0016684311,0.002742462],"category_scores_gemma":[0.008658197,0.00045648578,0.0009652862,0.0007931477,0.0017007435,0.0016768737,0.0027233632,0.0036975855,0.0008435112],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00012708834,0.00012278675,0.001208947,0.00010266117,0.000085471795,0.00011434434,0.00006730758,0.9220305,0.0016026841,0.0064179273,0.0033458183,0.06477435],"study_design_scores_gemma":[0.0000047123385,0.000027689139,0.00005332479,0.000006302926,0.0000046770538,0.000010895321,0.0000036310064,0.9926347,0.0004403153,0.006583239,0.00022585415,0.000004714015],"about_ca_topic_score_codex":0.0034081114,"about_ca_topic_score_gemma":0.001891509,"teacher_disagreement_score":0.0034081114,"about_ca_system_score_codex":0.0012422156,"about_ca_system_score_gemma":0.0011715925,"threshold_uncertainty_score":0.014155865},"labels":[],"label_agreement":null},{"id":"W3042874372","doi":"10.1093/bioinformatics/btaa456","title":"Graph neural representational learning of RNA secondary structures for predicting RNA-protein interactions","year":2020,"lang":"en","type":"article","venue":"Bioinformatics","topic":"RNA and protein synthesis mechanisms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":53,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; Mila - Quebec Artificial Intelligence Institute","funders":"Institut de Valorisation des Données","keywords":"In silico; Computer science; RNA; Source code; Graph; Sequence (biology); Computational biology; RNase P; Artificial intelligence; Machine learning; Representation (politics); Exploit; Theoretical computer science; Biology; Genetics; Gene; Programming language","score_opus":0.01989641482810419,"score_gpt":0.26378689942756817,"score_spread":0.24389048459946397,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3042874372","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.13917398,0.0015743913,0.84622926,0.0008838595,0.00016377254,0.00013855085,0.0020657252,0.0056552934,0.0041152886],"genre_scores_gemma":[0.80748594,0.0006938336,0.18129484,0.00037001437,0.00012620419,0.00021474056,0.0060916548,0.00029968764,0.0034231213],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99974984,0.000072774266,0.00001008668,0.00008405948,0.000052038697,0.00003109222],"domain_scores_gemma":[0.9990656,0.0006089897,0.00007677645,0.00006835356,0.00013199364,0.00004826329],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00071934547,0.0009777402,0.0005993679,0.0010651889,0.00029017,0.00057640293,0.0011072436,0.00090043765,0.0026417421],"category_scores_gemma":[0.002692216,0.0003044652,0.00057697814,0.0009327622,0.0005254003,0.0010079286,0.0006069891,0.001363137,0.0008353661],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00014074998,0.00010969338,0.0018059366,0.00014207294,0.0000716153,0.000057870264,0.000029298566,0.8906621,0.0042647645,0.0041064606,0.0047426578,0.09386687],"study_design_scores_gemma":[0.0000037123598,0.00001168611,0.000113581584,0.000003593506,0.0000037849331,0.0000038076878,0.0000022528714,0.99616843,0.00046790694,0.0030987847,0.00011990303,0.000002532899],"about_ca_topic_score_codex":0.005493246,"about_ca_topic_score_gemma":0.008695969,"teacher_disagreement_score":0.005493246,"about_ca_system_score_codex":0.000970294,"about_ca_system_score_gemma":0.0007452669,"threshold_uncertainty_score":0.010922551},"labels":[],"label_agreement":null},{"id":"W3043192906","doi":"10.1093/bioinformatics/btaa453","title":"Identification of conserved evolutionary trajectories in tumors","year":2020,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Cancer Genomics and Diagnostics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":22,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Genome British Columbia; University of British Columbia; Simon Fraser University","funders":"National Cancer Institute; National Institutes of Health; Francis Crick Institute","keywords":"Phylogenetic tree; Identification (biology); Conserved sequence; Set (abstract data type); Computational biology; Computer science; Biology; Evolutionary biology; Gene; Genetics; Base sequence; Ecology","score_opus":0.012552316219340496,"score_gpt":0.2282734148533736,"score_spread":0.2157210986340331,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3043192906","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6239643,0.0022163528,0.35600784,0.0007692609,0.00004394966,0.00020536683,0.012423462,0.0017313935,0.0026380476],"genre_scores_gemma":[0.8713383,0.00058821915,0.111561224,0.00009878132,0.000032984353,0.00012922211,0.015191552,0.00023313149,0.0008265969],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9993222,0.00014734273,0.000061303195,0.0002771945,0.00012544327,0.000066476314],"domain_scores_gemma":[0.9971343,0.001456152,0.0005701444,0.00025696043,0.0003969158,0.00018558626],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011542125,0.00048593592,0.0005355939,0.0017305829,0.00051529246,0.0007698019,0.00065847713,0.0004952632,0.0015751611],"category_scores_gemma":[0.005844572,0.0002151557,0.0007274869,0.0018226227,0.00045900958,0.0006710069,0.0006706505,0.00057065795,0.00041142],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010904163,0.00026706085,0.29298785,0.0011854569,0.00045228456,0.0014446994,0.0005861851,0.42655072,0.024716793,0.012818414,0.013550237,0.22434995],"study_design_scores_gemma":[0.000046570764,0.00012963988,0.046512943,0.000114718714,0.00012388328,0.0014651333,0.000232694,0.90988654,0.010287591,0.023249593,0.007898146,0.00005251354],"about_ca_topic_score_codex":0.0054231146,"about_ca_topic_score_gemma":0.006769822,"teacher_disagreement_score":0.0054231146,"about_ca_system_score_codex":0.00083017047,"about_ca_system_score_gemma":0.0012660313,"threshold_uncertainty_score":0.010783136},"labels":[],"label_agreement":null},{"id":"W3043435966","doi":"10.1093/bioinformatics/btaa469","title":"Finding the direct optimal RNA barrier energy and improving pathways with an arbitrary energy model","year":2020,"lang":"en","type":"article","venue":"Bioinformatics","topic":"RNA and protein synthesis mechanisms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Core Research for Evolutional Science and Technology; Ministry of Education, Culture, Sports, Science and Technology; Japan Society for the Promotion of Science; Institute of Genetics; Artificial Intelligence Research Center; National Institute of Advanced Industrial Science and Technology","keywords":"Heuristics; Computer science; Source code; Hamming distance; Code (set theory); Folding (DSP implementation); Energy (signal processing); RNA; Theoretical computer science; Algorithm; Chemistry; Mathematics","score_opus":0.015246508073107491,"score_gpt":0.1968845897067588,"score_spread":0.1816380816336513,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3043435966","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.13825677,0.0015566456,0.84735495,0.00074707874,0.00009028444,0.00012651802,0.0008429083,0.001144277,0.009880492],"genre_scores_gemma":[0.51544344,0.00087143475,0.47698656,0.00020584924,0.000038956754,0.00031746173,0.0012885777,0.00058924087,0.0042584892],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9996462,0.00010575478,0.000016193755,0.0001042038,0.000067434616,0.000060197515],"domain_scores_gemma":[0.99839276,0.0010850732,0.00011400326,0.00013836769,0.00014404286,0.00012578664],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008259738,0.0012573669,0.001037437,0.0013800865,0.0008910563,0.0011089187,0.0015350548,0.0017555242,0.0063544326],"category_scores_gemma":[0.004238181,0.0005410312,0.0015245271,0.0009333209,0.0008515952,0.002049565,0.0014078517,0.0014531534,0.00085070555],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00009763042,0.00008971811,0.0009430121,0.00026291274,0.000033832068,0.000093682494,0.00004511181,0.94974023,0.0026010782,0.0257707,0.0011673382,0.019154873],"study_design_scores_gemma":[0.000019097974,0.000022216727,0.00010641424,0.000019834239,0.0000116795345,0.00003116584,0.000018158617,0.97714007,0.00090292096,0.021064678,0.0006539288,0.000009805448],"about_ca_topic_score_codex":0.0049631638,"about_ca_topic_score_gemma":0.004995402,"teacher_disagreement_score":0.0063544326,"about_ca_system_score_codex":0.0011743893,"about_ca_system_score_gemma":0.002413757,"threshold_uncertainty_score":0.021257639},"labels":[],"label_agreement":null},{"id":"W3043467106","doi":"10.1093/bioinformatics/btaa434","title":"Identifying diagnosis-specific genotype–phenotype associations via joint multitask sparse canonical correlation analysis and classification","year":2020,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Functional Brain Connectivity Studies","field":"Neuroscience","cited_by":38,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Institute on Aging; National Institute of Biomedical Imaging and Bioengineering; Canadian Institutes of Health Research; H. Lundbeck A/S; Servier; National Natural Science Foundation of China; Eisai; China Postdoctoral Science Foundation; National Institutes of Health; Northwestern Polytechnical University; Genentech; IXICO; Northern California Institute for Research and Education; Novartis Pharmaceuticals Corporation; BioClinica; Biogen; Pfizer; Northwestern University; University of Pennsylvania; University of Southern California; U.S. Department of Defense; Eli Lilly and Company; Bristol-Myers Squibb; Meso Scale Diagnostics; Alzheimer's Association","keywords":"Canonical correlation; Correlation; Phenotype; Joint (building); Genotype; Computational biology; Computer science; Genetics; Artificial intelligence; Biology; Mathematics; Gene","score_opus":0.12460659311314032,"score_gpt":0.28276698208212503,"score_spread":0.1581603889689847,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3043467106","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.06613004,0.000804496,0.92928106,0.001075214,0.000082234226,0.00008721997,0.00039442434,0.0009978865,0.001147518],"genre_scores_gemma":[0.7916319,0.00060147943,0.20120482,0.00053513603,0.0003143785,0.0002524349,0.0021790667,0.0001685429,0.0031123147],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99805707,0.0007941022,0.00011611099,0.0005382883,0.0003216717,0.00017273317],"domain_scores_gemma":[0.9961915,0.0018090783,0.00045757942,0.0006044477,0.00068697456,0.00025037755],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002683418,0.0012217427,0.0014722561,0.0010881983,0.00054756616,0.00085803645,0.0016868174,0.001345194,0.0012675849],"category_scores_gemma":[0.008654412,0.00050028006,0.0014195708,0.0013063493,0.0009005226,0.00085459807,0.0014344704,0.0015622125,0.00081970595],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00051156746,0.0005132191,0.025109053,0.00034279894,0.00041548713,0.00056856684,0.00027153318,0.5153018,0.009729903,0.014580993,0.016058346,0.41659665],"study_design_scores_gemma":[0.000008998195,0.000016521482,0.0008251143,0.000005075174,0.000017127146,0.00004296698,0.000008579033,0.99474007,0.0005953077,0.0033979246,0.00033197706,0.0000103588945],"about_ca_topic_score_codex":0.0077746785,"about_ca_topic_score_gemma":0.008299542,"teacher_disagreement_score":0.0077746785,"about_ca_system_score_codex":0.000600573,"about_ca_system_score_gemma":0.002054157,"threshold_uncertainty_score":0.015458882},"labels":[],"label_agreement":null},{"id":"W3046214778","doi":"10.1093/bioinformatics/btaa683","title":"Hive Panel Explorer: an interactive network visualization tool","year":2020,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"BC Cancer Agency; Genome British Columbia; University of British Columbia","funders":"Genome British Columbia; Genome Canada","keywords":"Computer science; Visualization; License; Representation (politics); MIT License; Software; Interactive visualization; Function (biology); Data science; Data visualization; Data mining; World Wide Web; Programming language; Biology","score_opus":0.026032707409416673,"score_gpt":0.2550425047605303,"score_spread":0.22900979735111363,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3046214778","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0114187235,0.0012245189,0.40717956,0.0017125998,0.00041436378,0.00052565295,0.082134314,0.48242408,0.01296625],"genre_scores_gemma":[0.1326041,0.0022537424,0.6703477,0.0015803955,0.0003166167,0.0036998193,0.0998901,0.07379344,0.015514068],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.999218,0.00022289982,0.000056444875,0.00015925466,0.00027340592,0.00007002459],"domain_scores_gemma":[0.9952939,0.003623636,0.00017594192,0.00033603082,0.00033098305,0.0002394408],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0022676585,0.0017242588,0.0009852172,0.0026694592,0.00070280745,0.0021166974,0.0024182287,0.0011865128,0.072769],"category_scores_gemma":[0.009301567,0.00092216703,0.0014015877,0.0016152875,0.00057126646,0.003246073,0.004030257,0.002251423,0.011331422],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011156978,0.00019829093,0.0048154336,0.003265373,0.00040707033,0.0010879376,0.0018495186,0.009848303,0.018351872,0.017482877,0.75935787,0.18221974],"study_design_scores_gemma":[0.0008682903,0.00020687537,0.00948163,0.0011742854,0.00021633736,0.0015229628,0.000607136,0.2109416,0.033236887,0.079295866,0.66195107,0.0004970729],"about_ca_topic_score_codex":0.0023285244,"about_ca_topic_score_gemma":0.0032988098,"teacher_disagreement_score":0.072769,"about_ca_system_score_codex":0.00040546825,"about_ca_system_score_gemma":0.0010017,"threshold_uncertainty_score":0.24343657},"labels":[],"label_agreement":null},{"id":"W3047830475","doi":"10.1093/bioinformatics/btaa700","title":"FastBMD: an online tool for rapid benchmark dose–response analysis of transcriptomics data","year":2020,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":45,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"Natural Sciences and Engineering Research Council of Canada; Canada Research Chairs; Genome Canada","keywords":"Benchmark (surveying); Computer science; Transcriptome; Data mining; Biology","score_opus":0.07772542409987128,"score_gpt":0.3109088631813704,"score_spread":0.23318343908149913,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3047830475","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0026004808,0.00071844045,0.34370506,0.0005630226,0.00044589757,0.00062755705,0.15211913,0.49619082,0.003029661],"genre_scores_gemma":[0.023625644,0.00082533807,0.643727,0.0013740917,0.00023518127,0.0059613865,0.20115067,0.11691364,0.00618712],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99144214,0.0019467376,0.0008166688,0.0017424534,0.0035872087,0.00046478785],"domain_scores_gemma":[0.9786981,0.013627416,0.0017224854,0.002446949,0.002873563,0.0006314885],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01592587,0.0045236964,0.003938648,0.005860208,0.0015183679,0.003977318,0.006547988,0.0026341495,0.09442213],"category_scores_gemma":[0.03182748,0.0029120136,0.004035449,0.004306855,0.0011981403,0.0034408025,0.0048882347,0.004247718,0.050864574],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0019753245,0.00031021493,0.0050681997,0.005336093,0.0008386518,0.00057861506,0.00044665585,0.008213363,0.04094696,0.0066408687,0.82121575,0.10842926],"study_design_scores_gemma":[0.0020977654,0.000644087,0.0139941815,0.0012793597,0.00046024192,0.0015735958,0.00020909843,0.102314666,0.11814343,0.034533344,0.7236229,0.0011272884],"about_ca_topic_score_codex":0.0037882673,"about_ca_topic_score_gemma":0.004139081,"teacher_disagreement_score":0.09442213,"about_ca_system_score_codex":0.0018351376,"about_ca_system_score_gemma":0.0043716165,"threshold_uncertainty_score":0.3158735},"labels":[],"label_agreement":null},{"id":"W3080498684","doi":"10.1093/bioinformatics/btaa750","title":"DELPHI: accurate deep ensemble model for protein interaction sites prediction","year":2020,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Protein Structure and Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":167,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University; Western University","funders":"Natural Sciences and Engineering Research Council of Canada; Compute Canada","keywords":"Computer science; Delphi; Deep learning; Delphi method; Ensemble learning; Data mining; Artificial intelligence; Machine learning; Ensemble forecasting; Web server; The Internet","score_opus":0.02283886874922888,"score_gpt":0.25817084603009527,"score_spread":0.23533197728086638,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3080498684","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.11794404,0.0036673131,0.8531819,0.0011666417,0.00021787963,0.00016133531,0.0048720203,0.013863665,0.00492528],"genre_scores_gemma":[0.7328888,0.0013770098,0.23483767,0.0011493317,0.00023763491,0.00048113748,0.018828504,0.0005437002,0.009656181],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99964035,0.00007880746,0.000013665043,0.000104305,0.00009980855,0.00006290778],"domain_scores_gemma":[0.9995339,0.0002040885,0.00004332145,0.000056193916,0.00011097454,0.000051501756],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009666152,0.0013811607,0.0012999851,0.0008187048,0.00036171728,0.0006121542,0.0023455825,0.0012832773,0.0025217575],"category_scores_gemma":[0.0018126416,0.00061733567,0.0009045997,0.0007101064,0.0003029243,0.0012820093,0.0011931803,0.0020630155,0.0008428744],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00029528333,0.00016276223,0.003798926,0.00013235415,0.00023180526,0.00015161648,0.00003790781,0.86264175,0.0031352544,0.002809557,0.020796375,0.105806395],"study_design_scores_gemma":[0.0000065653758,0.000010957931,0.00014417872,0.0000030720569,0.0000055173396,0.0000105641,0.0000020129949,0.99797815,0.0003407991,0.0011530926,0.0003421667,0.0000029768753],"about_ca_topic_score_codex":0.011266169,"about_ca_topic_score_gemma":0.011648443,"teacher_disagreement_score":0.011266169,"about_ca_system_score_codex":0.0009462986,"about_ca_system_score_gemma":0.0011404988,"threshold_uncertainty_score":0.022401214},"labels":[],"label_agreement":null},{"id":"W3085471204","doi":"10.1093/bioinformatics/btaa781","title":"CHESPA/CHESCA-SPARKY: automated NMR data analysis plugins for SPARKY to map protein allostery","year":2020,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Metabolomics and Mass Spectrometry Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":14,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University","funders":"National Institute of General Medical Sciences; National Heart, Lung, and Blood Institute; University of Colorado Denver; Natural Sciences and Engineering Research Council of Canada; Canadian Institutes of Health Research; National Science Foundation","keywords":"Plug-in; Computer science; Visualization; Allosteric regulation; Software; Download; Data mining; Nuclear magnetic resonance; World Wide Web; Programming language; Physics","score_opus":0.047450940346056465,"score_gpt":0.28628587133260136,"score_spread":0.2388349309865449,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3085471204","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.010240573,0.0005572503,0.12562652,0.0008126972,0.00043345214,0.0004583151,0.114513464,0.74223983,0.005117887],"genre_scores_gemma":[0.057305947,0.0012238704,0.38849187,0.0013817454,0.00033921667,0.00285905,0.34152767,0.19967116,0.0071994374],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99883,0.00015891342,0.000092675786,0.00042828053,0.00034745585,0.00014266932],"domain_scores_gemma":[0.99716336,0.0011432124,0.00024619847,0.000596814,0.00044518878,0.0004051946],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0042134686,0.004048233,0.0026087575,0.002377445,0.0015687937,0.0031254087,0.005654544,0.0017305533,0.04492882],"category_scores_gemma":[0.009754101,0.002194333,0.0025064645,0.0021716002,0.0010355351,0.0048032613,0.005312911,0.0036012975,0.026811415],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0015223364,0.0002017271,0.0035555577,0.0016123339,0.00038855095,0.00043319355,0.00041342518,0.004148721,0.01116141,0.005507638,0.9327853,0.03826993],"study_design_scores_gemma":[0.0041342736,0.00036642692,0.014188732,0.00072794134,0.0003095784,0.0014207333,0.0003790886,0.24154474,0.057355247,0.053709507,0.6250208,0.0008428668],"about_ca_topic_score_codex":0.0062355804,"about_ca_topic_score_gemma":0.00968914,"teacher_disagreement_score":0.04492882,"about_ca_system_score_codex":0.0009992536,"about_ca_system_score_gemma":0.003373905,"threshold_uncertainty_score":0.15030187},"labels":[],"label_agreement":null},{"id":"W3089154911","doi":"10.1093/bioinformatics/btaa829","title":"IsoResolve: predicting splice isoform functions by integrating gene and isoform-level features with domain adaptation","year":2020,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":20,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Saskatchewan","funders":"National Cancer Institute; National Institutes of Health; National Institute of Environmental Health Sciences; National Key Research and Development Program of China; Higher Education Discipline Innovation Project; National Natural Science Foundation of China","keywords":"Gene isoform; splice; Adaptation (eye); Alternative splicing; Computational biology; Domain (mathematical analysis); Focus (optics); Gene; Biology; Computer science; Genetics; Neuroscience; Mathematics","score_opus":0.011538196552200689,"score_gpt":0.20273656914766616,"score_spread":0.19119837259546546,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3089154911","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.1481683,0.0028508864,0.74920136,0.00053965615,0.000149342,0.00023176314,0.01743207,0.076639704,0.0047869855],"genre_scores_gemma":[0.40280008,0.0015495189,0.49986354,0.0006013623,0.00009736041,0.0005763605,0.08392477,0.004850307,0.005736754],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99950147,0.00010384747,0.000025043524,0.00021098326,0.000111632224,0.000047068963],"domain_scores_gemma":[0.9992849,0.00038039437,0.000072186325,0.00011663696,0.00009720006,0.00004872308],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015545164,0.0018183136,0.0011328474,0.0019532943,0.00040064697,0.0011606794,0.0021328644,0.000987822,0.003596103],"category_scores_gemma":[0.0025640542,0.00053599075,0.0017844398,0.0012328611,0.0004792995,0.0012224697,0.0012270726,0.0015276064,0.002389941],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0017829852,0.0007076878,0.03894436,0.0015810292,0.00080988224,0.00080738286,0.0003005848,0.3765113,0.05945587,0.010163005,0.07910684,0.42982903],"study_design_scores_gemma":[0.00006794204,0.0001465274,0.0029924347,0.00004638828,0.00007403635,0.00026514498,0.000039915056,0.9621122,0.015414587,0.0071226005,0.011660657,0.000057547884],"about_ca_topic_score_codex":0.0029367523,"about_ca_topic_score_gemma":0.00571724,"teacher_disagreement_score":0.003596103,"about_ca_system_score_codex":0.0006172009,"about_ca_system_score_gemma":0.0009236271,"threshold_uncertainty_score":0.012030125},"labels":[],"label_agreement":null},{"id":"W3092157588","doi":"10.1093/bioinformatics/btaa867","title":"Particle Gibbs sampling for Bayesian phylogenetic inference","year":2020,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Markov chain Monte Carlo; Gibbs sampling; Phylogenetic tree; Computer science; Particle filter; Inference; Tree (set theory); Bayesian inference; Bayesian probability; Markov chain; Sampling (signal processing); Monte Carlo method; Algorithm; Data mining; Theoretical computer science; Artificial intelligence; Mathematics; Machine learning; Statistics; Biology; Combinatorics","score_opus":0.04092308051789186,"score_gpt":0.275039594293375,"score_spread":0.23411651377548312,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3092157588","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0008743641,0.00029687947,0.99718446,0.00016092809,0.000052609845,0.000052199786,0.00016102896,0.0004775949,0.0007398366],"genre_scores_gemma":[0.0978601,0.0014172366,0.89411294,0.00040058172,0.00042407296,0.0008804603,0.0015690649,0.00070729165,0.0026282168],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99767274,0.0012698139,0.00009771105,0.00042340992,0.00043088334,0.00010533754],"domain_scores_gemma":[0.9901839,0.007941609,0.00034694833,0.00060568703,0.00069817144,0.00022374693],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0044816057,0.0013466273,0.0016566679,0.0018366483,0.0010664408,0.0020176107,0.0026962801,0.0019138313,0.008481938],"category_scores_gemma":[0.023539228,0.0011821908,0.0015522802,0.0025662454,0.0021482601,0.002193308,0.0022151354,0.0038515874,0.0023626923],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00014814535,0.00006847046,0.0022085838,0.00053266395,0.00018076798,0.00017503746,0.00015855237,0.5647923,0.0010629249,0.32680318,0.009809979,0.09405945],"study_design_scores_gemma":[0.000037041693,0.000009363152,0.00016338639,0.00004045867,0.000017452694,0.000040256724,0.000008049426,0.8637179,0.00034815742,0.13165025,0.0039505456,0.000017083232],"about_ca_topic_score_codex":0.010269615,"about_ca_topic_score_gemma":0.0072362297,"teacher_disagreement_score":0.010269615,"about_ca_system_score_codex":0.0018019197,"about_ca_system_score_gemma":0.00313313,"threshold_uncertainty_score":0.02837497},"labels":[],"label_agreement":null},{"id":"W3092159560","doi":"10.1093/bioinformatics/btaa882","title":"DoubleRecViz: a web-based tool for visualizing transcript–gene–species tree reconciliation","year":2020,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Sherbrooke","funders":"","keywords":"Phylogenetic tree; Tree (set theory); Visualization; Toolbox; Python (programming language); Phylogenetic network; Biology; Computer science; Phylogenetics; Gene; Computational biology; Data mining; Genetics; Programming language","score_opus":0.040083956390118754,"score_gpt":0.25121907701317914,"score_spread":0.2111351206230604,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3092159560","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.011632452,0.0011414116,0.36264595,0.0008481082,0.00058053236,0.0002982067,0.07924086,0.5356597,0.00795285],"genre_scores_gemma":[0.10227591,0.0014993241,0.6058292,0.0012808015,0.00021127974,0.0020721487,0.12680565,0.15103742,0.008988377],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99905664,0.00016319004,0.000079658945,0.00027179607,0.00031340483,0.000115195726],"domain_scores_gemma":[0.99832803,0.0008518625,0.00020678712,0.00021640312,0.00025206234,0.00014478035],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0023779913,0.0028586567,0.0017601295,0.004536791,0.0009987006,0.0031666998,0.0031076712,0.0020205767,0.052710112],"category_scores_gemma":[0.005804485,0.0013173567,0.0025418499,0.0029285052,0.00056710106,0.003442539,0.0037824037,0.003017873,0.015874537],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001558505,0.00024449654,0.007981116,0.005906671,0.00082572096,0.0016074848,0.0032121579,0.015836518,0.05306857,0.021764375,0.69534713,0.19264728],"study_design_scores_gemma":[0.0006835234,0.00022521848,0.013805919,0.0014189938,0.0003544482,0.0020081077,0.00090082886,0.20052549,0.07092528,0.061675727,0.64666706,0.00080933183],"about_ca_topic_score_codex":0.0031198289,"about_ca_topic_score_gemma":0.003397192,"teacher_disagreement_score":0.052710112,"about_ca_system_score_codex":0.0008011655,"about_ca_system_score_gemma":0.001362418,"threshold_uncertainty_score":0.17633289},"labels":[],"label_agreement":null},{"id":"W3093585277","doi":"10.1093/bioinformatics/btaa906","title":"Leveraging heterogeneous network embedding for metabolic pathway prediction","year":2020,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Genome British Columbia; University of British Columbia","funders":"Natural Sciences and Engineering Research Council of Canada; Genome British Columbia; Compute Canada; Genome Canada","keywords":"Computer science; Embedding; Computational biology; Artificial intelligence; Biology","score_opus":0.017715284303641308,"score_gpt":0.22994934628557556,"score_spread":0.21223406198193426,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3093585277","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.27789897,0.0011904441,0.6897218,0.0007516695,0.000109112174,0.00009887036,0.011964565,0.014417272,0.0038472302],"genre_scores_gemma":[0.77624804,0.000523146,0.19699323,0.00013892143,0.000047302612,0.00012264217,0.023434449,0.00051218097,0.0019801788],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.999721,0.00007342635,0.000012305287,0.000103345315,0.00006159747,0.000028351662],"domain_scores_gemma":[0.9992391,0.00041967825,0.00006928215,0.00009849093,0.00012746963,0.00004599706],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00052797084,0.0010498381,0.0003987842,0.0012938073,0.00028518765,0.00070735277,0.0006067305,0.00046499938,0.0023406814],"category_scores_gemma":[0.0029907126,0.00026676664,0.00067916256,0.0011261662,0.00029375992,0.0013373889,0.00077216874,0.0007642079,0.0007946918],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00047143054,0.00020651348,0.014852157,0.00052053825,0.00024399764,0.00034439945,0.00013198421,0.72897106,0.022781376,0.0089468965,0.013327723,0.20920195],"study_design_scores_gemma":[0.000009166482,0.000033816883,0.0010334037,0.0000144270225,0.000021348575,0.00005256336,0.000016357424,0.983978,0.00498192,0.008157011,0.001690733,0.000011296599],"about_ca_topic_score_codex":0.0036754764,"about_ca_topic_score_gemma":0.004826885,"teacher_disagreement_score":0.0036754764,"about_ca_system_score_codex":0.0004599383,"about_ca_system_score_gemma":0.00052474916,"threshold_uncertainty_score":0.007830322},"labels":[],"label_agreement":null},{"id":"W3095372394","doi":"10.1093/bioinformatics/btaa945","title":"ALeS: adaptive-length spaced-seed design","year":2020,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Metaheuristic Optimization Algorithms Research","field":"Computer Science","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"Natural Sciences and Engineering Research Council of Canada; Compute Canada","keywords":"Computer science; Heuristic; Software; Source code; Computation; Similarity (geometry); Algorithm; Code (set theory); Sensitivity (control systems); Sequence (biology); Data mining; Theoretical computer science; Programming language; Artificial intelligence; Biology","score_opus":0.08403751598559044,"score_gpt":0.27139443497950566,"score_spread":0.1873569189939152,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3095372394","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.023521563,0.0006738083,0.96039164,0.00027561412,0.00012905817,0.00029681355,0.0004698941,0.007922452,0.006319134],"genre_scores_gemma":[0.20692307,0.00030247972,0.7834681,0.00040658537,0.000049661387,0.0007813765,0.00091173867,0.0016717155,0.005485326],"study_design_codex":"simulation_or_modeling","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9989779,0.00028360635,0.00006719501,0.0002530094,0.00031607875,0.00010234102],"domain_scores_gemma":[0.9977496,0.0011262072,0.00028249368,0.00030942974,0.00036995168,0.00016232586],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016053581,0.0011958982,0.00087229186,0.00085780467,0.0006930369,0.000969878,0.002123242,0.0012669698,0.013255127],"category_scores_gemma":[0.005622638,0.0005731103,0.00090561394,0.00075209216,0.0010995996,0.0011262681,0.0014798329,0.0014198918,0.0046362597],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0016657836,0.0005207332,0.0061207633,0.0014124466,0.00018134505,0.0007167296,0.0003632985,0.4344411,0.0865021,0.041333973,0.020410867,0.4063309],"study_design_scores_gemma":[0.00037701832,0.00074026186,0.00066566886,0.00009988343,0.00010182538,0.00045960335,0.00006761661,0.8862899,0.038229167,0.042227134,0.030666271,0.00007573829],"about_ca_topic_score_codex":0.0006572752,"about_ca_topic_score_gemma":0.0010730517,"teacher_disagreement_score":0.013255127,"about_ca_system_score_codex":0.0006284075,"about_ca_system_score_gemma":0.0013530523,"threshold_uncertainty_score":0.044342816},"labels":[],"label_agreement":null},{"id":"W3097039570","doi":"10.1093/bioinformatics/btad167","title":"Finite mixtures of matrix variate Poisson-log normal distributions for three-way count data","year":2023,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bayesian Methods and Mixture Models","field":"Computer Science","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University; Carleton University; University of Guelph","funders":"Natural Sciences and Engineering Research Council of Canada; Canada Research Chairs; University of Toronto; Carleton University","keywords":"Random variate; Cluster analysis; Computer science; Poisson distribution; Markov chain Monte Carlo; Covariance matrix; Algorithm; Gaussian; Mixture model; Data mining; Mathematics; Statistics; Artificial intelligence; Bayesian probability; Random variable","score_opus":0.04948721312213769,"score_gpt":0.31552877440143295,"score_spread":0.26604156127929524,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3097039570","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0044083935,0.0002670478,0.99259746,0.00019828208,0.0000546248,0.00012531927,0.0011555775,0.00084250484,0.0003507316],"genre_scores_gemma":[0.21088864,0.0013221455,0.76477975,0.00058588665,0.00032600595,0.002230993,0.013301829,0.0011304063,0.0054343757],"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.98935026,0.0050775423,0.0006401815,0.0026452064,0.0018135664,0.00047329743],"domain_scores_gemma":[0.9568599,0.032055594,0.0029097092,0.0044145654,0.0031252173,0.00063496217],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.016368702,0.0016733499,0.0024457297,0.0030107014,0.0010741968,0.0037082974,0.006105636,0.003169272,0.0070117027],"category_scores_gemma":[0.06502598,0.0014434727,0.0036698338,0.0045051323,0.0032458077,0.004231021,0.0028142435,0.00469467,0.0032808133],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00050576875,0.0001628634,0.018893413,0.00075810304,0.0003982871,0.0007380264,0.0008686159,0.5964735,0.0031488088,0.25831413,0.012186568,0.107551835],"study_design_scores_gemma":[0.000024950623,0.000030621788,0.0014306747,0.00007962483,0.000031023195,0.0001888668,0.000055156397,0.87848496,0.0006698685,0.11496011,0.003978139,0.000066078435],"about_ca_topic_score_codex":0.011453373,"about_ca_topic_score_gemma":0.009429896,"teacher_disagreement_score":0.016368702,"about_ca_system_score_codex":0.0024925403,"about_ca_system_score_gemma":0.0025198597,"threshold_uncertainty_score":0.086566985},"labels":[],"label_agreement":null},{"id":"W3102269903","doi":"10.1093/bioinformatics/btaa287","title":"DTF: Deep Tensor Factorization for predicting anticancer drug synergy","year":2020,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Computational Drug Discovery Methods","field":"Computer Science","cited_by":85,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"CancerCare Manitoba; Research Institute in Oncology and Hematology; University of Manitoba","funders":"Natural Sciences and Engineering Research Council of Canada; Mitacs","keywords":"Computer science; Artificial intelligence; Machine learning; Logistic regression; Drug; Classifier (UML); Artificial neural network; Data mining; Pharmacology; Biology","score_opus":0.029312767236159363,"score_gpt":0.2798065312882639,"score_spread":0.2504937640521045,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3102269903","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.06606337,0.0060961354,0.9006283,0.0019911174,0.00040829953,0.00031765542,0.008848651,0.011138268,0.0045082653],"genre_scores_gemma":[0.5365198,0.0027053137,0.44106993,0.0006450446,0.0002160849,0.00040570885,0.0128055625,0.0004095699,0.005222923],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99948895,0.00010405634,0.000048914993,0.000116110685,0.00017027969,0.000071596514],"domain_scores_gemma":[0.9989478,0.00043090485,0.00018901167,0.00008888756,0.000252375,0.00009113303],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013645751,0.0019849997,0.0011008374,0.0013297215,0.00031972898,0.000744152,0.0009901716,0.0010155864,0.004064596],"category_scores_gemma":[0.0038534761,0.0005193504,0.0016039051,0.0008704272,0.00038655027,0.0013708195,0.00071289664,0.0014416258,0.001208274],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00056810584,0.00029087946,0.009408165,0.0007728697,0.000318713,0.00030369693,0.000070567585,0.64859134,0.01409739,0.005565049,0.028653597,0.29135963],"study_design_scores_gemma":[0.000016739064,0.00006267718,0.00039342526,0.000017258613,0.000022139542,0.000046554636,0.0000058355713,0.9926421,0.0020451648,0.003146467,0.0015887119,0.000012858849],"about_ca_topic_score_codex":0.0165924,"about_ca_topic_score_gemma":0.019684665,"teacher_disagreement_score":0.0165924,"about_ca_system_score_codex":0.0013733774,"about_ca_system_score_gemma":0.002540838,"threshold_uncertainty_score":0.032991648},"labels":[],"label_agreement":null},{"id":"W3104008342","doi":"10.1093/bioinformatics/btaa973","title":"MetaADEDB 2.0: a comprehensive database on adverse drug events","year":2020,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Pharmacovigilance and Adverse Drug Reactions","field":"Pharmacology, Toxicology and Pharmaceutics","cited_by":14,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"National Key Research and Development Program of China; China Postdoctoral Science Foundation; National Natural Science Foundation of China","keywords":"Adverse Event Reporting System; Computer science; Database; Drug; Online database; Drug reaction; Adverse effect; Information retrieval; World Wide Web; Data mining; Medicine; Pharmacology","score_opus":0.1810652966885001,"score_gpt":0.4259213948965154,"score_spread":0.24485609820801532,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3104008342","genre_codex":"dataset","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0026796511,0.002959221,0.0073366235,0.00094683276,0.00012253861,0.00045517282,0.97026676,0.010894507,0.004338686],"genre_scores_gemma":[0.018317029,0.0029562141,0.02457144,0.0008571651,0.00025170055,0.0018367573,0.94669175,0.002707258,0.0018106945],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99393624,0.0015536672,0.0024348474,0.00075329933,0.0010937682,0.00022822745],"domain_scores_gemma":[0.96920246,0.014652211,0.007497018,0.0030405244,0.003444743,0.0021630004],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0069982037,0.0016397054,0.0033928438,0.021069335,0.00061447575,0.0039960626,0.003347285,0.0016617364,0.06862343],"category_scores_gemma":[0.040347874,0.0013328933,0.0025552195,0.017524403,0.00042666355,0.0032962253,0.003796466,0.001905772,0.024096057],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0030090332,0.00018141262,0.023271555,0.022147793,0.00226421,0.0006274288,0.00047441392,0.0023242803,0.002488575,0.0064232354,0.79819554,0.13859247],"study_design_scores_gemma":[0.0012668555,0.00019886282,0.033402983,0.0031585228,0.0015934833,0.0011314757,0.00020824611,0.0033385365,0.0022577057,0.009670735,0.9434421,0.00033047175],"about_ca_topic_score_codex":0.0046766438,"about_ca_topic_score_gemma":0.004847838,"teacher_disagreement_score":0.06862343,"about_ca_system_score_codex":0.001178845,"about_ca_system_score_gemma":0.0055281715,"threshold_uncertainty_score":0.22956824},"labels":[],"label_agreement":null},{"id":"W3108043540","doi":"10.1093/bioinformatics/btab158","title":"Conditional canonical correlation estimation based on covariates with random forests","year":2021,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Functional Brain Connectivity Studies","field":"Neuroscience","cited_by":28,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal; Centre Hospitalier Universitaire Sainte-Justine; HEC Montréal","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Covariate; Canonical correlation; Correlation; Statistics; Random forest; Mathematics; Computer science; Data mining; Artificial intelligence","score_opus":0.022750724464176218,"score_gpt":0.2514645990948229,"score_spread":0.22871387463064666,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3108043540","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.008553715,0.00036858217,0.9891396,0.00014679416,0.00006669061,0.00008522395,0.00047877317,0.00086581055,0.0002948469],"genre_scores_gemma":[0.24044842,0.0006336551,0.75057274,0.00025171056,0.0003287891,0.00077709084,0.004810475,0.0006058717,0.0015713087],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99462116,0.0032390947,0.00019037107,0.001135203,0.00048469147,0.0003294057],"domain_scores_gemma":[0.98648304,0.008759096,0.00094252074,0.0017185651,0.0016965066,0.0004002824],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0117808515,0.0017706958,0.002145987,0.001686013,0.0012112026,0.0014503311,0.0024122337,0.0012300671,0.003871097],"category_scores_gemma":[0.031349376,0.00086969096,0.0027484873,0.003040842,0.0014575709,0.001932373,0.0019673775,0.0030405198,0.0017293898],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007702115,0.00025856515,0.034766667,0.000618339,0.00093446573,0.00051277457,0.0004411894,0.5484943,0.0039826664,0.068042755,0.025802854,0.3153753],"study_design_scores_gemma":[0.000050565464,0.000041019124,0.0025744291,0.00006213925,0.00008075692,0.00012545461,0.00002697514,0.9407927,0.0009900687,0.05215687,0.0030539378,0.000045058645],"about_ca_topic_score_codex":0.010699096,"about_ca_topic_score_gemma":0.015640559,"teacher_disagreement_score":0.0117808515,"about_ca_system_score_codex":0.0007642221,"about_ca_system_score_gemma":0.003724305,"threshold_uncertainty_score":0.0623039},"labels":[],"label_agreement":null},{"id":"W3111754743","doi":"10.1093/bioinformatics/btaa1019","title":"<i>MorphOT</i> : transport-based interpolation between EM maps with UCSF <i>ChimeraX</i>","year":2020,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Advanced Electron Microscopy Techniques and Applications","field":"Biochemistry, Genetics and Molecular Biology","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Morphing; Computer science; Interpolation (computer graphics); Documentation; Code (set theory); Source code; Software; Metric (unit); Algorithm; Linear interpolation; Computer graphics (images); Data mining; Computational science; Theoretical computer science; Image (mathematics); Artificial intelligence; Programming language; Pattern recognition (psychology); Set (abstract data type)","score_opus":0.009067360275982882,"score_gpt":0.2572769591131348,"score_spread":0.24820959883715193,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3111754743","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.014998873,0.0010714044,0.5226716,0.002434795,0.0008227613,0.00027236823,0.06308262,0.38351616,0.011129476],"genre_scores_gemma":[0.060693815,0.0012444496,0.63011426,0.0010882315,0.0001430106,0.0009807622,0.106877506,0.19165272,0.0072052726],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99879706,0.00021061994,0.00009313788,0.00035771157,0.00041924743,0.00012224427],"domain_scores_gemma":[0.9979582,0.00071348157,0.00021232286,0.0005798711,0.00029313384,0.0002430297],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0029254884,0.0026561487,0.002625331,0.0015585545,0.0024766899,0.0038947375,0.007759442,0.0030406278,0.0743667],"category_scores_gemma":[0.009785608,0.0021417253,0.002209504,0.0021694605,0.0012786527,0.0034345197,0.0050676647,0.005378353,0.03564931],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006452234,0.00019866398,0.0033516877,0.0031908238,0.00051713386,0.000594098,0.0007874311,0.032859474,0.019044932,0.025637146,0.82623,0.0869434],"study_design_scores_gemma":[0.0005282543,0.00013367579,0.0027998427,0.00077260437,0.00019457153,0.0012813903,0.0002588737,0.47609612,0.05964065,0.04437883,0.41350928,0.00040576773],"about_ca_topic_score_codex":0.011968403,"about_ca_topic_score_gemma":0.015295039,"teacher_disagreement_score":0.0743667,"about_ca_system_score_codex":0.0018878363,"about_ca_system_score_gemma":0.0032082286,"threshold_uncertainty_score":0.24878138},"labels":[],"label_agreement":null},{"id":"W3112246030","doi":"10.1093/bioinformatics/btaa1038","title":"Identification of differentially expressed gene modules in heterogeneous diseases","year":2020,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Identification (biology); Gene; Computational biology; Genetics; Biology; Computer science","score_opus":0.01023522611631228,"score_gpt":0.2154059868898163,"score_spread":0.20517076077350402,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3112246030","genre_codex":"empirical","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.49736732,0.0037262917,0.4645072,0.0013644625,0.00011762508,0.00046359748,0.024206387,0.004677117,0.0035700453],"genre_scores_gemma":[0.7760394,0.0009060864,0.20388488,0.0003022687,0.000087740744,0.00039659784,0.01647343,0.00031867833,0.0015909115],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9993794,0.00014342666,0.000040945928,0.00027266284,0.00011975132,0.00004381481],"domain_scores_gemma":[0.9986632,0.00064520387,0.00025771372,0.00014923351,0.00016525575,0.00011940502],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011081626,0.0005960387,0.0005746812,0.002438117,0.00044492015,0.00065235747,0.0005957238,0.00031628038,0.002491278],"category_scores_gemma":[0.0036479346,0.00025836335,0.00066906476,0.0018849801,0.00058153627,0.000454275,0.0012002381,0.0004830638,0.0006273025],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0025899145,0.00034459308,0.20676535,0.0033669278,0.0013801348,0.0016047809,0.0012649053,0.08999686,0.23387562,0.018206665,0.025552772,0.41505143],"study_design_scores_gemma":[0.00025336558,0.000489579,0.25838578,0.0003802768,0.0007156776,0.0030795797,0.0005012999,0.51989806,0.0927733,0.08822363,0.035122484,0.00017697131],"about_ca_topic_score_codex":0.0010831016,"about_ca_topic_score_gemma":0.0016694639,"teacher_disagreement_score":0.002491278,"about_ca_system_score_codex":0.0005024938,"about_ca_system_score_gemma":0.0006515973,"threshold_uncertainty_score":0.0083341},"labels":[],"label_agreement":null},{"id":"W3112841179","doi":"10.1093/bioinformatics/btaa1008","title":"Prioritizing genes for systematic variant effect mapping","year":2020,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Rare Diseases","field":"Biochemistry, Genetics and Molecular Biology","cited_by":18,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Sinai Health System; Lunenfeld-Tanenbaum Research Institute; University of Toronto","funders":"National Human Genome Research Institute; Canada Excellence Research Chairs, Government of Canada","keywords":"Missense mutation; Prioritization; Gene; Computational biology; Pathogenicity; Protein function; Source code; Biology; Genetics; Mutation; Computer science","score_opus":0.013396985121383712,"score_gpt":0.2290038695373218,"score_spread":0.2156068844159381,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3112841179","genre_codex":"empirical","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.54817975,0.0065401928,0.3766508,0.0049418933,0.0005973091,0.0015470347,0.032188304,0.018420001,0.010934752],"genre_scores_gemma":[0.61327785,0.0008242207,0.3605995,0.0011632214,0.00024463216,0.000531472,0.019757817,0.0015467476,0.0020546366],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99835867,0.00044180482,0.00012816866,0.0005552414,0.00036592688,0.00015006901],"domain_scores_gemma":[0.99248743,0.0049769655,0.0005638971,0.0005509064,0.0010346206,0.00038610434],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003803972,0.0012643654,0.0013367516,0.0032602823,0.00070551684,0.0018521547,0.0012604131,0.0006859112,0.0062314803],"category_scores_gemma":[0.013836062,0.00030412115,0.001253957,0.0023395491,0.00069001224,0.0008850089,0.0014629903,0.0009517709,0.0021105062],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0036192029,0.00044724767,0.22138092,0.0041925353,0.0010602159,0.0032775698,0.0011961688,0.027340284,0.18689837,0.012990176,0.052874386,0.48472297],"study_design_scores_gemma":[0.0015598952,0.0017448061,0.19334605,0.0010399474,0.0028443679,0.009117446,0.0020575775,0.3221231,0.26348066,0.07159927,0.13066442,0.00042248072],"about_ca_topic_score_codex":0.0015905473,"about_ca_topic_score_gemma":0.0033432883,"teacher_disagreement_score":0.0062314803,"about_ca_system_score_codex":0.00069335033,"about_ca_system_score_gemma":0.0018595495,"threshold_uncertainty_score":0.020846426},"labels":[],"label_agreement":null},{"id":"W3115522577","doi":"10.1093/bioinformatics/btaa842","title":"Supervised learning on phylogenetically distributed data","year":2020,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Cell Image Analysis Techniques","field":"Biochemistry, Genetics and Molecular Biology","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia; McGill University","funders":"Genome Canada","keywords":"Computer science; Generalization; Phylogenetic tree; Context (archaeology); Machine learning; Artificial intelligence; Independent and identically distributed random variables; Flexibility (engineering); Artificial neural network; Tree (set theory); Data mining; Biology; Mathematics; Statistics; Genetics","score_opus":0.026320987439237244,"score_gpt":0.26800879019179735,"score_spread":0.2416878027525601,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3115522577","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.04031307,0.0008107859,0.9524298,0.0010500078,0.00009535074,0.00010575705,0.0015065573,0.0021355539,0.0015531365],"genre_scores_gemma":[0.58623517,0.0007480216,0.40143484,0.0006944542,0.00034228314,0.00039937953,0.007242489,0.00040157253,0.0025018228],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99644226,0.0018036551,0.00021992177,0.00095543236,0.00046426774,0.000114536095],"domain_scores_gemma":[0.97521484,0.018394273,0.0013493503,0.003167481,0.0015157016,0.0003583533],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0060802274,0.0009770409,0.0011044372,0.0010673902,0.00061360345,0.0011688035,0.0025248278,0.0017876619,0.0017051453],"category_scores_gemma":[0.028831879,0.00055589835,0.00083633413,0.0011922698,0.0015035943,0.0024054248,0.001712705,0.0024443401,0.0010938408],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00016900741,0.00015899898,0.00456337,0.00044588925,0.00015573093,0.00017026716,0.00012358752,0.8760961,0.0021367685,0.011409291,0.006406932,0.098163985],"study_design_scores_gemma":[0.000015331832,0.000028295324,0.00042792622,0.000022667127,0.0000071754835,0.0000399979,0.000011610584,0.97706485,0.0010273041,0.020406283,0.00094176765,0.000006850408],"about_ca_topic_score_codex":0.0016308344,"about_ca_topic_score_gemma":0.0023089568,"teacher_disagreement_score":0.0060802274,"about_ca_system_score_codex":0.0009960419,"about_ca_system_score_gemma":0.0010685863,"threshold_uncertainty_score":0.032155693},"labels":[],"label_agreement":null},{"id":"W3115554950","doi":"10.1093/bioinformatics/btab189","title":"The NRGTEN Python package: an extensible toolkit for coarse-grained normal mode analysis of proteins, nucleic acids, small molecules and their complexes","year":2021,"lang":"en","type":"preprint","venue":"Bioinformatics","topic":"Protein Structure and Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal; Institute for Research in Immunology and Cancer","funders":"","keywords":"Python (programming language); Computer science; Extensibility; Docking (animal); Chemistry; Programming language","score_opus":0.013682821071897035,"score_gpt":0.24898436254100595,"score_spread":0.23530154146910892,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3115554950","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0031424651,0.0005473027,0.68397427,0.0004824249,0.00041320024,0.00024626357,0.046055734,0.25520396,0.00993428],"genre_scores_gemma":[0.042809013,0.0010378634,0.7073664,0.0006806663,0.00024798836,0.0017404247,0.082332715,0.14752175,0.01626321],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99890935,0.00020709193,0.00007943957,0.00015709442,0.00052688614,0.000120057404],"domain_scores_gemma":[0.99845254,0.0005132488,0.000121611854,0.0004120382,0.00033793753,0.00016262666],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0020995282,0.0018321814,0.0016653658,0.0018624692,0.0008993552,0.0019980643,0.0049302243,0.0009766137,0.06261817],"category_scores_gemma":[0.005886097,0.0012623888,0.0018803915,0.001326256,0.0007083261,0.002611553,0.003937576,0.0032735735,0.044496667],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002883977,0.00016062094,0.0020615768,0.0020073224,0.0005167494,0.0006089908,0.00031193337,0.036621485,0.009762102,0.055073686,0.7486542,0.14393292],"study_design_scores_gemma":[0.00033223513,0.000059180613,0.0025403528,0.0003508712,0.000094535455,0.00075168855,0.00008513934,0.350455,0.011143029,0.13844278,0.4954337,0.00031156282],"about_ca_topic_score_codex":0.004306101,"about_ca_topic_score_gemma":0.0060404637,"teacher_disagreement_score":0.06261817,"about_ca_system_score_codex":0.0006769574,"about_ca_system_score_gemma":0.0027377426,"threshold_uncertainty_score":0.20947862},"labels":[],"label_agreement":null},{"id":"W3117084002","doi":"10.1093/bioinformatics/btaa1077","title":"Improving circRNA–disease association prediction by sequence and ontology representations with convolutional and recurrent neural networks","year":2020,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Circular RNAs in diseases","field":"Biochemistry, Genetics and Molecular Biology","cited_by":46,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Saskatchewan","funders":"Higher Education Discipline Innovation Project; National Natural Science Foundation of China","keywords":"Computer science; Ontology; Convolutional neural network; Identification (biology); Sequence (biology); Artificial intelligence; Disease; Machine learning; Computational biology; Biology; Genetics; Medicine","score_opus":0.012069895630337916,"score_gpt":0.2344749164493988,"score_spread":0.22240502081906088,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3117084002","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.37738138,0.0030857688,0.60655266,0.0010088898,0.00019443789,0.00011773084,0.002160151,0.0068992227,0.0025997867],"genre_scores_gemma":[0.8578329,0.0008999926,0.13192424,0.0004556089,0.00014705652,0.00011955574,0.005422686,0.00012656093,0.0030714122],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99964106,0.000052208747,0.00002627535,0.00014037696,0.00007711142,0.00006306388],"domain_scores_gemma":[0.9993406,0.00027950693,0.00013631892,0.00007617139,0.0001243436,0.000043024767],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00055280153,0.0010588185,0.00057243987,0.0012894823,0.0002741163,0.0005276569,0.00074094685,0.000698256,0.00075663364],"category_scores_gemma":[0.0017750291,0.0002695953,0.00081526116,0.0008447599,0.00028861855,0.0008452881,0.0006055075,0.00089303456,0.00036972252],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00075112394,0.0006157069,0.040726017,0.00034091793,0.00043570768,0.00075286656,0.0001650544,0.29487392,0.04119503,0.0041785934,0.013193723,0.60277146],"study_design_scores_gemma":[0.000011616993,0.000038008642,0.0014718984,0.000008538657,0.000044291366,0.000060768176,0.000010508419,0.9924085,0.0037268838,0.0016718335,0.000537763,0.000009338399],"about_ca_topic_score_codex":0.010879203,"about_ca_topic_score_gemma":0.015465361,"teacher_disagreement_score":0.010879203,"about_ca_system_score_codex":0.00070192147,"about_ca_system_score_gemma":0.0009911191,"threshold_uncertainty_score":0.021631718},"labels":[],"label_agreement":null},{"id":"W3118845805","doi":"10.1093/bioinformatics/btaa1105","title":"PhyloCorrelate: inferring bacterial gene–gene functional associations through large-scale phylogenetic profiling","year":2020,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":26,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"Natural Sciences and Engineering Research Council of Canada; University of Waterloo","keywords":"Phylogenetic tree; Computational biology; Profiling (computer programming); Gene; Biology; Genetics; Computer science","score_opus":0.017719928573947816,"score_gpt":0.2259968109679526,"score_spread":0.20827688239400477,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3118845805","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.30150136,0.0015438796,0.60921395,0.00068480114,0.00010949149,0.00032451563,0.031885844,0.051326226,0.0034099186],"genre_scores_gemma":[0.46144336,0.000648709,0.5011007,0.00030118925,0.00006173656,0.00034167533,0.032680422,0.0025239117,0.00089826377],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9985921,0.00043061297,0.00007702359,0.0005057228,0.00029607935,0.00009854775],"domain_scores_gemma":[0.9974112,0.0014519145,0.00042646454,0.00035408398,0.00016994742,0.00018646072],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0031773567,0.0015813546,0.0014562758,0.0025152985,0.0009330662,0.001717629,0.0016995935,0.000836474,0.0037389158],"category_scores_gemma":[0.00865755,0.00084187795,0.0015727804,0.0021977925,0.0009432134,0.0017347628,0.002538689,0.0011643935,0.001429883],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0042684204,0.0006525606,0.19083045,0.0048521687,0.0019444969,0.0013920548,0.0017591921,0.21890405,0.19802895,0.018304346,0.05502311,0.30404016],"study_design_scores_gemma":[0.00018343228,0.00038822653,0.032062158,0.00015407013,0.0002219537,0.0009089407,0.0002222034,0.9063531,0.028598787,0.014569211,0.016196012,0.00014186016],"about_ca_topic_score_codex":0.00323708,"about_ca_topic_score_gemma":0.0045797364,"teacher_disagreement_score":0.0037389158,"about_ca_system_score_codex":0.0008118753,"about_ca_system_score_gemma":0.0013499613,"threshold_uncertainty_score":0.016803682},"labels":[],"label_agreement":null},{"id":"W3122382017","doi":"10.1093/bioinformatics/btab022","title":"PrInCE: an R/Bioconductor package for protein–protein interaction network inference from co-fractionation mass spectrometry data","year":2021,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":20,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canada's Michael Smith Genome Sciences Centre; University of British Columbia","funders":"","keywords":"Bioconductor; Computer science; Source code; Scripting language; Software; Plug-in; Inference; Workflow; Data mining; Programming language; Database; Artificial intelligence; Chemistry","score_opus":0.033643065196177366,"score_gpt":0.3009317678640216,"score_spread":0.2672887026678442,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3122382017","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0023366192,0.002088528,0.48620486,0.0015171407,0.00091269746,0.000515096,0.17261955,0.32757884,0.0062267007],"genre_scores_gemma":[0.02775804,0.0022888975,0.552996,0.0021982773,0.00047831316,0.005050628,0.2074249,0.18823412,0.013570908],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99679476,0.001027347,0.00030753427,0.000953518,0.00070773397,0.00020913017],"domain_scores_gemma":[0.9900413,0.0052454853,0.0010084292,0.0017325517,0.001570095,0.00040226066],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0066564847,0.0041861623,0.0036135265,0.0046252506,0.001173911,0.0039708233,0.007016425,0.0018768848,0.10675401],"category_scores_gemma":[0.027523622,0.0026012533,0.004299751,0.0036368573,0.001367738,0.0034930215,0.0046373624,0.0054365364,0.089958504],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00045395776,0.00006210665,0.0022938075,0.004772744,0.0017370676,0.0003952699,0.00023799333,0.007143359,0.003519471,0.011988851,0.91933304,0.04806234],"study_design_scores_gemma":[0.0007668488,0.00016711504,0.0058775456,0.0012411908,0.0011321127,0.0013937487,0.00014278588,0.10293291,0.01152908,0.107641265,0.76679593,0.00037956046],"about_ca_topic_score_codex":0.004142337,"about_ca_topic_score_gemma":0.005668231,"teacher_disagreement_score":0.10675401,"about_ca_system_score_codex":0.0011946288,"about_ca_system_score_gemma":0.004497041,"threshold_uncertainty_score":0.35712773},"labels":[],"label_agreement":null},{"id":"W3124991397","doi":"10.1093/bioinformatics/btab018","title":"Mouse Quantitative Proteomics Knowledgebase: reference protein concentration ranges in 20 mouse tissues using 5000 quantitative proteomics assays","year":2021,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Advanced Proteomics Techniques and Applications","field":"Chemistry","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; Jewish General Hospital; Genome British Columbia; University of Victoria","funders":"McGill University","keywords":"Proteomics; Computational biology; Quantitative proteomics; Biology; Laboratory mouse; Gene; Bioinformatics; Genetics","score_opus":0.05352775581482376,"score_gpt":0.3313772763677119,"score_spread":0.27784952055288814,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3124991397","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0030845224,0.005925704,0.04629256,0.0006256704,0.00025474725,0.0004107693,0.8978713,0.03783968,0.0076950276],"genre_scores_gemma":[0.0068900227,0.0039620386,0.049878396,0.0005996963,0.00004726072,0.0009987449,0.9318708,0.0033499964,0.0024031198],"study_design_codex":"not_applicable","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9973787,0.00030223996,0.00061382254,0.00064792467,0.00091497804,0.00014227479],"domain_scores_gemma":[0.9945741,0.001683377,0.0008920708,0.0010385514,0.0015373083,0.00027465622],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004285155,0.0041270773,0.004010497,0.011622915,0.0010776684,0.004433633,0.006832874,0.00516323,0.041127946],"category_scores_gemma":[0.010306699,0.0023253895,0.0019371335,0.009525778,0.00069536,0.0037910037,0.0032448913,0.0020332779,0.04103348],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0017473572,0.000524936,0.00592908,0.025245259,0.00083681755,0.0009502001,0.0003105697,0.007467213,0.042340353,0.008667598,0.7476442,0.15833639],"study_design_scores_gemma":[0.0003457763,0.00019345232,0.008273959,0.0028811216,0.0008170802,0.0013492578,0.00010809508,0.0056490237,0.024648797,0.011052723,0.9444385,0.0002422415],"about_ca_topic_score_codex":0.0054635974,"about_ca_topic_score_gemma":0.0064467923,"teacher_disagreement_score":0.041127946,"about_ca_system_score_codex":0.002020523,"about_ca_system_score_gemma":0.0039020274,"threshold_uncertainty_score":0.13758671},"labels":[],"label_agreement":null},{"id":"W3126228859","doi":"10.1093/bioinformatics/btab077","title":"Detecting high-scoring local alignments in pangenome graphs","year":2021,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"German Network for Bioinformatics Infrastructure; Michael Smith Health Research BC; Bundesministerium für Bildung und Forschung; Deutsche Forschungsgemeinschaft; Natural Sciences and Engineering Research Council of Canada; European Commission; California Department of Fish and Game","keywords":"De Bruijn graph; Computer science; Spurious relationship; De Bruijn sequence; Graph; Sequence (biology); Similarity (geometry); Theoretical computer science; Artificial intelligence; Mathematics; Combinatorics; Biology; Machine learning; Genetics","score_opus":0.01144780682644667,"score_gpt":0.21991272058270583,"score_spread":0.20846491375625917,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3126228859","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.08359839,0.0009943178,0.8985146,0.00034087588,0.0000507711,0.00022427188,0.0025180122,0.012296197,0.0014625564],"genre_scores_gemma":[0.35673252,0.00055849674,0.63249147,0.00017221368,0.0000572223,0.00021179432,0.0071148295,0.0013031872,0.001358234],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9980987,0.00050724304,0.00013967347,0.0006201849,0.0005050479,0.00012916674],"domain_scores_gemma":[0.99238163,0.0043913657,0.0011846249,0.0008321265,0.0008636007,0.00034662333],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016611908,0.0013113943,0.0010164444,0.0034352182,0.00075436325,0.0017328556,0.001887413,0.0011607839,0.003711916],"category_scores_gemma":[0.013240373,0.00067902927,0.00088957936,0.0028946097,0.001017867,0.0021376936,0.0020467376,0.0009785034,0.0019616121],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013703128,0.00045644457,0.024267053,0.004541544,0.0005681211,0.0018546911,0.0011610478,0.22192177,0.23403236,0.021531574,0.01666482,0.4716302],"study_design_scores_gemma":[0.000101928366,0.00028604284,0.010681172,0.00021490974,0.00014315174,0.0025805715,0.0005501873,0.82930917,0.068051316,0.07597988,0.011968199,0.00013344074],"about_ca_topic_score_codex":0.001158961,"about_ca_topic_score_gemma":0.0022205485,"teacher_disagreement_score":0.003711916,"about_ca_system_score_codex":0.00066602754,"about_ca_system_score_gemma":0.0008236297,"threshold_uncertainty_score":0.012417555},"labels":[],"label_agreement":null},{"id":"W3127276189","doi":"10.1093/bioinformatics/btab075","title":"RAD: a web application to identify region associated differentially expressed genes","year":2021,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":23,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"Fundamental Research Funds for the Central Universities; Canadian Institutes of Health Research","keywords":"Biology; Gene; Computational biology; Python (programming language); Genetics; Web server; Genome; Chromatin; Computer science; The Internet; World Wide Web","score_opus":0.011409763270810988,"score_gpt":0.24695676855460863,"score_spread":0.23554700528379763,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3127276189","genre_codex":"software","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.009317775,0.0010095162,0.120796554,0.0003568562,0.00038188486,0.00028387015,0.20655662,0.6550544,0.0062425565],"genre_scores_gemma":[0.07320963,0.0009983354,0.42785722,0.0014727225,0.00029008783,0.0026476453,0.38907903,0.08939588,0.01504943],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9990151,0.00014504396,0.000074399104,0.00042843915,0.00023477757,0.000102213446],"domain_scores_gemma":[0.9986046,0.0008193117,0.00015968614,0.00015768658,0.00013949421,0.00011930298],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0019533564,0.0024408035,0.0018798286,0.003571841,0.0008238983,0.0020132354,0.0024541682,0.0013521179,0.061258182],"category_scores_gemma":[0.003236074,0.0010969469,0.0019455234,0.0019397776,0.000471829,0.0015125868,0.002675757,0.0018425883,0.045642633],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0028744426,0.000527806,0.016638894,0.0056167813,0.001164213,0.0013820014,0.0007593799,0.006012127,0.05951558,0.0052141044,0.71797407,0.18232064],"study_design_scores_gemma":[0.001526352,0.0004819982,0.04722323,0.0007027218,0.00064574124,0.0023979617,0.0004127997,0.080891564,0.09886139,0.024118803,0.74211085,0.00062653015],"about_ca_topic_score_codex":0.0009289414,"about_ca_topic_score_gemma":0.0015457091,"teacher_disagreement_score":0.061258182,"about_ca_system_score_codex":0.000640664,"about_ca_system_score_gemma":0.0009812402,"threshold_uncertainty_score":0.20492905},"labels":[],"label_agreement":null},{"id":"W3127513279","doi":"10.1093/bioinformatics/btab076","title":"WgLink: reconstructing whole-genome viral haplotypes using L0+L1-regularization","year":2021,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary; Alberta Children's Hospital","funders":"Canada Foundation for Innovation","keywords":"Haplotype; Regularization (linguistics); Computational biology; Genome; Genetics; Computer science; Biology; Virology; Artificial intelligence; Gene; Genotype","score_opus":0.017051819490013154,"score_gpt":0.2314050820654523,"score_spread":0.21435326257543913,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3127513279","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.025326844,0.00041587884,0.9347478,0.00035458335,0.00015048293,0.000105874424,0.003496073,0.03351444,0.0018880217],"genre_scores_gemma":[0.15720275,0.00026852934,0.8027682,0.0005315416,0.000117949414,0.0003668336,0.022134596,0.011277999,0.005331568],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99939597,0.00017918649,0.000025654665,0.00019242155,0.00016040343,0.000046503574],"domain_scores_gemma":[0.9988255,0.0006201002,0.00007832875,0.00025825808,0.00012975,0.00008795222],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0020290078,0.0015742892,0.00090617925,0.001094325,0.00059344625,0.0017177672,0.0026602964,0.0019619993,0.006622535],"category_scores_gemma":[0.006003223,0.000809221,0.0013928019,0.0010492251,0.00070959434,0.0013850433,0.0023050737,0.001972134,0.0036355394],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00059688813,0.00024731335,0.0079579735,0.00086161366,0.0007141717,0.00042920504,0.00034543092,0.5938129,0.024963541,0.013359045,0.0779371,0.2787749],"study_design_scores_gemma":[0.000052058833,0.000026319782,0.00043736811,0.00001863048,0.0000142391655,0.00004117603,0.00002542413,0.98705846,0.003210843,0.005019153,0.004077956,0.000018389752],"about_ca_topic_score_codex":0.005566791,"about_ca_topic_score_gemma":0.008338834,"teacher_disagreement_score":0.006622535,"about_ca_system_score_codex":0.00050770794,"about_ca_system_score_gemma":0.0011800841,"threshold_uncertainty_score":0.02215457},"labels":[],"label_agreement":null},{"id":"W3127538277","doi":"10.1093/bioinformatics/btab072","title":"Motif-Raptor: a cell type-specific and transcription factor centric approach for post-GWAS prioritization of causal regulators","year":2021,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal; Montreal Heart Institute","funders":"National Human Genome Research Institute; National Heart, Lung, and Blood Institute; National Institutes of Health","keywords":"Computational biology; Genome-wide association study; Chromatin; Biology; Transcription factor; Gene; Genetics; Single-nucleotide polymorphism; Genotype","score_opus":0.01624833803372405,"score_gpt":0.22994435342991612,"score_spread":0.21369601539619207,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3127538277","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.06989729,0.0011958593,0.77022666,0.0007906881,0.00032319874,0.0003747512,0.036666185,0.11717468,0.0033506607],"genre_scores_gemma":[0.26055568,0.00053741067,0.6877041,0.000943214,0.00021069887,0.0011874229,0.032628898,0.013082501,0.0031501453],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99941206,0.00021791326,0.00003256637,0.00020516106,0.00008438507,0.000047990783],"domain_scores_gemma":[0.99725276,0.0020474172,0.00015587363,0.0002505546,0.00015353902,0.00013986087],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0022524719,0.0014687651,0.0016313198,0.0016755603,0.0005024219,0.0011420937,0.0024156384,0.001282507,0.016622849],"category_scores_gemma":[0.0090268925,0.0008097466,0.002303576,0.0010230755,0.00038697355,0.00074753485,0.0015349392,0.0015149429,0.0040931706],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.006321519,0.0007712234,0.0857589,0.0046920455,0.0058238925,0.002604206,0.00087855285,0.26891845,0.059754275,0.024869032,0.18204828,0.3575597],"study_design_scores_gemma":[0.00047100047,0.00033984086,0.0065226513,0.00012890369,0.00047755474,0.00085647794,0.00008333854,0.93950427,0.010170685,0.018397825,0.022932095,0.000115512725],"about_ca_topic_score_codex":0.0027925647,"about_ca_topic_score_gemma":0.0078011192,"teacher_disagreement_score":0.016622849,"about_ca_system_score_codex":0.00047045987,"about_ca_system_score_gemma":0.0015759593,"threshold_uncertainty_score":0.05560893},"labels":[],"label_agreement":null},{"id":"W3133006703","doi":"10.1093/bioinformatics/btab112","title":"Accurate deep learning off-target prediction with novel sgRNA-DNA sequence encoding in CRISPR-Cas9 gene editing","year":2021,"lang":"en","type":"article","venue":"Bioinformatics","topic":"CRISPR and Genetic Engineering","field":"Biochemistry, Genetics and Molecular Biology","cited_by":54,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; McGill University and Génome Québec Innovation Centre; Université du Québec à Montréal","funders":"Natural Sciences and Engineering Research Council of Canada; Fonds Québécois de la Recherche sur la Nature et les Technologies","keywords":"Computer science; Deep learning; Encoding (memory); Recurrent neural network; Artificial intelligence; Convolutional neural network; CRISPR; Artificial neural network; Machine learning; Gene; Biology; Genetics","score_opus":0.01337359534881089,"score_gpt":0.2686154771607254,"score_spread":0.25524188181191454,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3133006703","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.34277442,0.0016667953,0.6402513,0.0006701033,0.00016594946,0.000083473285,0.0011028455,0.010003544,0.0032815607],"genre_scores_gemma":[0.8971259,0.00029107652,0.099355966,0.00017687578,0.000019264604,0.0000600329,0.0010551135,0.00012261097,0.0017932609],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9997626,0.00004409276,0.000013029757,0.0000739285,0.00007007251,0.000036314956],"domain_scores_gemma":[0.99937254,0.00032794042,0.000064693646,0.0000635453,0.00013405744,0.000037321315],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00066426245,0.00087326573,0.000503927,0.00024704132,0.00022246726,0.00046078928,0.0008231453,0.0007937957,0.0008988656],"category_scores_gemma":[0.0022761207,0.00019722432,0.00035653327,0.00023033314,0.00026932915,0.00073863193,0.0003753026,0.000985348,0.0003546043],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000329497,0.00016675738,0.0043517696,0.00013985285,0.00005772739,0.00016391954,0.00003326412,0.84562075,0.021550158,0.0010093658,0.0022671174,0.1243098],"study_design_scores_gemma":[0.0000042705155,0.000029763247,0.0003111121,0.000005380643,0.0000060213983,0.0000149409725,0.0000034066454,0.9908178,0.008130974,0.0004971054,0.00017448413,0.0000048233383],"about_ca_topic_score_codex":0.008342403,"about_ca_topic_score_gemma":0.008283185,"teacher_disagreement_score":0.008342403,"about_ca_system_score_codex":0.00075944903,"about_ca_system_score_gemma":0.0007751202,"threshold_uncertainty_score":0.016587734},"labels":[],"label_agreement":null},{"id":"W3134330704","doi":"10.1093/bioinformatics/btab157","title":"KibioR &amp; Kibio: a new architecture for next-generation data querying and sharing in big biology","year":2021,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université Laval","funders":"","keywords":"Computer science; Scalability; Data management; Identification (biology); Data exchange; Database; Biological database; Big data; Data science; Simple (philosophy); Biological data; Data access; World Wide Web; Data mining; Bioinformatics","score_opus":0.08786989134044719,"score_gpt":0.2923718289039058,"score_spread":0.20450193756345858,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3134330704","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0026300654,0.0027272445,0.6937466,0.0042312457,0.000885001,0.0005540059,0.006866293,0.27749285,0.010866743],"genre_scores_gemma":[0.043970533,0.0033024005,0.8436122,0.005228783,0.0005174563,0.0019387187,0.053673398,0.03632369,0.011432879],"study_design_codex":"not_applicable","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9920666,0.0015969556,0.0008068662,0.002022676,0.0028874537,0.00061936484],"domain_scores_gemma":[0.9872295,0.0025241275,0.00085131155,0.0061346716,0.0013583618,0.0019019265],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.014094418,0.0025816704,0.002843656,0.0045414316,0.0031416963,0.010309298,0.012744096,0.0041875457,0.017550752],"category_scores_gemma":[0.024596345,0.0033905003,0.0030441235,0.00728597,0.003787213,0.020505901,0.022300009,0.0071196863,0.02644636],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0025427057,0.0003418375,0.0036145374,0.0026237501,0.0007783747,0.00076807925,0.0017793404,0.01589566,0.018102033,0.14700742,0.5662601,0.24028613],"study_design_scores_gemma":[0.00040804365,0.0002229744,0.0013963233,0.00063269323,0.00022870662,0.00078525324,0.00041975616,0.15814927,0.017554976,0.22465558,0.59499985,0.0005465977],"about_ca_topic_score_codex":0.006940353,"about_ca_topic_score_gemma":0.004793936,"teacher_disagreement_score":0.017550752,"about_ca_system_score_codex":0.0026817853,"about_ca_system_score_gemma":0.005163411,"threshold_uncertainty_score":0.074539244},"labels":[],"label_agreement":null},{"id":"W3138390968","doi":"10.1093/bioinformatics/btab188","title":"PoseFilter: a PyMOL plugin for filtering and analyzing small molecule docking in symmetric binding sites","year":2021,"lang":"en","type":"article","venue":"Bioinformatics","topic":"14-3-3 protein interactions","field":"Biochemistry, Genetics and Molecular Biology","cited_by":18,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"Canada First Research Excellence Fund","keywords":"Plug-in; Docking (animal); Computer science; Graphical user interface; Software; Interface (matter); Computational biology; Data mining; Programming language; Operating system; Biology","score_opus":0.0231698618666592,"score_gpt":0.26680804518820156,"score_spread":0.24363818332154236,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3138390968","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0127560785,0.0009147981,0.30477735,0.00050479453,0.00038599913,0.00025203847,0.066907465,0.6070839,0.006417537],"genre_scores_gemma":[0.12776822,0.0022751878,0.3983299,0.0011959466,0.00015792406,0.0024065117,0.21943565,0.22263722,0.025793366],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.999188,0.000110335684,0.000048076712,0.00016531786,0.00035831123,0.0001299124],"domain_scores_gemma":[0.9993274,0.0002838669,0.0000612292,0.00012208369,0.00012576542,0.00007964126],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016478765,0.0026913288,0.00234767,0.002063366,0.0011854097,0.0020307861,0.0037148157,0.0010309878,0.08294735],"category_scores_gemma":[0.0034628229,0.0014313727,0.0014763128,0.0013733931,0.00038033712,0.0020408293,0.0021592653,0.0029082738,0.034133665],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009437033,0.00015929133,0.0027426714,0.0018964551,0.00036299496,0.00046345568,0.00023718341,0.0059226505,0.026019782,0.005521863,0.8345279,0.12120205],"study_design_scores_gemma":[0.0006963436,0.0003134032,0.01037769,0.00040591712,0.00025156923,0.0016357547,0.00023002768,0.17193678,0.17687553,0.017299796,0.6192104,0.0007668891],"about_ca_topic_score_codex":0.0042901956,"about_ca_topic_score_gemma":0.005413163,"teacher_disagreement_score":0.08294735,"about_ca_system_score_codex":0.00066021556,"about_ca_system_score_gemma":0.0017634664,"threshold_uncertainty_score":0.2774865},"labels":[],"label_agreement":null},{"id":"W3138529574","doi":"10.1093/bioinformatics/btac326","title":"Building alternative consensus trees and supertrees using <i>k</i> -means and Robinson and Foulds distance","year":2022,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Montréal; Université de Sherbrooke","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Computer science","score_opus":0.015281997913291155,"score_gpt":0.24037222629088925,"score_spread":0.2250902283775981,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3138529574","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.01133663,0.00011696793,0.986521,0.000090199035,0.000035323257,0.00008116121,0.00025257267,0.001074479,0.00049169926],"genre_scores_gemma":[0.041902445,0.00007081163,0.9559222,0.00005652664,0.00002510944,0.00013059028,0.0011270622,0.0003279661,0.0004374187],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9963462,0.0009281755,0.0002720721,0.0012278644,0.0009822269,0.0002434825],"domain_scores_gemma":[0.99060786,0.0043202057,0.0009936331,0.0011878436,0.0024360511,0.00045440887],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0046436153,0.0015483489,0.0021816627,0.005503193,0.0025275021,0.0022201089,0.004584206,0.0026012708,0.0038841558],"category_scores_gemma":[0.021145208,0.0013905884,0.0029842104,0.0038689978,0.0012568233,0.00377558,0.0027623693,0.0026625518,0.0019258206],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00036237197,0.00018912254,0.005338436,0.0005902878,0.00038269948,0.00037238534,0.0011924152,0.5695617,0.0071389657,0.042409964,0.010419895,0.3620417],"study_design_scores_gemma":[0.000024537288,0.00003146347,0.0004220584,0.00003406824,0.000029524772,0.00007124239,0.00013073362,0.9598744,0.0018248091,0.035300776,0.002221451,0.000034884033],"about_ca_topic_score_codex":0.009601192,"about_ca_topic_score_gemma":0.014444961,"teacher_disagreement_score":0.009601192,"about_ca_system_score_codex":0.001930688,"about_ca_system_score_gemma":0.003292876,"threshold_uncertainty_score":0.024558127},"labels":[],"label_agreement":null},{"id":"W3149048453","doi":"10.1093/bioinformatics/btab215","title":"MaveRegistry: a collaboration platform for multiplexed assays of variant effect","year":2021,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Rare Diseases","field":"Biochemistry, Genetics and Molecular Biology","cited_by":27,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Lunenfeld-Tanenbaum Research Institute; University of Toronto","funders":"National Institutes of Health; National Human Genome Research Institute; Canada Excellence Research Chairs, Government of Canada","keywords":"Computer science; Source code; Multiplexing; Workflow; Computational biology; NOMINATE; Data mining; Biology; Database; Machine learning; Programming language; Telecommunications","score_opus":0.008754319354611684,"score_gpt":0.24459441916459593,"score_spread":0.23584009980998424,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3149048453","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.010752775,0.0014104486,0.34433007,0.0015294558,0.00067361904,0.00075652485,0.101643436,0.51742655,0.021477176],"genre_scores_gemma":[0.12527245,0.0012341558,0.51549906,0.0031835753,0.00047767177,0.0038239954,0.21551257,0.10849741,0.026499057],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9948849,0.0012785988,0.0003868813,0.0015675417,0.0014488641,0.0004332846],"domain_scores_gemma":[0.98770124,0.0055230786,0.001214186,0.0030890866,0.0012856198,0.0011867393],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0075862785,0.0017457844,0.0018277037,0.0034128665,0.0009409958,0.004083217,0.0033870025,0.0018086892,0.055047426],"category_scores_gemma":[0.023574376,0.0014955157,0.001722582,0.0020809378,0.00074177515,0.0030211462,0.0073220367,0.0028888255,0.043422528],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0039360183,0.0002587314,0.016264128,0.001595492,0.0010396085,0.0014708213,0.0011996899,0.0045511066,0.04266666,0.019923303,0.7272441,0.1798504],"study_design_scores_gemma":[0.0014755414,0.00046733094,0.018064562,0.0005162296,0.0004089964,0.0021183465,0.0002782792,0.04644739,0.07435281,0.065449364,0.7897207,0.0007004],"about_ca_topic_score_codex":0.002660104,"about_ca_topic_score_gemma":0.003551564,"teacher_disagreement_score":0.055047426,"about_ca_system_score_codex":0.0008283175,"about_ca_system_score_gemma":0.0024651883,"threshold_uncertainty_score":0.18415195},"labels":[],"label_agreement":null},{"id":"W3159344430","doi":"10.1093/bioinformatics/btab249","title":"MetaFusion: a high-confidence metacaller for filtering and prioritizing RNA-seq gene fusion candidates","year":2021,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":16,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University Health Network; University of Toronto; SickKids Foundation; Hospital for Sick Children","funders":"","keywords":"Computer science; Benchmarking; Data mining; Graph; Fusion; Cluster analysis; Precision and recall; Majority rule; Selection (genetic algorithm); Machine learning; Information retrieval; Artificial intelligence; Theoretical computer science","score_opus":0.01830368811237904,"score_gpt":0.25958593528051555,"score_spread":0.2412822471681365,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3159344430","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.044251475,0.00146262,0.50466245,0.00060758507,0.00048152485,0.0005248226,0.05507886,0.38344482,0.009485912],"genre_scores_gemma":[0.15678331,0.0004548906,0.7070666,0.001489097,0.00022472968,0.0015092393,0.07067277,0.055033755,0.006765532],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99500173,0.0008098916,0.00038955605,0.0015684182,0.0018456354,0.00038481626],"domain_scores_gemma":[0.9914419,0.0049622175,0.00084188604,0.0013725578,0.0009522902,0.0004292275],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008896324,0.0029647409,0.001757489,0.005289178,0.0022667213,0.0039563146,0.003097639,0.0020405932,0.021111988],"category_scores_gemma":[0.021864086,0.0016602514,0.002672115,0.0024975676,0.00071776053,0.0029321108,0.003954851,0.002270569,0.012867315],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.004173928,0.0003540928,0.061362855,0.0043335184,0.0017389915,0.0014883457,0.0019749822,0.0144827645,0.26315284,0.012204034,0.29530287,0.33943072],"study_design_scores_gemma":[0.0005502776,0.0006195387,0.042687133,0.00080081687,0.0009375775,0.0032926148,0.0007017947,0.28111923,0.44956842,0.020625835,0.19810842,0.0009883601],"about_ca_topic_score_codex":0.0023459923,"about_ca_topic_score_gemma":0.0057709482,"teacher_disagreement_score":0.021111988,"about_ca_system_score_codex":0.001068294,"about_ca_system_score_gemma":0.0022200583,"threshold_uncertainty_score":0.070626676},"labels":[],"label_agreement":null},{"id":"W3162343445","doi":"10.1093/bioinformatics/btab363","title":"PaIntDB: network-based omics integration and visualization using protein–protein interactions in <i>Pseudomonas aeruginosa</i>","year":2021,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"Canadian Institutes of Health Research","keywords":"Interactome; Visualization; Computer science; Source code; Computational biology; Pseudomonas aeruginosa; Web server; Data mining; Bioinformatics; Biology; World Wide Web; The Internet; Genetics; Gene; Programming language","score_opus":0.014712646126482511,"score_gpt":0.25610344437096333,"score_spread":0.24139079824448081,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3162343445","genre_codex":"dataset","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.027040932,0.0014445181,0.1643673,0.0015325847,0.0005459799,0.0005540658,0.46444866,0.3244519,0.015614013],"genre_scores_gemma":[0.10085933,0.001975889,0.25941396,0.0008137534,0.0001388462,0.0019136113,0.58512414,0.04066141,0.009098959],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9993299,0.000078506986,0.000045249984,0.00016745906,0.00028979755,0.00008904801],"domain_scores_gemma":[0.99933213,0.00022479994,0.00008240208,0.000111007925,0.00012448682,0.00012507408],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010756672,0.0017440767,0.0012165126,0.0026657144,0.0010623499,0.0027316683,0.0025024775,0.00090422493,0.04985386],"category_scores_gemma":[0.0027140598,0.0010396849,0.0015628367,0.0023091917,0.00038566723,0.0021611822,0.0032373308,0.0019424177,0.013285103],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0017928372,0.00020841659,0.00670001,0.004979443,0.0007404981,0.0010465363,0.0009331661,0.008356611,0.10397199,0.010190825,0.77790827,0.083171405],"study_design_scores_gemma":[0.001387827,0.00035996895,0.037573155,0.0012444953,0.00037260674,0.001790819,0.0006148035,0.18608262,0.08317018,0.031091241,0.6557198,0.0005924954],"about_ca_topic_score_codex":0.0069423914,"about_ca_topic_score_gemma":0.008479224,"teacher_disagreement_score":0.04985386,"about_ca_system_score_codex":0.0009213068,"about_ca_system_score_gemma":0.0016317096,"threshold_uncertainty_score":0.16677779},"labels":[],"label_agreement":null},{"id":"W3163233738","doi":"10.1093/bioinformatics/btab340","title":"CLEP: a hybrid data- and knowledge-driven framework for generating patient representations","year":2021,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Machine Learning in Healthcare","field":"Computer Science","cited_by":16,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Institute of Biomedical Imaging and Bioengineering; Canadian Institutes of Health Research; National Institutes of Health; Genentech; IXICO; H. Lundbeck A/S; Servier; Eisai; Bundesministerium für Bildung und Forschung; Northern California Institute for Research and Education; Pfizer; Biogen; BioClinica; F. Hoffmann-La Roche; University of Southern California; Eli Lilly and Company; U.S. Department of Defense; Meso Scale Diagnostics; Alzheimer's Disease Neuroimaging Initiative; Novartis Pharmaceuticals Corporation; Bristol-Myers Squibb; National Institute on Aging; Alzheimer's Association; Foundation for the National Institutes of Health","keywords":"Computer science; Python (programming language); Embedding; Cluster analysis; Documentation; Machine learning; Domain knowledge; Artificial intelligence; Graph; Knowledge graph; Data mining; Theoretical computer science; Programming language","score_opus":0.05388756787387535,"score_gpt":0.35343735952302524,"score_spread":0.2995497916491499,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3163233738","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0028823945,0.00023128044,0.9491279,0.0014262532,0.00012671358,0.00044662942,0.015023593,0.028776798,0.0019583642],"genre_scores_gemma":[0.06275849,0.0003118946,0.88642615,0.0012484724,0.00012316093,0.0012930742,0.0417883,0.0032978235,0.0027526075],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99824834,0.0006904104,0.00011644932,0.00045930344,0.00038028977,0.00010512166],"domain_scores_gemma":[0.99521554,0.0027907297,0.00024792735,0.0009262796,0.0005626376,0.0002569039],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0035442729,0.0013387334,0.00086301594,0.0025093332,0.00065307197,0.0019827564,0.003096457,0.0019441249,0.018617157],"category_scores_gemma":[0.017940719,0.00076727883,0.0024039827,0.0018475051,0.0008169588,0.0019452707,0.0043362672,0.0034285805,0.006831353],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00079845515,0.00040951878,0.008712297,0.0011786366,0.00047538866,0.0011909604,0.00063442753,0.1214907,0.004047835,0.046555612,0.23081735,0.58368874],"study_design_scores_gemma":[0.00023412865,0.00011229985,0.0016947718,0.00022154239,0.00007598457,0.00077794323,0.00014627705,0.75205845,0.005463444,0.16123387,0.077873476,0.000107875116],"about_ca_topic_score_codex":0.005314221,"about_ca_topic_score_gemma":0.012209789,"teacher_disagreement_score":0.018617157,"about_ca_system_score_codex":0.0012033347,"about_ca_system_score_gemma":0.003275952,"threshold_uncertainty_score":0.062280595},"labels":[],"label_agreement":null},{"id":"W3174345118","doi":"10.1093/bioinformatics/btab457","title":"VSS: variance-stabilized signals for sequencing-based genomic signals","year":2021,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Fractal and DNA sequence analysis","field":"Biochemistry, Genetics and Molecular Biology","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Variance (accounting); Computer science; Imputation (statistics); Data mining; Computational biology; Biology; Machine learning; Missing data","score_opus":0.021658282613532025,"score_gpt":0.2577184315385112,"score_spread":0.23606014892497917,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3174345118","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.009669426,0.00013267413,0.9689612,0.00022274873,0.00009585914,0.000058982132,0.0011994862,0.018943358,0.00071636],"genre_scores_gemma":[0.1660948,0.00018862633,0.8172778,0.00043588012,0.00015346361,0.00043751477,0.0067259516,0.0057365187,0.0029494434],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99804986,0.00049533264,0.00010108164,0.00058740756,0.0006746148,0.00009166002],"domain_scores_gemma":[0.9961133,0.0018107098,0.00047115356,0.00078721705,0.0006534224,0.00016405754],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0028630404,0.0012427635,0.0007945199,0.0012421694,0.0004898982,0.0011940139,0.0020572022,0.0013630079,0.0065366793],"category_scores_gemma":[0.0150203,0.00063781824,0.001124337,0.00121563,0.0014025831,0.0013518848,0.001689865,0.001986218,0.0043229554],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013046367,0.00026518258,0.008060289,0.0008687671,0.00027510666,0.00037515527,0.00051916,0.19076477,0.19011928,0.054472476,0.0474545,0.5055207],"study_design_scores_gemma":[0.000091171714,0.00017437397,0.0016665823,0.00004653901,0.00002981668,0.00022022345,0.000050010414,0.87426424,0.06223768,0.04419845,0.016950112,0.00007087545],"about_ca_topic_score_codex":0.0010901983,"about_ca_topic_score_gemma":0.0016248015,"teacher_disagreement_score":0.0065366793,"about_ca_system_score_codex":0.0007373649,"about_ca_system_score_gemma":0.0010210649,"threshold_uncertainty_score":0.021867394},"labels":[],"label_agreement":null},{"id":"W3174992442","doi":"10.1093/bioinformatics/btab459","title":"JBrowseR: an R interface to the JBrowse 2 genome browser","year":2021,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":26,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ontario Institute for Cancer Research","funders":"National Institute of General Medical Sciences; National Cancer Institute; National Human Genome Research Institute; National Institutes of Health","keywords":"Genome browser; Computer science; Interface (matter); Web browser; Variety (cybernetics); Visualization; Download; World Wide Web; Source code; Genome; User interface; Software; Code (set theory); The Internet; Genomics; Operating system; Biology; Programming language; Set (abstract data type); Genetics; Data mining","score_opus":0.013206211296484414,"score_gpt":0.2460707830635001,"score_spread":0.2328645717670157,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3174992442","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0021166576,0.0022512244,0.34524634,0.0017308935,0.000738033,0.00042779907,0.17461437,0.46300274,0.009871902],"genre_scores_gemma":[0.015864803,0.0021667024,0.49038818,0.0018019007,0.00038756375,0.0025539696,0.17932263,0.29555237,0.011961918],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9930229,0.0026015337,0.00070777565,0.001464671,0.0017522854,0.00045068248],"domain_scores_gemma":[0.9881742,0.0052697714,0.0015767749,0.0024598958,0.0017786411,0.0007407426],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0099900905,0.0036321227,0.003724636,0.0052378355,0.0014569529,0.0046102703,0.0077315676,0.0025945245,0.1365757],"category_scores_gemma":[0.031465475,0.004118872,0.0029751132,0.0043936796,0.0014510578,0.0047947867,0.005880542,0.0058684703,0.21963224],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006700961,0.00006268123,0.00225024,0.0026296228,0.0005303748,0.00035087275,0.00034729252,0.0020779693,0.010412771,0.007856626,0.930541,0.042270444],"study_design_scores_gemma":[0.00052186113,0.00006969451,0.0033670359,0.00094686385,0.00024847346,0.0011068617,0.00012750152,0.0098454505,0.014664691,0.025133723,0.94356066,0.0004072315],"about_ca_topic_score_codex":0.004454178,"about_ca_topic_score_gemma":0.0036388054,"teacher_disagreement_score":0.1365757,"about_ca_system_score_codex":0.00081694126,"about_ca_system_score_gemma":0.004058767,"threshold_uncertainty_score":0.45689124},"labels":[],"label_agreement":null},{"id":"W3175203932","doi":"10.1093/bioinformatics/btab482","title":"phyLoSTM: a novel deep learning model on disease prediction from longitudinal microbiome data","year":2021,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gut microbiota and health","field":"Biochemistry, Genetics and Molecular Biology","cited_by":56,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Public Health Ontario; University of Toronto; Princess Margaret Cancer Centre; University Health Network","funders":"Natural Sciences and Engineering Research Council of Canada; Leona M. and Harry B. Helmsley Charitable Trust","keywords":"Microbiome; Computer science; Artificial intelligence; Machine learning; Random forest; Convolutional neural network; Longitudinal study; Deep learning; Bioinformatics; Biology; Medicine; Pathology","score_opus":0.037897150233829044,"score_gpt":0.27322617826641643,"score_spread":0.23532902803258737,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3175203932","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.110791534,0.0021485551,0.87050277,0.0025954242,0.00028288437,0.00014819729,0.0045375233,0.007256576,0.0017364462],"genre_scores_gemma":[0.78397036,0.0009343458,0.20216037,0.0011489593,0.00022225591,0.00055783987,0.0060801916,0.00023707672,0.0046886923],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9997359,0.00008796382,0.000017982571,0.00007764839,0.00003827329,0.000042174317],"domain_scores_gemma":[0.9992747,0.00041547508,0.00008191995,0.000047909878,0.00012427231,0.000055730983],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016804065,0.000978909,0.000819134,0.00084177003,0.00032310918,0.00068433926,0.0018827503,0.0012758839,0.0019952888],"category_scores_gemma":[0.0034066378,0.00046576528,0.0009829833,0.00073152885,0.00041973332,0.0008959127,0.0013303427,0.0017002643,0.0005502389],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00062681356,0.00024020235,0.017069755,0.00018322405,0.0003089393,0.00023940559,0.000066178545,0.7773853,0.0015408522,0.0031664586,0.010036395,0.18913637],"study_design_scores_gemma":[0.000011083083,0.00002444708,0.00022262581,0.000008102573,0.000010322178,0.000014810917,0.0000027958304,0.99767655,0.00022185878,0.0015235738,0.0002801203,0.0000036438917],"about_ca_topic_score_codex":0.009186469,"about_ca_topic_score_gemma":0.009598367,"teacher_disagreement_score":0.009186469,"about_ca_system_score_codex":0.00096421404,"about_ca_system_score_gemma":0.0014712388,"threshold_uncertainty_score":0.018266022},"labels":[],"label_agreement":null},{"id":"W3180080324","doi":"10.1093/bioinformatics/btab524","title":"Refget: standardized access to reference sequences","year":2021,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Rare Diseases","field":"Biochemistry, Genetics and Molecular Biology","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ontario Institute for Cancer Research","funders":"European Molecular Biology Laboratory; National Institute of General Medical Sciences; Wellcome Trust","keywords":"Implementation; Computer science; Identifier; Cloud computing; Suite; Identification (biology); Database; Data mining; Programming language; Biology; Operating system","score_opus":0.02716707637083181,"score_gpt":0.29908076257383037,"score_spread":0.27191368620299855,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3180080324","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.013923266,0.0016028251,0.6174359,0.0031244082,0.0010193216,0.0017779343,0.11890251,0.21764289,0.02457087],"genre_scores_gemma":[0.045567494,0.0017358377,0.39804325,0.0050522527,0.00050737994,0.004144153,0.46526676,0.064131506,0.01555145],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.98146343,0.00705385,0.0041501964,0.002347338,0.0039299796,0.0010551874],"domain_scores_gemma":[0.9290098,0.028927984,0.0048362557,0.020036738,0.015508641,0.0016806123],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.032740764,0.0026636524,0.0015647835,0.00576541,0.0016443572,0.0048212376,0.006026803,0.004014575,0.03731217],"category_scores_gemma":[0.09940531,0.002197732,0.0018256415,0.0048800376,0.0013946301,0.00650202,0.0068158777,0.004655866,0.06193992],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0036149034,0.00048064563,0.009916281,0.0052333907,0.00026050833,0.00167524,0.0021464308,0.005569547,0.043112636,0.08210444,0.6252553,0.22063072],"study_design_scores_gemma":[0.0004271298,0.0003707526,0.0029685604,0.0018979462,0.00011051862,0.0014205996,0.0003252647,0.0056156935,0.043966327,0.032467242,0.910146,0.00028395455],"about_ca_topic_score_codex":0.0036413837,"about_ca_topic_score_gemma":0.0029033124,"teacher_disagreement_score":0.03731217,"about_ca_system_score_codex":0.0015791722,"about_ca_system_score_gemma":0.00500468,"threshold_uncertainty_score":0.17315173},"labels":[],"label_agreement":null},{"id":"W3180876344","doi":"10.1093/bioinformatics/btab292","title":"Expected 10-anonymity of HyperLogLog sketches for federated queries of clinical data repositories","year":2021,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Privacy-Preserving Technologies in Data","field":"Computer Science","cited_by":12,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"The Scarborough Hospital; University of Toronto","funders":"National Cancer Institute; Division of Mathematical Sciences; University of Toronto","keywords":"Anonymity; Computer science; Confidentiality; Sketch; Probabilistic logic; Aggregate (composite); Sorting; Protocol (science); Code (set theory); Information retrieval; Data mining; Computer security; Algorithm; Programming language","score_opus":0.13502470347922566,"score_gpt":0.36424811112855326,"score_spread":0.2292234076493276,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3180876344","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.13339554,0.0012067864,0.84120315,0.008212109,0.00016403054,0.00043676095,0.003913159,0.0030842184,0.008384318],"genre_scores_gemma":[0.91568184,0.00062010705,0.07636975,0.001068562,0.0001555832,0.0004576031,0.0015131467,0.0003729251,0.0037606088],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9844141,0.0061638164,0.0013022382,0.002628948,0.0036217466,0.0018691756],"domain_scores_gemma":[0.80589515,0.150467,0.0067789746,0.027850715,0.0053542145,0.0036539147],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.021235596,0.0013456342,0.0022581834,0.0017068117,0.0025003739,0.008495756,0.003671787,0.0041617854,0.008360478],"category_scores_gemma":[0.13227065,0.0016818029,0.0027604182,0.002028525,0.004521084,0.016119156,0.0080939215,0.005522515,0.0019162394],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0025967958,0.00031777457,0.0096624615,0.0010306172,0.00036183067,0.0008766025,0.0021155002,0.395434,0.005947082,0.52151775,0.011846804,0.048292875],"study_design_scores_gemma":[0.00013361087,0.00012712719,0.0005907344,0.00013160844,0.00007715243,0.00051216845,0.00027532657,0.54455256,0.0036317662,0.44724622,0.0026340699,0.00008775628],"about_ca_topic_score_codex":0.0014612031,"about_ca_topic_score_gemma":0.001074184,"teacher_disagreement_score":0.021235596,"about_ca_system_score_codex":0.00525926,"about_ca_system_score_gemma":0.0039282697,"threshold_uncertainty_score":0.11230594},"labels":[],"label_agreement":null},{"id":"W3182507784","doi":"10.1093/bioinformatics/btab328","title":"Gene tree and species tree reconciliation with endosymbiotic gene transfer","year":2021,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":15,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal; Université de Sherbrooke","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Biology; Genome; Gene duplication; Horizontal gene transfer; Mitochondrial DNA; Gene; Nuclear gene; Genetics; Genome evolution; Computational biology; Eukaryote; Evolutionary biology","score_opus":0.013586332068311958,"score_gpt":0.19533522625582644,"score_spread":0.1817488941875145,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3182507784","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.20700604,0.0010605439,0.7786772,0.0015288696,0.0001489793,0.00028136227,0.0014502335,0.00426666,0.005580103],"genre_scores_gemma":[0.3459969,0.00023222333,0.6472078,0.00019949739,0.0000911407,0.00027131618,0.0028869961,0.0008505506,0.0022636123],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9984862,0.0005211306,0.00006932,0.00043633106,0.0002514619,0.00023554744],"domain_scores_gemma":[0.9946238,0.003386629,0.00046629895,0.0006624304,0.0005213911,0.0003394687],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0025116366,0.0013828586,0.0019704772,0.0029301334,0.0015597637,0.002033455,0.0032568583,0.0025393916,0.0057992707],"category_scores_gemma":[0.012010179,0.00064512395,0.0018425031,0.0027506407,0.0014512084,0.0033616994,0.0023501061,0.002023036,0.0012455093],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00070698006,0.00022225502,0.003367733,0.00039468112,0.0001458659,0.00029490094,0.0005078029,0.8042471,0.003507057,0.023435805,0.007229403,0.15594047],"study_design_scores_gemma":[0.00007253142,0.00009846884,0.0005975659,0.00003196193,0.000045151377,0.00015392434,0.00012067485,0.95565796,0.0018471247,0.03923545,0.0021194501,0.000019617362],"about_ca_topic_score_codex":0.0032085488,"about_ca_topic_score_gemma":0.0044606826,"teacher_disagreement_score":0.0057992707,"about_ca_system_score_codex":0.0021705993,"about_ca_system_score_gemma":0.0022191606,"threshold_uncertainty_score":0.019400477},"labels":[],"label_agreement":null},{"id":"W3185042382","doi":"10.1093/bioinformatics/btab527","title":"miRAnno—network-based functional microRNA annotation","year":2021,"lang":"en","type":"article","venue":"Bioinformatics","topic":"MicroRNA in disease regulation","field":"Biochemistry, Genetics and Molecular Biology","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto; Discovery Centre; University Health Network","funders":"Canada Foundation for Innovation; University Health Network; International Business Machines Corporation","keywords":"Annotation; microRNA; Computer science; Computational biology; Noise (video); Data mining; Artificial intelligence; Biology; Genetics; Gene","score_opus":0.012221069776055561,"score_gpt":0.22650080378046908,"score_spread":0.21427973400441352,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3185042382","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.035256725,0.0023398297,0.7166308,0.0007309242,0.00033026875,0.0004890388,0.054809913,0.17889237,0.010520201],"genre_scores_gemma":[0.21610579,0.0014272343,0.6519153,0.000616058,0.00018140991,0.0017036247,0.1089545,0.011755699,0.007340365],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99848324,0.00039239577,0.000088193585,0.00056688045,0.0003651937,0.000104158236],"domain_scores_gemma":[0.99881566,0.0005289272,0.00018703302,0.00017713572,0.00018650472,0.00010474626],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0017550648,0.0020650427,0.0009618033,0.0026484532,0.00090002554,0.0014797399,0.0016520732,0.00078914565,0.0070185917],"category_scores_gemma":[0.0041756975,0.0007138076,0.0012921921,0.0013123646,0.00056899496,0.0014907172,0.002010457,0.0010564191,0.0058643627],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0039764154,0.00037129573,0.027625676,0.008832152,0.001448825,0.0011192524,0.001003678,0.07750259,0.2531369,0.04158749,0.2424736,0.34092218],"study_design_scores_gemma":[0.0001542154,0.00034135574,0.013012666,0.00040031882,0.0004111066,0.0015102531,0.0001544328,0.5678986,0.1458786,0.05170643,0.21823841,0.00029366234],"about_ca_topic_score_codex":0.0015265116,"about_ca_topic_score_gemma":0.0020762526,"teacher_disagreement_score":0.0070185917,"about_ca_system_score_codex":0.000647302,"about_ca_system_score_gemma":0.0010067774,"threshold_uncertainty_score":0.023479521},"labels":[],"label_agreement":null},{"id":"W3186634360","doi":"10.1093/bioinformatics/btab534","title":"CCIP: predicting CTCF-mediated chromatin loops with transitivity","year":2021,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":12,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia, Okanagan Campus; University of British Columbia","funders":"Fundamental Research Funds for the Central Universities; National Natural Science Foundation of China","keywords":"CTCF; Chromatin; Computer science; Transitive relation; Mechanism (biology); Context (archaeology); Computational biology; Topology (electrical circuits); Biology; Enhancer; Genetics; Physics; Mathematics; Transcription factor; DNA; Gene; Combinatorics","score_opus":0.0047850384362429195,"score_gpt":0.19274127875333955,"score_spread":0.18795624031709662,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3186634360","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.23579904,0.0010694641,0.7537103,0.00023689655,0.000054037188,0.00021670418,0.001920042,0.0057814657,0.0012120032],"genre_scores_gemma":[0.8450485,0.00023960497,0.1469577,0.0001248604,0.000074312215,0.00020424939,0.005992607,0.00011109285,0.0012470249],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99961567,0.00006185676,0.000018664583,0.00016359481,0.00007886792,0.00006122006],"domain_scores_gemma":[0.9988493,0.00067366107,0.00013865292,0.000094742805,0.00015570119,0.000087983506],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00087699556,0.001007858,0.00070013804,0.0010466928,0.00036672765,0.0004114292,0.0011850981,0.0010009587,0.0010783164],"category_scores_gemma":[0.0020832336,0.0002671591,0.0008572184,0.0006742005,0.00036663254,0.0006315371,0.0004965608,0.0011419827,0.0003734453],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007437582,0.00042307438,0.04368152,0.0003358858,0.00018976952,0.00044423778,0.000071720206,0.6868228,0.03031739,0.0025510418,0.0066534104,0.22776546],"study_design_scores_gemma":[0.000012984321,0.00004315779,0.001302069,0.000003980554,0.000012256924,0.000057179357,0.0000036580918,0.99459225,0.0027980227,0.00094613014,0.0002219384,0.000006285889],"about_ca_topic_score_codex":0.004452903,"about_ca_topic_score_gemma":0.00353948,"teacher_disagreement_score":0.004452903,"about_ca_system_score_codex":0.0004104142,"about_ca_system_score_gemma":0.0007582368,"threshold_uncertainty_score":0.008853972},"labels":[],"label_agreement":null},{"id":"W3194988201","doi":"10.1093/bioinformatics/btab589","title":"PyJAMAS: open-source, multimodal segmentation and analysis of microscopy images","year":2021,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Cell Image Analysis Techniques","field":"Biochemistry, Genetics and Molecular Biology","cited_by":26,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ted Rogers Centre for Heart Research; Hospital for Sick Children; University of Toronto","funders":"Natural Sciences and Engineering Research Council of Canada; Canada Foundation for Innovation; Canadian Institutes of Health Research; Government of Ontario","keywords":"Python (programming language); Scripting language; Computer science; Graphical user interface; Open source; Segmentation; Software; Extensibility; Mac OS; Image segmentation; Open source software; Interface (matter); Image processing; Computer graphics (images); User interface; Artificial intelligence; Data mining; Programming language; Operating system; Image (mathematics)","score_opus":0.00716577433905271,"score_gpt":0.2882043148524268,"score_spread":0.2810385405133741,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3194988201","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00488565,0.0004598456,0.32717398,0.00038769405,0.00035362752,0.0002880717,0.024541108,0.637195,0.004715022],"genre_scores_gemma":[0.05711214,0.0008680553,0.6524603,0.00080922607,0.00025321604,0.0024891412,0.07118923,0.20205575,0.012763003],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9986034,0.00013947568,0.000097474614,0.0004190874,0.0005760683,0.00016453069],"domain_scores_gemma":[0.99824727,0.00050412666,0.00019119233,0.00038507496,0.00038864402,0.00028366124],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0023545327,0.0019062445,0.0017211837,0.00244721,0.0012313443,0.002997582,0.005241091,0.0012492114,0.05885307],"category_scores_gemma":[0.00698467,0.001521247,0.0017265782,0.0014404963,0.000923911,0.0034828747,0.0035637333,0.0031489565,0.0375746],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0015270882,0.00021336702,0.0024142987,0.0027442302,0.00051077,0.00053804024,0.00062504626,0.0067462143,0.08638001,0.009485158,0.6998262,0.1889896],"study_design_scores_gemma":[0.00044563174,0.00018558881,0.008300924,0.0004950078,0.00016817557,0.0013742653,0.00015485348,0.17071903,0.17807408,0.035147656,0.6043747,0.0005601705],"about_ca_topic_score_codex":0.0023597134,"about_ca_topic_score_gemma":0.0028750177,"teacher_disagreement_score":0.05885307,"about_ca_system_score_codex":0.0007654903,"about_ca_system_score_gemma":0.0022606407,"threshold_uncertainty_score":0.19688314},"labels":[],"label_agreement":null},{"id":"W3196432935","doi":"10.1093/bioinformatics/btab844","title":"RNAglib: a python package for RNA 2.5 D graphs","year":2021,"lang":"en","type":"preprint","venue":"Bioinformatics","topic":"RNA and protein synthesis mechanisms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"McGill University","funders":"Agence Nationale de la Recherche","keywords":"Python (programming language); Computer science; Theoretical computer science; Graph; Representation (politics); Artificial intelligence; Programming language","score_opus":0.018474810047135474,"score_gpt":0.25248114482782946,"score_spread":0.234006334780694,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3196432935","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0020510773,0.0003496523,0.31838855,0.00042386338,0.00028358513,0.00018916467,0.1059654,0.5618648,0.010483898],"genre_scores_gemma":[0.03933888,0.00086648494,0.39799193,0.0013604468,0.00016476877,0.0023749613,0.24896908,0.29158023,0.017353263],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9992065,0.00013142262,0.00005897392,0.00018815693,0.00030275167,0.000112255104],"domain_scores_gemma":[0.99911255,0.0004028864,0.00006851877,0.00017125813,0.0001622397,0.00008257079],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010945674,0.0026249879,0.0019396787,0.0017258782,0.0010886696,0.002502276,0.004625918,0.0011842811,0.11518774],"category_scores_gemma":[0.0040057665,0.0017297567,0.002522342,0.0018518216,0.0007077074,0.0021300185,0.0028257135,0.0037142339,0.06345397],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003438311,0.000092988186,0.0014526475,0.0019907504,0.00025322844,0.00029664004,0.00025837714,0.019116834,0.008453585,0.025551118,0.8675409,0.0746492],"study_design_scores_gemma":[0.00041881434,0.00007954702,0.0020572834,0.00027888143,0.00011196519,0.00043075447,0.0001286992,0.21716282,0.019777104,0.09425706,0.66506076,0.0002364494],"about_ca_topic_score_codex":0.0056867804,"about_ca_topic_score_gemma":0.008391227,"teacher_disagreement_score":0.11518774,"about_ca_system_score_codex":0.00095665565,"about_ca_system_score_gemma":0.001800758,"threshold_uncertainty_score":0.38534135},"labels":[],"label_agreement":null},{"id":"W3197661809","doi":"10.1093/bioinformatics/btab636","title":"mspack: efficient lossless and lossy mass spectrometry data compression","year":2021,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Advanced Proteomics Techniques and Applications","field":"Chemistry","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Lossy compression; Lossless compression; Computer science; Mass spectrometry; Compression (physics); Data compression; Algorithm; Chemistry; Artificial intelligence; Materials science; Chromatography","score_opus":0.024762895791205896,"score_gpt":0.2898678943558391,"score_spread":0.2651049985646332,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3197661809","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.09281241,0.0041920105,0.55055773,0.0014067338,0.0009050408,0.0010290407,0.015011836,0.32302073,0.011064348],"genre_scores_gemma":[0.3225453,0.0024376272,0.580822,0.0014649256,0.0004825809,0.0017039351,0.057945207,0.018461207,0.014137257],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9977921,0.00016773031,0.0002497971,0.00032958662,0.0012623766,0.00019843916],"domain_scores_gemma":[0.99706537,0.0008403352,0.00028642954,0.0008122433,0.00087272935,0.00012280379],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0018531608,0.002004284,0.00077223865,0.0024921326,0.00079119706,0.0022324445,0.0032101858,0.0011944138,0.0076157125],"category_scores_gemma":[0.008750061,0.0005470873,0.0007574914,0.002619624,0.0010401128,0.0033807552,0.0027025754,0.0015500387,0.007198806],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.003976865,0.0005161213,0.0062553235,0.0016312804,0.0003223693,0.0013561633,0.0008409826,0.029610716,0.117718264,0.012395513,0.22706215,0.5983143],"study_design_scores_gemma":[0.0005150369,0.0005970762,0.0052629746,0.00026908328,0.00010353823,0.0017576271,0.0003012518,0.39195576,0.48437288,0.014010052,0.1005646,0.00029018137],"about_ca_topic_score_codex":0.0018829018,"about_ca_topic_score_gemma":0.0012515903,"teacher_disagreement_score":0.0076157125,"about_ca_system_score_codex":0.0006971984,"about_ca_system_score_gemma":0.0010668042,"threshold_uncertainty_score":0.025477111},"labels":[],"label_agreement":null},{"id":"W3199498600","doi":"10.1093/bioinformatics/btab664","title":"timeOmics: an R package for longitudinal multi-omics data integration","year":2021,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":41,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université Laval","funders":"National Health and Medical Research Council; Medical Research Council","keywords":"R package; Computer science; Longitudinal data; Computational biology; Omics; Data integration; Data mining; Data science; Biology; Bioinformatics; Computational science","score_opus":0.04835382042161563,"score_gpt":0.29535634491152235,"score_spread":0.24700252448990673,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3199498600","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.003759615,0.0012952727,0.46277696,0.0015113567,0.0006905272,0.00062227336,0.20856358,0.31588882,0.0048915516],"genre_scores_gemma":[0.052209333,0.0018131295,0.511221,0.0020297975,0.00051007525,0.0065800264,0.2137539,0.20382625,0.008056545],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99600595,0.0015340081,0.0004344342,0.00093220326,0.00078493357,0.00030845476],"domain_scores_gemma":[0.9870214,0.007752056,0.001347332,0.001816196,0.0014400702,0.00062289863],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007967666,0.0036839584,0.002406818,0.0032067162,0.0008590737,0.0040449104,0.003583204,0.0011918503,0.06869573],"category_scores_gemma":[0.035515808,0.0019412384,0.0040470758,0.0034857027,0.0011900442,0.003096348,0.0049463892,0.0038969729,0.039968736],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0018544919,0.00010992708,0.012594421,0.0055509023,0.0030546666,0.00070042943,0.00071957364,0.0149243735,0.0073193065,0.021498818,0.8440116,0.08766143],"study_design_scores_gemma":[0.0009926927,0.0002962008,0.013159592,0.0010204613,0.0012470871,0.0011389911,0.00026781674,0.08614065,0.013560577,0.11078666,0.7708987,0.00049070694],"about_ca_topic_score_codex":0.003422574,"about_ca_topic_score_gemma":0.003958386,"teacher_disagreement_score":0.06869573,"about_ca_system_score_codex":0.00091860397,"about_ca_system_score_gemma":0.004644475,"threshold_uncertainty_score":0.22981012},"labels":[],"label_agreement":null},{"id":"W3210392179","doi":"10.1093/bioinformatics/btab742","title":"SigTools: exploratory visualization for genomic signals","year":2021,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Fractal and DNA sequence analysis","field":"Biochemistry, Genetics and Molecular Biology","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Visualization; Computer science; Scope (computer science); Source code; Software; Autocorrelation; Data mining; Data science; Programming language","score_opus":0.021617557391683015,"score_gpt":0.2717709809975664,"score_spread":0.2501534236058834,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3210392179","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.007552118,0.0004010008,0.3582203,0.0009544304,0.0003551962,0.0002309768,0.041243877,0.583992,0.0070500127],"genre_scores_gemma":[0.097290464,0.0011588552,0.60305244,0.0013418626,0.000347867,0.0017177183,0.093906134,0.18715976,0.014024914],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99857056,0.00030167328,0.00011506635,0.00026458837,0.00057615235,0.00017184368],"domain_scores_gemma":[0.9929482,0.0044852966,0.00039697415,0.00072573405,0.0010228544,0.0004209801],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0026400515,0.00284742,0.0012004662,0.004515441,0.000995505,0.0037194726,0.0029475286,0.0012476156,0.068482734],"category_scores_gemma":[0.011241964,0.00097420905,0.0020275416,0.0031674302,0.00089992007,0.0031936713,0.0040142704,0.0022998268,0.022232419],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0016810208,0.00014863897,0.0038861437,0.002487069,0.00033858133,0.00089684455,0.0021319801,0.007214687,0.022706727,0.014905655,0.6914702,0.25213245],"study_design_scores_gemma":[0.00061742635,0.00024662892,0.008586964,0.0010188984,0.00021093723,0.0013616532,0.0010940143,0.16260241,0.07257171,0.065174215,0.68588555,0.00062969443],"about_ca_topic_score_codex":0.004105182,"about_ca_topic_score_gemma":0.0037563164,"teacher_disagreement_score":0.068482734,"about_ca_system_score_codex":0.0007066059,"about_ca_system_score_gemma":0.0020084362,"threshold_uncertainty_score":0.22909755},"labels":[],"label_agreement":null},{"id":"W3214075907","doi":"10.1093/bioinformatics/btab768","title":"<scp>Verna</scp>l: a tool for mining fuzzy network motifs in RNA","year":2021,"lang":"en","type":"article","venue":"Bioinformatics","topic":"RNA and protein synthesis mechanisms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"McGill University; Mila - Quebec Artificial Intelligence Institute","funders":"Natural Sciences and Engineering Research Council of Canada; Agence Nationale de la Recherche","keywords":"Motif (music); Computer science; Structural motif; Cluster analysis; Source code; Web server; Data mining; Theoretical computer science; Artificial intelligence; The Internet; World Wide Web; Biology","score_opus":0.013765716585605228,"score_gpt":0.23308839328544625,"score_spread":0.219322676699841,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3214075907","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.016053226,0.00052740047,0.6099739,0.00036001895,0.00013349303,0.00040773398,0.03385363,0.3301416,0.008548966],"genre_scores_gemma":[0.10649845,0.00034820865,0.8098052,0.0002896085,0.00006509207,0.0010646079,0.05222147,0.020979397,0.008727905],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9995598,0.000048235157,0.00003393903,0.00015434323,0.00017312645,0.000030543688],"domain_scores_gemma":[0.99861073,0.0008605877,0.00011499309,0.0001643185,0.00017291111,0.00007636864],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000854313,0.001566055,0.00074378436,0.003604461,0.00084090896,0.001970766,0.0025745106,0.00093441666,0.053719684],"category_scores_gemma":[0.0054521365,0.0007936811,0.0014447792,0.0018895034,0.0008089665,0.0015697089,0.0016536391,0.0010223215,0.009994511],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008561032,0.00022464896,0.009787796,0.0034256477,0.00041602494,0.0015501576,0.0008041637,0.058423795,0.03348994,0.025722748,0.3897911,0.4755078],"study_design_scores_gemma":[0.00026341397,0.0001343385,0.003898782,0.00025308284,0.000084084044,0.0013480937,0.0001770421,0.7765343,0.032565363,0.04007979,0.14449774,0.00016398719],"about_ca_topic_score_codex":0.0053395773,"about_ca_topic_score_gemma":0.013171424,"teacher_disagreement_score":0.053719684,"about_ca_system_score_codex":0.0008461391,"about_ca_system_score_gemma":0.0010573082,"threshold_uncertainty_score":0.17971027},"labels":[],"label_agreement":null},{"id":"W3214341251","doi":"10.1093/bioinformatics/btab771","title":"DD-GUI: a graphical user interface for deep learning-accelerated virtual screening of large chemical libraries (Deep Docking)","year":2021,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Computational Drug Discovery Methods","field":"Computer Science","cited_by":19,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"Canadian Institutes of Health Research","keywords":"Graphical user interface; Computer science; MIT License; License; Docking (animal); Virtual screening; Deep learning; User interface; Drug discovery; Software; Human–computer interaction; Data mining; Artificial intelligence; Operating system; Bioinformatics","score_opus":0.02893742099918154,"score_gpt":0.3022683980524978,"score_spread":0.2733309770533162,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3214341251","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.004408245,0.00039682924,0.37932503,0.00044217348,0.00020700284,0.0002654876,0.026621165,0.58056587,0.0077681732],"genre_scores_gemma":[0.15297778,0.0013201756,0.5542536,0.0038513786,0.00022638429,0.0033193217,0.1040899,0.14061262,0.039348822],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99947864,0.000108527434,0.000048812853,0.00010164754,0.00018902033,0.00007345246],"domain_scores_gemma":[0.99818796,0.0010389894,0.00007230258,0.00020654054,0.000302755,0.0001914552],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013503407,0.0019696935,0.000895855,0.0012375519,0.00025251877,0.0009493542,0.0028955112,0.001021168,0.12068978],"category_scores_gemma":[0.005259964,0.00070399157,0.0010276228,0.00062331493,0.0003355233,0.0012253772,0.0026825122,0.0021231752,0.03645019],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013472823,0.00025407973,0.0026562943,0.0010612959,0.00023148059,0.00041376293,0.00022663844,0.010126087,0.017413624,0.007962305,0.770307,0.1880002],"study_design_scores_gemma":[0.0018713558,0.00036411887,0.007290397,0.00045063495,0.00013719907,0.001058908,0.00008027195,0.3483663,0.061739556,0.0307967,0.5473723,0.00047234178],"about_ca_topic_score_codex":0.0019150996,"about_ca_topic_score_gemma":0.0029055467,"teacher_disagreement_score":0.12068978,"about_ca_system_score_codex":0.0005658271,"about_ca_system_score_gemma":0.00082410075,"threshold_uncertainty_score":0.40374756},"labels":[],"label_agreement":null},{"id":"W3214361435","doi":"10.1093/bioinformatics/btab754","title":"Mian: interactive web-based microbiome data table visualization and machine learning platform","year":2021,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gut microbiota and health","field":"Biochemistry, Genetics and Molecular Biology","cited_by":21,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Computer science; Context (archaeology); Categorical variable; Source code; Visualization; Machine learning; Metadata; Data mining; Artificial intelligence; MIT License; Table (database); License; Information retrieval; World Wide Web; Programming language","score_opus":0.024662368003439315,"score_gpt":0.2864016975794296,"score_spread":0.26173932957599033,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3214361435","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.005611674,0.0004291663,0.10962398,0.0008830094,0.00045114444,0.00046688927,0.22535764,0.6462479,0.010928639],"genre_scores_gemma":[0.0639533,0.0006167029,0.44571257,0.0018762161,0.0003307937,0.004638734,0.36876982,0.0965382,0.017563658],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9990362,0.00016798188,0.00007960575,0.00022517145,0.00038302335,0.00010816154],"domain_scores_gemma":[0.99793917,0.0008765501,0.00016205874,0.00033467807,0.00040921193,0.00027830747],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002521375,0.002047977,0.0012573729,0.0035652525,0.0009197057,0.0024420726,0.0027387463,0.0009789333,0.12860394],"category_scores_gemma":[0.007999006,0.0010510334,0.0016115446,0.001974586,0.0003928853,0.0030935185,0.00406537,0.0023496398,0.03916667],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0012182079,0.00018556888,0.0035847146,0.0011699537,0.0003293215,0.0002806443,0.00042276888,0.0022802986,0.009976575,0.0064506778,0.8960186,0.07808261],"study_design_scores_gemma":[0.0008849519,0.0002238715,0.01468858,0.00066369656,0.00018974014,0.0005712224,0.0002685126,0.13091443,0.03278505,0.048171766,0.7701928,0.00044525307],"about_ca_topic_score_codex":0.0036604062,"about_ca_topic_score_gemma":0.007091119,"teacher_disagreement_score":0.12860394,"about_ca_system_score_codex":0.00085362146,"about_ca_system_score_gemma":0.0019389044,"threshold_uncertainty_score":0.43022305},"labels":[],"label_agreement":null},{"id":"W4200168537","doi":"10.1093/bioinformatics/btab836","title":"TMBleR: a bioinformatic tool to optimize TMB estimation and predictive power","year":2021,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Cancer Immunotherapy and Biomarkers","field":"Medicine","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Izaak Walton Killam Health Centre; Dalhousie University","funders":"Ministero della Salute","keywords":"MIT License; Computer science; License; Container (type theory); Exome; Open source; Interface (matter); Heuristic; Predictive power; Data mining; Computational biology; Exome sequencing; Software; Gene; Operating system; Artificial intelligence; Mutation; Biology","score_opus":0.01103806359430913,"score_gpt":0.2568978572876185,"score_spread":0.24585979369330935,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4200168537","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.017936096,0.0017193474,0.5841761,0.0014404813,0.0006326501,0.00045990737,0.05013258,0.3396864,0.0038164242],"genre_scores_gemma":[0.1091047,0.0009963186,0.75778526,0.0013548312,0.00045638834,0.0020702197,0.0597217,0.06408349,0.0044271415],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99766326,0.0008614676,0.00021965707,0.0007176799,0.00039233538,0.00014556169],"domain_scores_gemma":[0.9900029,0.007995926,0.00047814244,0.00051154057,0.0007148217,0.00029677275],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.006951717,0.0039896662,0.0024282006,0.0036977967,0.0011371023,0.003777638,0.0028846818,0.0019610762,0.033860136],"category_scores_gemma":[0.036834016,0.0017008624,0.00268265,0.0023738658,0.00074943394,0.002384474,0.002973309,0.003180622,0.017670732],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.003639996,0.00056468457,0.03225319,0.004656156,0.002604445,0.0011210865,0.0010442791,0.06650082,0.017328978,0.0129945725,0.54605496,0.3112368],"study_design_scores_gemma":[0.00155723,0.00049811375,0.010074023,0.0007462246,0.00087138684,0.0014760334,0.0003176398,0.7560208,0.022440715,0.07398727,0.13153566,0.00047483845],"about_ca_topic_score_codex":0.0027041156,"about_ca_topic_score_gemma":0.004173398,"teacher_disagreement_score":0.033860136,"about_ca_system_score_codex":0.00075995224,"about_ca_system_score_gemma":0.0024830832,"threshold_uncertainty_score":0.1132735},"labels":[],"label_agreement":null},{"id":"W4200253063","doi":"10.1093/bioinformatics/btab826","title":"A network-based drug repurposing method via non-negative matrix factorization","year":2021,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Computational Drug Discovery Methods","field":"Computer Science","cited_by":44,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Windsor","funders":"","keywords":"Computer science; Non-negative matrix factorization; Drug repositioning; Matrix decomposition; Repurposing; Factorization; Matrix (chemical analysis); Drug; Algorithm; Artificial intelligence; Pharmacology; Chemistry; Medicine; Engineering; Chromatography","score_opus":0.018187499973003937,"score_gpt":0.3095715465829911,"score_spread":0.2913840466099871,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4200253063","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0051039536,0.0005152483,0.9911221,0.00029854744,0.00007327182,0.00011037,0.00030716427,0.0014010865,0.001068183],"genre_scores_gemma":[0.13748607,0.0006126024,0.8544182,0.00032954497,0.00021332083,0.00027433704,0.0017455404,0.00017132933,0.0047489973],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9992525,0.00016951314,0.000042012976,0.00022904415,0.00024045043,0.00006632344],"domain_scores_gemma":[0.99876356,0.0005506285,0.0001279796,0.000116660965,0.00036260247,0.0000786054],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011984125,0.0013576018,0.0013124133,0.0021414594,0.00081153907,0.0006860047,0.0015231813,0.0013634792,0.0044068377],"category_scores_gemma":[0.0032934318,0.0005638532,0.0016064345,0.0015425817,0.00043353683,0.0011073052,0.0007385052,0.0013417759,0.0019848486],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000293817,0.00025107627,0.0035603554,0.00040898152,0.0002843586,0.00045918344,0.00018963261,0.3646662,0.014856603,0.0079093315,0.022146916,0.58497345],"study_design_scores_gemma":[0.000018733137,0.000035119643,0.00025319515,0.0000109183175,0.000026646974,0.0001183228,0.000014031059,0.99391246,0.0013380215,0.0023387433,0.0019206483,0.000013144632],"about_ca_topic_score_codex":0.01828676,"about_ca_topic_score_gemma":0.024788953,"teacher_disagreement_score":0.01828676,"about_ca_system_score_codex":0.00081684854,"about_ca_system_score_gemma":0.0016915366,"threshold_uncertainty_score":0.03636062},"labels":[],"label_agreement":null},{"id":"W4200301447","doi":"10.1093/bioinformatics/btab852","title":"geoCancerPrognosticDatasetsRetriever: a bioinformatics tool to easily identify cancer prognostic datasets on Gene Expression Omnibus (GEO)","year":2021,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":40,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Perl; Computer science; Data mining; DNA microarray; MIT License; Download; Bioinformatics; Computational biology; Information retrieval; Software; Gene; Gene expression; World Wide Web; Biology; Programming language","score_opus":0.01894596870205884,"score_gpt":0.2948328055811739,"score_spread":0.2758868368791151,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4200301447","genre_codex":"dataset","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0023777126,0.0006315537,0.059415385,0.00085330766,0.0003967649,0.00043353878,0.4956638,0.43495724,0.005270723],"genre_scores_gemma":[0.0125802625,0.0007933337,0.10677537,0.001444875,0.00016463116,0.0026717598,0.8099811,0.060203984,0.005384739],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9964384,0.0005824032,0.0004929592,0.0009535453,0.0011949375,0.00033778162],"domain_scores_gemma":[0.99315554,0.0027906178,0.0009214797,0.0015028046,0.0011225311,0.0005070817],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0054847943,0.0031470368,0.003039443,0.008329907,0.0016948001,0.003830023,0.0055654417,0.0024082975,0.06038751],"category_scores_gemma":[0.016128045,0.0028088614,0.002513537,0.009219492,0.00097089325,0.0036305906,0.0062042163,0.0036912586,0.06334682],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00087369845,0.000098166485,0.002559881,0.0024464414,0.00025027772,0.00042797983,0.00044705166,0.0015590934,0.008997347,0.0046151634,0.93925,0.038474917],"study_design_scores_gemma":[0.00087942026,0.0001638145,0.0075884443,0.0005005888,0.00016130402,0.00060142763,0.00026691298,0.011867709,0.02556372,0.021088334,0.9309691,0.00034908787],"about_ca_topic_score_codex":0.0053960024,"about_ca_topic_score_gemma":0.007738518,"teacher_disagreement_score":0.06038751,"about_ca_system_score_codex":0.0016189149,"about_ca_system_score_gemma":0.004108706,"threshold_uncertainty_score":0.20201635},"labels":[],"label_agreement":null},{"id":"W4206241205","doi":"10.1093/bioinformatics/btab874","title":"SPEQ: quality assessment of peptide tandem mass spectra with deep learning","year":2021,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Advanced Proteomics Techniques and Applications","field":"Chemistry","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Tandem; Computer science; Quality (philosophy); Peptide; Quality assurance; Quality assessment; Artificial intelligence; Chemistry; Evaluation methods; Materials science; Reliability engineering; Physics; Biochemistry; Engineering; External quality assessment; Operations management","score_opus":0.021111677440230032,"score_gpt":0.3115698301347636,"score_spread":0.29045815269453357,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4206241205","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.10566539,0.0008914602,0.7772368,0.0009003589,0.00028303004,0.00053560303,0.010002967,0.10167098,0.0028134135],"genre_scores_gemma":[0.36068666,0.0005111985,0.6079799,0.00074668555,0.000095919575,0.0006338041,0.019969301,0.00589154,0.0034850305],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99801505,0.00037702205,0.00017910246,0.0004118626,0.0008674307,0.00014959591],"domain_scores_gemma":[0.9941553,0.002339028,0.00086052663,0.0007227196,0.0016478703,0.00027449633],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007146463,0.0020308364,0.00079745374,0.0024096165,0.00064576924,0.0017410475,0.002645364,0.0010540496,0.0077364864],"category_scores_gemma":[0.015876824,0.0005977226,0.0011537452,0.0015021662,0.0009032218,0.0028712323,0.0026375842,0.0021351562,0.0021236644],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0037174392,0.0009871242,0.051797602,0.0025425777,0.0012047462,0.0009769069,0.0005227375,0.1508634,0.08457478,0.009344081,0.14942496,0.54404366],"study_design_scores_gemma":[0.00013346411,0.00023107596,0.006693471,0.00007166351,0.00005330005,0.00032952608,0.000062682775,0.93169165,0.04493975,0.008235564,0.0074402653,0.000117621814],"about_ca_topic_score_codex":0.0031606953,"about_ca_topic_score_gemma":0.003328348,"teacher_disagreement_score":0.0077364864,"about_ca_system_score_codex":0.0010716119,"about_ca_system_score_gemma":0.0016670323,"threshold_uncertainty_score":0.03779453},"labels":[],"label_agreement":null},{"id":"W4206552974","doi":"10.1093/bioinformatics/btaa924","title":"Scool: a new data storage format for single-cell Hi-C data","year":2020,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Single-cell and spatial transcriptomics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Deutsche Forschungsgemeinschaft; Royal Bank of Canada; National Bar Institute","keywords":"Computer science; File format; Reuse; Database; Data format; Data file; Computer data storage; Raw data; Operating system; Programming language; Computer hardware","score_opus":0.12624920128489087,"score_gpt":0.27267642649154566,"score_spread":0.1464272252066548,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4206552974","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0054099844,0.00056390854,0.16201769,0.0009635633,0.0010936131,0.0004859434,0.34225896,0.47696236,0.010244009],"genre_scores_gemma":[0.031562537,0.00071692053,0.16323301,0.0014310419,0.00043478183,0.002957486,0.63221824,0.15182774,0.015618254],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99702877,0.00034637013,0.000534788,0.0007565077,0.0009489352,0.00038471387],"domain_scores_gemma":[0.98662484,0.0035055273,0.00095224835,0.004682194,0.0034546764,0.00078057556],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0053815553,0.0023773261,0.0017689604,0.0036133735,0.0019341597,0.0055747735,0.0055580763,0.0022238016,0.10506283],"category_scores_gemma":[0.01989207,0.0015691238,0.001777025,0.004274148,0.00097009656,0.0058611967,0.0055894824,0.0036177437,0.07839408],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0029917536,0.00008479983,0.002340647,0.0021198075,0.00019726594,0.00033197197,0.0005722152,0.0018199873,0.026813125,0.0073407805,0.8902623,0.06512527],"study_design_scores_gemma":[0.0004828785,0.00014906087,0.0054983534,0.00060005434,0.00013520164,0.00047899163,0.000301494,0.013022144,0.09502698,0.013818173,0.8699556,0.0005311256],"about_ca_topic_score_codex":0.003860986,"about_ca_topic_score_gemma":0.0033378096,"teacher_disagreement_score":0.10506283,"about_ca_system_score_codex":0.0017649316,"about_ca_system_score_gemma":0.003689004,"threshold_uncertainty_score":0.35147017},"labels":[],"label_agreement":null},{"id":"W4206944890","doi":"10.1093/bioinformatics/btab854","title":"BATL: Bayesian annotations for targeted lipidomics","year":2021,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Metabolomics and Mass Spectrometry Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"National Research Council Canada; Ottawa Hospital; University of Ottawa","funders":"National Institute of Neurological Disorders and Stroke; Natural Sciences and Engineering Research Council of Canada","keywords":"Lipidomics; Bayesian probability; Computer science; Computational biology; Data mining; Bioinformatics; Artificial intelligence; Biology","score_opus":0.010961671639871308,"score_gpt":0.250181686187481,"score_spread":0.2392200145476097,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4206944890","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0032352698,0.0003909837,0.7729702,0.0005247379,0.00019038588,0.00022247666,0.031833548,0.18727838,0.0033541596],"genre_scores_gemma":[0.03774732,0.0005283396,0.8438751,0.0007404958,0.00017301737,0.001005558,0.087838896,0.023081202,0.005010058],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9956857,0.0013052694,0.00038556583,0.00084315176,0.0015122141,0.00026824913],"domain_scores_gemma":[0.99005145,0.0053429636,0.00092173286,0.0011277745,0.0020576932,0.00049831683],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0083103655,0.0026952073,0.0013581928,0.004480351,0.001656119,0.0034330687,0.003856488,0.0027148665,0.033230178],"category_scores_gemma":[0.021845223,0.001825605,0.0020457716,0.0023105938,0.0010815045,0.004301867,0.0034677794,0.0033095116,0.03486745],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.002753495,0.00041440615,0.01069985,0.004840971,0.0004906576,0.00096005603,0.00100692,0.04005792,0.05574315,0.037558977,0.5039597,0.34151378],"study_design_scores_gemma":[0.00033114644,0.00018096456,0.0036506427,0.0008605068,0.00017964396,0.0009500031,0.00021269037,0.58710444,0.041109458,0.09252649,0.27254733,0.0003467533],"about_ca_topic_score_codex":0.0059093544,"about_ca_topic_score_gemma":0.0070131603,"teacher_disagreement_score":0.033230178,"about_ca_system_score_codex":0.0019702597,"about_ca_system_score_gemma":0.0040733726,"threshold_uncertainty_score":0.11116606},"labels":[],"label_agreement":null},{"id":"W4210425469","doi":"10.1093/bioinformatics/btac037","title":"GenomeTornadoPlot: a novel R package for CNV visualization and focality analysis","year":2022,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomic variations and chromosomal abnormalities","field":"Biochemistry, Genetics and Molecular Biology","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Canadian Institutes of Health Research; Deutschen Konsortium für Translationale Krebsforschung; Deutsche Forschungsgemeinschaft","keywords":"R package; Visualization; Computer science; Parallel computing; Computational science; Artificial intelligence","score_opus":0.014098011942831874,"score_gpt":0.24234130736854387,"score_spread":0.228243295425712,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4210425469","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.004343575,0.001241011,0.2619965,0.0007687797,0.00041866646,0.0003911186,0.1126629,0.6135947,0.0045827725],"genre_scores_gemma":[0.05064531,0.0013222427,0.38247016,0.0016344078,0.00026957793,0.0036807018,0.17018884,0.3817861,0.008002698],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.997447,0.00076011504,0.00024249137,0.0007683845,0.00057328993,0.00020879655],"domain_scores_gemma":[0.9931676,0.0043042637,0.0006961429,0.00092403806,0.0006200741,0.0002879103],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0044778013,0.0038443895,0.0025396382,0.003095078,0.0008222164,0.0037050345,0.0043329643,0.0014025904,0.06638652],"category_scores_gemma":[0.021949513,0.0022038973,0.003344631,0.0023484274,0.00092586473,0.002502549,0.0050217956,0.0038156267,0.038422987],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014365208,0.0000944757,0.008768176,0.0046975426,0.0020148137,0.0009673195,0.00081423536,0.011259006,0.007791307,0.014649494,0.84284306,0.10466415],"study_design_scores_gemma":[0.0013581618,0.00023531933,0.012003192,0.0008823368,0.0009118318,0.0017294718,0.00023604375,0.089103244,0.019308481,0.06454785,0.8091218,0.0005622864],"about_ca_topic_score_codex":0.004247655,"about_ca_topic_score_gemma":0.004330353,"teacher_disagreement_score":0.06638652,"about_ca_system_score_codex":0.00084801327,"about_ca_system_score_gemma":0.0029211324,"threshold_uncertainty_score":0.22208506},"labels":[],"label_agreement":null},{"id":"W4210607727","doi":"10.1093/bioinformatics/btac077","title":"PDMDA: predicting deep-level miRNA–disease associations with graph neural networks and sequence features","year":2022,"lang":"en","type":"article","venue":"Bioinformatics","topic":"MicroRNA in disease regulation","field":"Biochemistry, Genetics and Molecular Biology","cited_by":38,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Saskatchewan","funders":"National Natural Science Foundation of China-Zhejiang Joint Fund for the Integration of Industrialization and Informatization; Higher Education Discipline Innovation Project; Guizhou Science and Technology Department; National Natural Science Foundation of China","keywords":"Softmax function; Computer science; Artificial neural network; Artificial intelligence; Disease; Computational biology; Feature (linguistics); Association (psychology); microRNA; Machine learning; Biology; Genetics; Gene; Medicine; Pathology","score_opus":0.013996187747695224,"score_gpt":0.23173156990943583,"score_spread":0.2177353821617406,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4210607727","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.38251325,0.0027766353,0.5995946,0.0015863419,0.00020023837,0.00021672521,0.003790864,0.007088294,0.0022329676],"genre_scores_gemma":[0.8870361,0.00043355805,0.105223715,0.0004678407,0.00007255065,0.0001880762,0.004098688,0.00008739001,0.002392065],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99972934,0.00006270977,0.000017128912,0.00010679404,0.000045546945,0.000038422397],"domain_scores_gemma":[0.9993631,0.0003420428,0.0000912553,0.000051661817,0.00009846455,0.0000533865],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007195385,0.0012930774,0.0006488625,0.0011613844,0.00032499834,0.00049974676,0.0010752974,0.0010747905,0.0013621333],"category_scores_gemma":[0.0018130833,0.00042723783,0.001054692,0.00048920664,0.00032686617,0.0006457606,0.0007217361,0.0012772725,0.0003412329],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00053601107,0.0004121196,0.036622215,0.00018257966,0.0005331988,0.00034976334,0.00006796132,0.7287294,0.007241472,0.0019075982,0.007838281,0.21557939],"study_design_scores_gemma":[0.000010116174,0.000025122587,0.00083679287,0.0000045168567,0.00001651449,0.000024792465,0.000003342157,0.9971738,0.0006328223,0.0010826137,0.00018539089,0.000004105673],"about_ca_topic_score_codex":0.010747876,"about_ca_topic_score_gemma":0.013212498,"teacher_disagreement_score":0.010747876,"about_ca_system_score_codex":0.0008076271,"about_ca_system_score_gemma":0.0008742673,"threshold_uncertainty_score":0.02137065},"labels":[],"label_agreement":null},{"id":"W4210808382","doi":"10.1093/bioinformatics/btac081","title":"SmMIP-tools: a computational toolset for processing and analysis of single-molecule molecular inversion probes-derived data","year":2022,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Acute Myeloid Leukemia Research","field":"Medicine","cited_by":12,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Princess Margaret Cancer Centre; University Health Network; University of Toronto; Ontario Institute for Cancer Research","funders":"Canada First Research Excellence Fund","keywords":"Computer science; Source code; Modular design; Raw data; DNA sequencing; Set (abstract data type); Data mining; Scripting language; Computational biology; Biology; Genetics; Programming language; DNA","score_opus":0.07001876067722557,"score_gpt":0.3211005807529847,"score_spread":0.25108182007575913,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4210808382","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.010726227,0.0016903846,0.6482559,0.0011116551,0.00067467516,0.00067052507,0.048121385,0.28295046,0.005798809],"genre_scores_gemma":[0.06789661,0.0017282611,0.7782872,0.00148304,0.0002330793,0.0044538807,0.08653541,0.05499219,0.0043903072],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9983191,0.00048615935,0.00018454657,0.00037342162,0.0004805967,0.00015610461],"domain_scores_gemma":[0.99517715,0.003429311,0.00031342736,0.00038186667,0.00043122366,0.00026696193],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0045531495,0.00356814,0.002077312,0.0020634395,0.0012475401,0.002992732,0.004557286,0.001822006,0.026937066],"category_scores_gemma":[0.016530858,0.0014411474,0.0024563884,0.0019857215,0.0012670084,0.0020693275,0.0037652298,0.0048341216,0.014135566],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0027630012,0.00041868666,0.008949559,0.00782629,0.0023253274,0.0021728347,0.0013493896,0.09414323,0.02735601,0.037065603,0.6161132,0.1995169],"study_design_scores_gemma":[0.0013837053,0.00024113062,0.0028880301,0.00075854466,0.00036276685,0.00093470555,0.00026340038,0.59584844,0.04154092,0.09026835,0.26507267,0.00043734378],"about_ca_topic_score_codex":0.0036064256,"about_ca_topic_score_gemma":0.005051692,"teacher_disagreement_score":0.026937066,"about_ca_system_score_codex":0.0008154418,"about_ca_system_score_gemma":0.0040437155,"threshold_uncertainty_score":0.09011346},"labels":[],"label_agreement":null},{"id":"W4213201540","doi":"10.1093/bioinformatics/btac096","title":"CNGPLD: case–control copy-number analysis using Gaussian process latent difference","year":2022,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomic variations and chromosomal abnormalities","field":"Biochemistry, Genetics and Molecular Biology","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Center for Advancing Translational Sciences; National Institute of General Medical Sciences; National Institutes of Health; National Cancer Institute; Canadian Institutes of Health Research; Cancer Prevention and Research Institute of Texas; U.S. Department of Defense","keywords":"False discovery rate; Copy-number variation; Genome; Computational biology; Covariance; Selection (genetic algorithm); Biology; Somatic cell; Genetics; Computer science; Gene; Mathematics; Statistics; Artificial intelligence","score_opus":0.011863232954302026,"score_gpt":0.24514270016203954,"score_spread":0.2332794672077375,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4213201540","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.03071819,0.0012133903,0.8093495,0.0016169567,0.0009681854,0.0027543344,0.09759373,0.053140953,0.0026447775],"genre_scores_gemma":[0.22639875,0.00047637525,0.67989576,0.001048395,0.0003562419,0.010794914,0.062516496,0.011261644,0.007251463],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99027187,0.004464577,0.00057433033,0.0031537272,0.001258901,0.0002766139],"domain_scores_gemma":[0.9635225,0.028402956,0.001625451,0.0047031697,0.0010793383,0.0006666484],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.017654309,0.0017038123,0.002522823,0.004324926,0.0011666473,0.0026747482,0.006360213,0.0018337187,0.092932954],"category_scores_gemma":[0.08714232,0.0015005398,0.002941354,0.0038154344,0.0016113758,0.0014463348,0.004654498,0.0031686018,0.0076614716],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.012478209,0.00075304986,0.11286646,0.0060739773,0.011650194,0.0040598125,0.0011870011,0.042915862,0.0057273335,0.034845937,0.35053957,0.4169026],"study_design_scores_gemma":[0.010115269,0.0014086307,0.06054648,0.00091504486,0.0049803746,0.0052470746,0.00037375028,0.59389883,0.009870321,0.1392972,0.17272606,0.00062102056],"about_ca_topic_score_codex":0.008988278,"about_ca_topic_score_gemma":0.007932844,"teacher_disagreement_score":0.092932954,"about_ca_system_score_codex":0.0013922515,"about_ca_system_score_gemma":0.0031230396,"threshold_uncertainty_score":0.3108917},"labels":[],"label_agreement":null},{"id":"W4213235856","doi":"10.1093/bioinformatics/btac100","title":"Pre-training graph neural networks for link prediction in biomedical networks","year":2022,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":90,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"BC Research (Canada)","funders":"National Natural Science Foundation of China","keywords":"Computer science; Initialization; Graph; Artificial intelligence; Convolutional neural network; Machine learning; Artificial neural network; Node (physics); Encoder; Deep learning; Data mining; Theoretical computer science","score_opus":0.012200227654294996,"score_gpt":0.2334871319472515,"score_spread":0.22128690429295653,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4213235856","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.14008416,0.0027909284,0.83887035,0.000959174,0.00020152333,0.00022876344,0.0016380864,0.01078935,0.0044376696],"genre_scores_gemma":[0.7922683,0.0011256681,0.19452226,0.0006043484,0.000106808744,0.00027820392,0.0064179967,0.00029575478,0.0043806145],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99953127,0.00009251039,0.000027231355,0.00019111756,0.000094378,0.00006352231],"domain_scores_gemma":[0.99877805,0.0006254133,0.00013794863,0.0001517807,0.00024835515,0.000058392525],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009021749,0.0020124344,0.000872139,0.0012283442,0.00042176706,0.000517285,0.001677324,0.0014038045,0.0022582684],"category_scores_gemma":[0.003641569,0.00056292955,0.0009252179,0.0011039586,0.0006602882,0.001763213,0.00095978164,0.0021002826,0.0009032566],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00014965062,0.00016986157,0.0030434963,0.00017551264,0.00007630297,0.0001469756,0.000042385404,0.79243183,0.0046661445,0.0020334527,0.0047416287,0.1923227],"study_design_scores_gemma":[0.000004939363,0.000023440702,0.00023901877,0.0000077280065,0.000010233241,0.000013563074,0.0000053889094,0.99605477,0.0016033761,0.0016919902,0.00034159454,0.000003985647],"about_ca_topic_score_codex":0.011812448,"about_ca_topic_score_gemma":0.017553704,"teacher_disagreement_score":0.011812448,"about_ca_system_score_codex":0.00118971,"about_ca_system_score_gemma":0.0011995267,"threshold_uncertainty_score":0.02348739},"labels":[],"label_agreement":null},{"id":"W4214521759","doi":"10.1093/bioinformatics/btac134","title":"HAMdetector: a Bayesian regression model that integrates information to detect HLA-associated mutations","year":2022,"lang":"en","type":"article","venue":"Bioinformatics","topic":"vaccines and immunoinformatics approaches","field":"Biochemistry, Genetics and Molecular Biology","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"AIDS Vancouver; Simon Fraser University","funders":"Canadian Institutes of Health Research; Deutsche Forschungsgemeinschaft; Michael Smith Health Research BC","keywords":"Human leukocyte antigen; Major histocompatibility complex; CTL*; Computational biology; Biology; Epitope; Bayesian probability; Bayes' theorem; Population; Bayesian inference; Computer science; Genetics; CD8; Artificial intelligence; Immune system; Antigen","score_opus":0.016174406778392554,"score_gpt":0.2369433670564221,"score_spread":0.22076896027802956,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4214521759","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.021003652,0.0007664254,0.97383547,0.0007383252,0.0000696136,0.00009882506,0.0007607225,0.0010989478,0.0016279714],"genre_scores_gemma":[0.596693,0.001202546,0.38443235,0.0012547211,0.0003983322,0.00070970383,0.0034338597,0.0003703645,0.011505125],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9989222,0.0005309208,0.000038799455,0.00024704795,0.00015737835,0.00010360355],"domain_scores_gemma":[0.9967596,0.0023837609,0.00024853635,0.00008510852,0.00041130587,0.00011155778],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0037220216,0.001340602,0.0021115558,0.0014087838,0.00046524857,0.0011924147,0.0030989014,0.0022187151,0.0040352596],"category_scores_gemma":[0.006975871,0.0010268971,0.0013744262,0.0009319657,0.0009014266,0.001050694,0.0012398326,0.0023947798,0.0012554474],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002617134,0.00007938143,0.0032590057,0.00008093104,0.00015351159,0.000093429524,0.000038129565,0.9307559,0.0010468996,0.010055734,0.0026183762,0.051556993],"study_design_scores_gemma":[0.000012911815,0.000012583854,0.00011249182,0.0000066287344,0.000011082614,0.000010470646,0.0000016181543,0.99739456,0.00011089406,0.0020396472,0.000279996,0.0000072174976],"about_ca_topic_score_codex":0.017727487,"about_ca_topic_score_gemma":0.015139747,"teacher_disagreement_score":0.017727487,"about_ca_system_score_codex":0.0012442091,"about_ca_system_score_gemma":0.0017424977,"threshold_uncertainty_score":0.035248637},"labels":[],"label_agreement":null},{"id":"W4220970247","doi":"10.1093/bioinformatics/btac157","title":"RevUP: an online scoring system for regulatory variants implicated in rare diseases","year":2022,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Rare Diseases","field":"Biochemistry, Genetics and Molecular Biology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"BC Children's Hospital; University of British Columbia","funders":"ZonMw","keywords":"Classifier (UML); Computer science; Source code; Scoring system; Phenotype; Data mining; Computational biology; Artificial intelligence; Biology; Genetics; Medicine; Gene","score_opus":0.017269702443983927,"score_gpt":0.2424494757252676,"score_spread":0.22517977328128366,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4220970247","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.047549468,0.0027995019,0.5449449,0.0019082224,0.001058894,0.0013748318,0.14458154,0.24288571,0.012896911],"genre_scores_gemma":[0.17910673,0.0011689183,0.6266244,0.0012487354,0.0004782981,0.0022331048,0.14718923,0.031935714,0.010014941],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99528104,0.0012324568,0.00077679387,0.0012343277,0.0012201807,0.00025525616],"domain_scores_gemma":[0.98859185,0.006864068,0.001132069,0.0010865537,0.0019731298,0.00035237373],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007014542,0.0019591015,0.0013964502,0.0060652182,0.001111754,0.0026279658,0.0029436199,0.0015831476,0.044692617],"category_scores_gemma":[0.035782527,0.0010823543,0.0019998308,0.0029293157,0.0004907114,0.0018719995,0.0034051167,0.0016045436,0.021621464],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0019619793,0.00022254244,0.053705975,0.003836048,0.0008435323,0.0018547635,0.00052265637,0.010123776,0.015733855,0.008290717,0.4391979,0.46370623],"study_design_scores_gemma":[0.0016769448,0.00071590894,0.07313311,0.0020426817,0.0013688912,0.012722701,0.00054475275,0.22904657,0.06627482,0.0674052,0.5441051,0.0009632294],"about_ca_topic_score_codex":0.0019131256,"about_ca_topic_score_gemma":0.0036378736,"teacher_disagreement_score":0.044692617,"about_ca_system_score_codex":0.0008178525,"about_ca_system_score_gemma":0.002036996,"threshold_uncertainty_score":0.14951164},"labels":[],"label_agreement":null},{"id":"W4221007687","doi":"10.1093/bioinformatics/btac128","title":"SomaticSiMu: a mutational signature simulator","year":2022,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo; Western University","funders":"Compute Canada","keywords":"Python (programming language); Computer science; Visualization; Data mining; Signature (topology); Graphical user interface; Open source; Source code; Software; Computational biology; Artificial intelligence; Programming language; Biology","score_opus":0.008138215293260502,"score_gpt":0.222538426452493,"score_spread":0.2144002111592325,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4221007687","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.10784496,0.00065427483,0.593205,0.0007725868,0.0005133826,0.00057061913,0.02957352,0.24141686,0.025448795],"genre_scores_gemma":[0.5709015,0.0007108388,0.33985737,0.0011250435,0.00010527898,0.002514633,0.034905594,0.03660979,0.013269964],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9996215,0.00011144246,0.000029965511,0.00008070097,0.00010546764,0.00005097062],"domain_scores_gemma":[0.99880326,0.00071331766,0.00006685201,0.0001528519,0.00016073388,0.00010300428],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009803823,0.0009035626,0.0007683347,0.0006656665,0.00051510165,0.001113846,0.002694892,0.0012994646,0.016738635],"category_scores_gemma":[0.0038396537,0.00067774596,0.0012069995,0.0005861474,0.00046602727,0.00097203767,0.0011718448,0.0012767726,0.0042651067],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001041236,0.00035711133,0.014013853,0.0014067473,0.0004276163,0.0006869382,0.0004936528,0.77391785,0.022086455,0.034286562,0.088086,0.06319595],"study_design_scores_gemma":[0.0001179542,0.000060590304,0.0005386745,0.000034215504,0.000031826123,0.00009465953,0.000028973234,0.9543601,0.008827586,0.009572733,0.026294468,0.000038178056],"about_ca_topic_score_codex":0.0030380432,"about_ca_topic_score_gemma":0.002943823,"teacher_disagreement_score":0.016738635,"about_ca_system_score_codex":0.0008491931,"about_ca_system_score_gemma":0.0015997833,"threshold_uncertainty_score":0.0559963},"labels":[],"label_agreement":null},{"id":"W4224287786","doi":"10.1093/bioinformatics/btac287","title":"SimPlot++: a Python application for representing sequence similarity and detecting recombination","year":2022,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":155,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Agriculture and Agri-Food Canada; Université du Québec à Montréal","funders":"Natural Sciences and Engineering Research Council of Canada; Fonds Québécois de la Recherche sur la Nature et les Technologies","keywords":"Python (programming language); Computer science; Sequence (biology); Computational biology; Programming language; Biology; Genetics","score_opus":0.026194200691998164,"score_gpt":0.27159187954014335,"score_spread":0.2453976788481452,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4224287786","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.009977757,0.0004057773,0.34806082,0.00053401274,0.0003635365,0.0006385993,0.04409027,0.584152,0.011777204],"genre_scores_gemma":[0.060646426,0.00064408686,0.71436614,0.0010669669,0.00017122798,0.0022029122,0.07346969,0.13294648,0.0144860335],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99820125,0.00025442598,0.00018372477,0.00046623172,0.0007148818,0.00017937862],"domain_scores_gemma":[0.99830174,0.00068561896,0.00025071128,0.00022100518,0.00036454154,0.00017633585],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0024602308,0.0027357044,0.0015311328,0.0016598287,0.0015160138,0.002298931,0.0045674164,0.00095780066,0.055184416],"category_scores_gemma":[0.007158084,0.0016028387,0.0020043917,0.0015199193,0.0011985229,0.0027031912,0.003753056,0.0025833577,0.032135636],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014340507,0.00023096007,0.007266421,0.004075817,0.0003983609,0.00091618986,0.0010634979,0.014617698,0.036044933,0.016492315,0.725526,0.1919338],"study_design_scores_gemma":[0.0005685921,0.00026408536,0.007752106,0.0005458625,0.00019872833,0.0022920459,0.00022518793,0.2156336,0.080194235,0.048683915,0.6430254,0.00061628455],"about_ca_topic_score_codex":0.0034767718,"about_ca_topic_score_gemma":0.0039453544,"teacher_disagreement_score":0.055184416,"about_ca_system_score_codex":0.000883635,"about_ca_system_score_gemma":0.0026332203,"threshold_uncertainty_score":0.1846103},"labels":[],"label_agreement":null},{"id":"W4224289048","doi":"10.1093/bioinformatics/btac286","title":"Tightly integrated multiomics-based deep tensor survival model for time-to-event prediction","year":2022,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Ferroptosis and cancer prognosis","field":"Medicine","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Manitoba; Princess Margaret Cancer Centre; CancerCare Manitoba; Public Health Ontario; University of Toronto","funders":"","keywords":"Computer science; Data mining; Concordance; Machine learning; Artificial intelligence; Bioinformatics; Biology","score_opus":0.025212904911393164,"score_gpt":0.26451984591089206,"score_spread":0.2393069409994989,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4224289048","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.07901101,0.00074980553,0.9139016,0.0014103828,0.00011163469,0.00008294973,0.0019162463,0.0014530579,0.0013633146],"genre_scores_gemma":[0.87132007,0.0006305617,0.11875708,0.00034582094,0.00013979374,0.00026056304,0.0032836122,0.00014826005,0.0051142476],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99951994,0.0001359006,0.000029625795,0.00014676458,0.0000783841,0.000089323534],"domain_scores_gemma":[0.99900657,0.00040319495,0.0001561674,0.00009525424,0.00023327222,0.00010554076],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0022180353,0.0011925987,0.0010128162,0.0006917749,0.00040652836,0.0008175701,0.0016107736,0.0008392128,0.002804964],"category_scores_gemma":[0.004138588,0.00034992932,0.00095416204,0.0008510266,0.00070573407,0.0009915223,0.001201877,0.001740257,0.0005883788],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002947658,0.00009957159,0.012586285,0.0000966304,0.00013170553,0.00014321023,0.0000945379,0.9074368,0.0017578816,0.0145590985,0.004029676,0.05876991],"study_design_scores_gemma":[0.000004653436,0.000008513319,0.00024985656,0.0000033613987,0.000008260358,0.000013042685,0.0000030609797,0.9954347,0.0001633119,0.0039246655,0.00018145998,0.0000051063275],"about_ca_topic_score_codex":0.021090506,"about_ca_topic_score_gemma":0.017495617,"teacher_disagreement_score":0.021090506,"about_ca_system_score_codex":0.0014619277,"about_ca_system_score_gemma":0.0021929396,"threshold_uncertainty_score":0.041935503},"labels":[],"label_agreement":null},{"id":"W4224309355","doi":"10.1093/bioinformatics/btac276","title":"Kernel-based hierarchical structural component models for pathway analysis","year":2022,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"","keywords":"Kernel (algebra); Computer science; Kernel method; Computational biology; Software; Data mining; Pathway analysis; Component (thermodynamics); Machine learning; Artificial intelligence; Bioinformatics; Biology; Support vector machine; Mathematics; Genetics","score_opus":0.01322497955221346,"score_gpt":0.23123613584337055,"score_spread":0.21801115629115708,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4224309355","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.007154382,0.0005009664,0.98947096,0.0002701738,0.000039974228,0.00007938998,0.00077551027,0.0010804571,0.00062815734],"genre_scores_gemma":[0.51001996,0.0020318052,0.46997935,0.00045280808,0.00023462009,0.001114911,0.0057008443,0.0009012161,0.009564467],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9980191,0.0009333007,0.00009261132,0.00046341005,0.00030269916,0.00018901954],"domain_scores_gemma":[0.9943777,0.0034734535,0.00053177576,0.00047326306,0.0009525567,0.00019129254],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00353634,0.0017511677,0.001672427,0.0023162332,0.0006471025,0.0015290573,0.0028008365,0.0013011557,0.006247585],"category_scores_gemma":[0.0116425315,0.0007271216,0.0028877696,0.002893963,0.0012525223,0.001983975,0.0018853211,0.0026368105,0.0021435807],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001948051,0.000068165034,0.0025886646,0.00023051421,0.00027167046,0.000089209374,0.000116720155,0.89093083,0.00091713306,0.047269233,0.0037027122,0.0536204],"study_design_scores_gemma":[0.000006895173,0.000009463483,0.00021830562,0.000007800887,0.000016197317,0.000011508313,0.0000063291714,0.9818912,0.00011858649,0.017088741,0.00061562186,0.000009343593],"about_ca_topic_score_codex":0.019566474,"about_ca_topic_score_gemma":0.013809196,"teacher_disagreement_score":0.019566474,"about_ca_system_score_codex":0.0019290486,"about_ca_system_score_gemma":0.0027793054,"threshold_uncertainty_score":0.038905203},"labels":[],"label_agreement":null},{"id":"W4225351187","doi":"10.1093/bioinformatics/btac236","title":"Characterizing domain-specific open educational resources by linking ISCB Communities of Special Interest to Wikipedia","year":2022,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Wikis in Education and Collaboration","field":"Social Sciences","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"BC Cancer Agency","funders":"","keywords":"Computer science; Resource (disambiguation); Representation (politics); Data science; Domain (mathematical analysis); Systems biology; Computational biology; Biology; Mathematics","score_opus":0.05021417730920626,"score_gpt":0.3220012158937981,"score_spread":0.27178703858459186,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4225351187","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.80240715,0.0017196989,0.060071595,0.00077877607,0.00016997781,0.0012772955,0.09256754,0.0020742458,0.038933717],"genre_scores_gemma":[0.74055743,0.0009434905,0.12981872,0.00012427242,0.00010905219,0.0011292403,0.12051948,0.00044181937,0.0063565252],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9975447,0.0005896966,0.00036774843,0.00047558313,0.0007821603,0.00024016773],"domain_scores_gemma":[0.976767,0.009453701,0.0042644325,0.0015038383,0.0063825296,0.0016284694],"candidate_categories":["scholarly_communication","open_science"],"consensus_categories":[],"category_scores_codex":[0.0014843782,0.00047616102,0.00032834942,0.02338621,0.0010876014,0.0025639592,0.00058153324,0.00045249998,0.003857767],"category_scores_gemma":[0.018808482,0.00017614619,0.00034111578,0.020767987,0.000623007,0.0027351934,0.0025398242,0.00046485843,0.001100396],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005435009,0.00057707465,0.4714551,0.0068615214,0.00038979,0.0017448987,0.012883942,0.008323851,0.018060219,0.041117176,0.07339446,0.36464846],"study_design_scores_gemma":[0.00009688551,0.00026715916,0.51602304,0.0018444868,0.00038855898,0.0020679398,0.027104821,0.054972418,0.023815574,0.034641203,0.338546,0.00023193622],"about_ca_topic_score_codex":0.008252528,"about_ca_topic_score_gemma":0.013349369,"teacher_disagreement_score":0.99941844,"about_ca_system_score_codex":0.00089389953,"about_ca_system_score_gemma":0.0016470781,"threshold_uncertainty_score":0.01640898},"labels":[],"label_agreement":null},{"id":"W4225604045","doi":"10.1093/bioinformatics/btac283","title":"Continuous chromatin state feature annotation of the human epigenome","year":2022,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"Natural Sciences and Engineering Research Council of Canada; Genome British Columbia; Michael Smith Health Research BC; Compute Canada; Genome Canada","keywords":"Epigenome; Chromatin; Epigenomics; Computer science; Annotation; Computational biology; Genome; Biology; Genetics; Artificial intelligence; Gene; DNA methylation; Gene expression","score_opus":0.004797717254827892,"score_gpt":0.20735783793867735,"score_spread":0.20256012068384946,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4225604045","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.055231657,0.0009880834,0.89997053,0.00046165523,0.000075235934,0.00011912724,0.02294576,0.016951201,0.0032568232],"genre_scores_gemma":[0.42454284,0.0005367431,0.5242663,0.0002436119,0.00012050541,0.00031719755,0.044791106,0.0014603178,0.0037214316],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99954283,0.00009662368,0.000025333688,0.00020768263,0.00009885762,0.000028579834],"domain_scores_gemma":[0.9987966,0.00057339674,0.00011746048,0.00024650717,0.0002142298,0.000051725434],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00061385264,0.0006082446,0.0004699366,0.0012921059,0.000396352,0.0008729263,0.0008328485,0.0007995562,0.0049481927],"category_scores_gemma":[0.003280054,0.00033186437,0.0007554704,0.0014778803,0.00036526623,0.0011353835,0.00091064494,0.0007497758,0.0015841253],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0016181294,0.0001850781,0.041486736,0.0019305545,0.00020558889,0.00064781395,0.0012662946,0.19889106,0.10760147,0.041805252,0.05091925,0.5534428],"study_design_scores_gemma":[0.0000648161,0.00009639149,0.018384293,0.00012808778,0.00006974513,0.00052339694,0.00018232118,0.81175053,0.056891188,0.064742275,0.047077287,0.000089653135],"about_ca_topic_score_codex":0.0036031764,"about_ca_topic_score_gemma":0.0043356353,"teacher_disagreement_score":0.0049481927,"about_ca_system_score_codex":0.0005228549,"about_ca_system_score_gemma":0.000516415,"threshold_uncertainty_score":0.016553402},"labels":[],"label_agreement":null},{"id":"W4225839207","doi":"10.1093/bioinformatics/btac259","title":"PhyloPGM: boosting regulatory function prediction accuracy using evolutionary information","year":2022,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"","keywords":"Computational biology; Computer science; Boosting (machine learning); Context (archaeology); Probabilistic logic; Sequence (biology); Function (biology); Data mining; Extant taxon; Genomics; DNA binding site; Gene regulatory network; Machine learning; Biology; Genome; Gene; Artificial intelligence; Genetics; Gene expression; Evolutionary biology; Promoter","score_opus":0.008650630458668988,"score_gpt":0.20584923273597067,"score_spread":0.19719860227730168,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4225839207","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.22198337,0.0061507206,0.5890898,0.0024837994,0.0004307035,0.00038099344,0.014613691,0.15817651,0.0066903746],"genre_scores_gemma":[0.46713892,0.0015297114,0.48777,0.00121969,0.00025308004,0.00068195164,0.031565003,0.007044186,0.002797527],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9986922,0.00041951254,0.000045611785,0.00046202858,0.0002641466,0.00011653488],"domain_scores_gemma":[0.99777967,0.0012893775,0.00016696504,0.00038928492,0.00025356485,0.000121107725],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0035996311,0.002706569,0.0018157414,0.0024092968,0.0012184045,0.0011339956,0.0030977747,0.0020432984,0.0053206244],"category_scores_gemma":[0.009357446,0.00092906447,0.0015557826,0.0020839176,0.0010721469,0.0021417977,0.002805978,0.0031121238,0.0031358493],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0026792255,0.00061424746,0.025978968,0.0020045717,0.0009423129,0.0004906764,0.00039181064,0.50653446,0.030533763,0.0073692165,0.09106396,0.33139676],"study_design_scores_gemma":[0.00014171527,0.00013249321,0.0015696738,0.000055602603,0.00009348691,0.00013319077,0.00003696395,0.9774975,0.0068890145,0.0069226255,0.0064909183,0.00003681257],"about_ca_topic_score_codex":0.004378402,"about_ca_topic_score_gemma":0.004294557,"teacher_disagreement_score":0.0053206244,"about_ca_system_score_codex":0.00088537374,"about_ca_system_score_gemma":0.0017659949,"threshold_uncertainty_score":0.01903689},"labels":[],"label_agreement":null},{"id":"W4226094768","doi":"10.1093/bioinformatics/btac266","title":"A graph neural network approach for molecule carcinogenicity prediction","year":2022,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Computational Drug Discovery Methods","field":"Computer Science","cited_by":34,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Vector Institute; University of Toronto","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Computer science; Machine learning; Artificial intelligence; Data mining; Source code; Graph; Theoretical computer science; Programming language","score_opus":0.0286934818020863,"score_gpt":0.25977673961487296,"score_spread":0.23108325781278666,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4226094768","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.055066172,0.0023007158,0.92731947,0.0015008827,0.00017613848,0.00012845462,0.0024348907,0.0032331406,0.007840178],"genre_scores_gemma":[0.7665316,0.0017995693,0.20934527,0.00091167,0.00020586092,0.00031326496,0.0065548904,0.00032057884,0.01401732],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99982244,0.00003431839,0.000007187737,0.000065739936,0.000047643134,0.000022750923],"domain_scores_gemma":[0.99947137,0.0002753382,0.000053484266,0.00004833597,0.00011337303,0.000038013146],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00048773878,0.0008127239,0.00067256944,0.0011113283,0.00027962093,0.0005935376,0.0015104522,0.0010168959,0.0038475865],"category_scores_gemma":[0.0017444745,0.00037488423,0.0006718116,0.00089474773,0.0004914349,0.0010854008,0.000651276,0.0015008678,0.00080972],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00009677017,0.00009270386,0.001619748,0.00011824584,0.00007687905,0.0000649774,0.000014970493,0.9060296,0.0019456986,0.007443217,0.004979027,0.077518165],"study_design_scores_gemma":[0.0000031299674,0.000008712207,0.00007728941,0.000004090025,0.000005872681,0.0000058730393,0.0000012104789,0.994156,0.00025741328,0.0050793174,0.00039883322,0.0000021333046],"about_ca_topic_score_codex":0.008663474,"about_ca_topic_score_gemma":0.0129016405,"teacher_disagreement_score":0.008663474,"about_ca_system_score_codex":0.0010993079,"about_ca_system_score_gemma":0.000978426,"threshold_uncertainty_score":0.0172261},"labels":[],"label_agreement":null},{"id":"W4226119586","doi":"10.1093/bioinformatics/btac226","title":"Syotti: scalable bait design for DNA enrichment","year":2022,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":18,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Dalhousie University","funders":"National Human Genome Research Institute; National Institute of Allergy and Infectious Diseases; Natural Sciences and Engineering Research Council of Canada; Agencia Nacional de Investigación y Desarrollo; United States - Israel Binational Science Foundation; National Institutes of Health; National Science Foundation","keywords":"Metagenomics; Scalability; Computer science; Heuristic; Set (abstract data type); Computational biology; DNA; Process (computing); Protocol (science); Data mining; Algorithm; Biology; Database; Genetics; Artificial intelligence; Programming language; Gene","score_opus":0.02003392150478879,"score_gpt":0.23204389337318718,"score_spread":0.21200997186839837,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4226119586","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.01344553,0.0013083003,0.84655076,0.00047524742,0.00023222479,0.00066140963,0.007911769,0.12644686,0.002967805],"genre_scores_gemma":[0.05838007,0.00069860317,0.90799433,0.0008385853,0.000075406846,0.0014277977,0.021823822,0.0052175396,0.0035438584],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9980648,0.00030463398,0.00016101333,0.00062681054,0.00067469315,0.00016808859],"domain_scores_gemma":[0.9978422,0.00105081,0.00015031184,0.0005438162,0.00027003107,0.00014282526],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0019324144,0.0024273193,0.0023879122,0.0015350598,0.0009680945,0.0021925839,0.0033049062,0.0021904025,0.014742099],"category_scores_gemma":[0.0061356346,0.0016428902,0.0024856238,0.0015714017,0.0013274457,0.0030932748,0.0036427751,0.0028770703,0.012163368],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0026918557,0.0005275969,0.0048714723,0.0054637305,0.0005119351,0.0007156405,0.00068018696,0.046069268,0.4304122,0.020838546,0.1118868,0.3753309],"study_design_scores_gemma":[0.0007590253,0.0006954388,0.002649271,0.00027988607,0.000192921,0.0011000491,0.00020777874,0.5002762,0.33991528,0.0462978,0.10731107,0.00031522018],"about_ca_topic_score_codex":0.0016331014,"about_ca_topic_score_gemma":0.002828652,"teacher_disagreement_score":0.014742099,"about_ca_system_score_codex":0.0012921399,"about_ca_system_score_gemma":0.0018381543,"threshold_uncertainty_score":0.04931724},"labels":[],"label_agreement":null},{"id":"W4226329328","doi":"10.1093/bioinformatics/btac195","title":"ELIXIR biovalidator for semantic validation of life science metadata","year":2022,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Biomedical Text Mining and Ontologies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ontario Institute for Cancer Research","funders":"Vlaamse regering; Fonds Wetenschappelijk Onderzoek; European Bioinformatics Institute","keywords":"Validator; Computer science; License; Schema (genetic algorithms); Elixir (programming language); Information retrieval; Metadata; JSON; Semantic integration; World Wide Web; Programming language; Semantic Web","score_opus":0.03119283094040965,"score_gpt":0.28790835011622395,"score_spread":0.2567155191758143,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4226329328","genre_codex":"software","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.007914277,0.0010484258,0.40264416,0.0013940595,0.0005767332,0.0009628487,0.058294646,0.5065172,0.020647744],"genre_scores_gemma":[0.061313387,0.00096941664,0.49450088,0.0022127957,0.00020579435,0.0021967965,0.2972729,0.12799862,0.013329467],"study_design_codex":"not_applicable","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.98710907,0.0029283701,0.0019172933,0.0021690046,0.005325114,0.0005511837],"domain_scores_gemma":[0.9682663,0.011029374,0.0024609189,0.010402225,0.007197386,0.0006437761],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.024666661,0.0025089383,0.0014377608,0.006342515,0.0025802942,0.0074865627,0.004270434,0.0024215402,0.03149138],"category_scores_gemma":[0.05426424,0.0016668623,0.0034140702,0.0032895505,0.0017419046,0.008659867,0.011805419,0.0038000944,0.030898647],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0031235563,0.00038725793,0.013917451,0.005561309,0.0007651137,0.0012739964,0.0028460408,0.0066527063,0.02653859,0.08968955,0.59228015,0.2569643],"study_design_scores_gemma":[0.00030377918,0.00016647647,0.006507653,0.0023086017,0.000249515,0.0009942097,0.00052408804,0.046883408,0.07674799,0.04975083,0.8151497,0.00041384457],"about_ca_topic_score_codex":0.006904767,"about_ca_topic_score_gemma":0.0067571593,"teacher_disagreement_score":0.03149138,"about_ca_system_score_codex":0.0028787956,"about_ca_system_score_gemma":0.0070295907,"threshold_uncertainty_score":0.13045138},"labels":[],"label_agreement":null},{"id":"W4229030659","doi":"10.1093/bioinformatics/btac309","title":"Targeted metabolomics analyses for brain tumor margin assessment during surgery","year":2022,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Metabolomics and Mass Spectrometry Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"Bruker BioSpin","keywords":"Brain tumor; Receiver operating characteristic; Biomarker; Metabolomics; Margin (machine learning); Computer science; Glioma; Magnetic resonance imaging; Artificial intelligence; Medicine; Chemistry; Machine learning; Bioinformatics; Radiology; Biology; Cancer research; Pathology","score_opus":0.02779288951151632,"score_gpt":0.30232438225387476,"score_spread":0.27453149274235844,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4229030659","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.25857276,0.0018773504,0.6980145,0.0011857109,0.00017763111,0.00043356966,0.009482201,0.026739832,0.0035164603],"genre_scores_gemma":[0.63269514,0.0011744375,0.35644525,0.00030784056,0.00006474852,0.00030257064,0.005085909,0.0009657108,0.0029584444],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99981123,0.00003263305,0.000007659776,0.000063612606,0.00006955194,0.000015264437],"domain_scores_gemma":[0.9995339,0.0001998382,0.00009709984,0.000036238845,0.00010694284,0.000025978808],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005484324,0.0009885756,0.00048156639,0.0006456745,0.00026289446,0.00053771696,0.0005467403,0.0006964467,0.0040535294],"category_scores_gemma":[0.0017033314,0.00021721363,0.0004988765,0.0004204758,0.00022475385,0.00048491402,0.00047858892,0.00050866,0.0020973354],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0031210498,0.0007403015,0.032544315,0.0014782257,0.00039597283,0.0006714208,0.00016133781,0.15493655,0.34462956,0.0017096152,0.012015959,0.4475958],"study_design_scores_gemma":[0.00008577604,0.0005894202,0.018070921,0.00007331749,0.000120895784,0.00045187393,0.000048981172,0.756761,0.21528845,0.003582598,0.004839555,0.000087217515],"about_ca_topic_score_codex":0.0026033276,"about_ca_topic_score_gemma":0.0035437476,"teacher_disagreement_score":0.0040535294,"about_ca_system_score_codex":0.00033942406,"about_ca_system_score_gemma":0.00065718126,"threshold_uncertainty_score":0.013560414},"labels":[],"label_agreement":null},{"id":"W4232460051","doi":"10.4018/978-1-4666-3604-0.ch067","title":"The Impact of Genetic Testing and Genetic Information on Ethical, Legal and Social Issues in North America","year":2013,"lang":"en","type":"book-chapter","venue":"Bioinformatics","topic":"Reproductive Health and Technologies","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Lakehead University","funders":"","keywords":"Genetic testing; Genetic discrimination; Eugenics; Autonomy; Legislation; Political science; Terminology; Internet privacy; Law; Computer science; Medicine","score_opus":0.030871836761586547,"score_gpt":0.304688814603027,"score_spread":0.27381697784144043,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4232460051","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0058500823,0.16248941,0.0014019003,0.090437815,0.0029092461,0.000022949898,0.00003299385,0.00003354697,0.7368221],"genre_scores_gemma":[0.19118841,0.22300866,0.004389027,0.076272875,0.0032701918,0.00014643247,0.00007793482,0.00008779964,0.5015586],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99857605,0.00055580627,0.00003249751,0.00011876546,0.00054153684,0.00017536402],"domain_scores_gemma":[0.9991204,0.0006405364,0.000037438145,0.000023413162,0.00012800592,0.000050158058],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015116314,0.00042286288,0.00033486454,0.0009998846,0.005756954,0.0050247363,0.00056095107,0.002772041,0.0055925553],"category_scores_gemma":[0.0014512726,0.00027121123,0.00024876342,0.0017135027,0.009463642,0.0038849358,0.0023341752,0.0041264603,0.0007310765],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000052393166,0.000026177653,0.00020185673,0.00012551057,0.0000036975616,0.00031593372,0.012890053,0.00028709386,0.00021066726,0.7872235,0.13056539,0.06814478],"study_design_scores_gemma":[0.000002187186,0.000006857007,0.00071215496,0.0004690403,0.0000041839726,0.0002540354,0.0035566138,0.00012226205,0.000108404056,0.09164617,0.9031037,0.00001440487],"about_ca_topic_score_codex":0.06032649,"about_ca_topic_score_gemma":0.11351953,"teacher_disagreement_score":0.9396735,"about_ca_system_score_codex":0.013714998,"about_ca_system_score_gemma":0.009435612,"threshold_uncertainty_score":0.11995071},"labels":[],"label_agreement":null},{"id":"W4232856118","doi":"10.36255/exonpublications.bioinformatics.2021.cont","title":"List of Contributors","year":2021,"lang":"en","type":"book-chapter","venue":"Bioinformatics","topic":"Genetically Modified Organisms Research","field":"Agricultural and Biological Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Library science; Humanities; Engineering; Art","score_opus":0.03265229458678334,"score_gpt":0.2262356959710034,"score_spread":0.19358340138422006,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4232856118","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.002008031,0.048253115,0.009643665,0.13628975,0.14631338,0.0012628762,0.092331745,0.0072404784,0.55665696],"genre_scores_gemma":[0.009698081,0.041485813,0.0074285544,0.044027053,0.03367163,0.0020681515,0.06775183,0.004129754,0.78973913],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9975515,0.00047265846,0.00031294487,0.0004023246,0.0009854829,0.00027504217],"domain_scores_gemma":[0.9857091,0.002308782,0.0006012115,0.0011308658,0.0069599994,0.0032901599],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0023879441,0.0011875705,0.0014750271,0.004515956,0.0019652178,0.005112844,0.0023279262,0.002331973,0.6259446],"category_scores_gemma":[0.030851504,0.0004121411,0.00075859774,0.0034300555,0.0005814232,0.0033847578,0.0030273583,0.0026060804,0.5180536],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000015090863,0.000005829209,0.00020058289,0.000114265036,0.000003224809,0.00007303191,0.000027713706,0.000014755645,0.000029882387,0.000980504,0.9690099,0.029525287],"study_design_scores_gemma":[0.0000065212744,0.0000059517083,0.00022772266,0.000330805,0.0000049768755,0.00029304822,0.000055105098,0.000020377069,0.000036406516,0.0007786725,0.9982339,0.0000064160336],"about_ca_topic_score_codex":0.0023712486,"about_ca_topic_score_gemma":0.0030242472,"teacher_disagreement_score":0.6259446,"about_ca_system_score_codex":0.0020311254,"about_ca_system_score_gemma":0.006404883,"threshold_uncertainty_score":0.5335446},"labels":[],"label_agreement":null},{"id":"W4233566421","doi":"10.1201/b16589-19","title":"Combined Analysis of Three Genome-Wide Association Studies on vWF and FVIII Plasma Levels","year":2014,"lang":"en","type":"book-chapter","venue":"Bioinformatics","topic":"Blood Coagulation and Thrombosis Mechanisms","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Canadian Institutes of Health Research; Fondation pour la Recherche Médicale; Institut National de la Santé et de la Recherche Médicale; Canada Research Chairs; Heart and Stroke Foundation of Canada","keywords":"Genome-wide association study; Genetic association; Medicine; Computational biology; Internal medicine; Genetics; Biology; Single-nucleotide polymorphism; Genotype; Gene","score_opus":0.06007262892958733,"score_gpt":0.2769431802463115,"score_spread":0.21687055131672417,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4233566421","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.77429754,0.07983294,0.046789568,0.004397441,0.0013332813,0.00023654553,0.06274736,0.0013621424,0.029003097],"genre_scores_gemma":[0.9037637,0.014998574,0.03417789,0.0013731186,0.0006890052,0.0002836512,0.024556657,0.00032334533,0.019834053],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99892896,0.00034914154,0.000074703945,0.00037561622,0.00020957274,0.00006198285],"domain_scores_gemma":[0.99834394,0.0010084697,0.00011689956,0.0001835955,0.00021363134,0.00013350132],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0019346864,0.00080194446,0.0010877718,0.0015787197,0.00041622002,0.0013932714,0.0005943066,0.0007698473,0.008368871],"category_scores_gemma":[0.002227776,0.00022764054,0.0021605901,0.0035450025,0.00017254491,0.00032879517,0.0007752126,0.0006979915,0.001444797],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0071522626,0.00028150025,0.6253892,0.0021495272,0.027443768,0.0043350104,0.00057996815,0.001098676,0.041769627,0.0019097251,0.021106256,0.2667844],"study_design_scores_gemma":[0.00029944192,0.00045644902,0.9328524,0.00025321447,0.023665193,0.004192812,0.00018984613,0.0013449576,0.004557039,0.002299783,0.02979784,0.000091050344],"about_ca_topic_score_codex":0.003327962,"about_ca_topic_score_gemma":0.007618473,"teacher_disagreement_score":0.008368871,"about_ca_system_score_codex":0.0002533993,"about_ca_system_score_gemma":0.00047083446,"threshold_uncertainty_score":0.027996719},"labels":[],"label_agreement":null},{"id":"W4239428642","doi":"10.1093/bioinformatics/btr647","title":"Sensitive and fast mapping of di-base encoded reads","year":2011,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Chemical Synthesis and Analysis","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Computer science; Base (topology); Software; Computational biology; Biology; Programming language; Mathematics","score_opus":0.016896724885756904,"score_gpt":0.20783175419916358,"score_spread":0.19093502931340667,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4239428642","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.04046227,0.001169186,0.8887709,0.00055232336,0.00064354955,0.0005216342,0.01162341,0.048598837,0.0076578334],"genre_scores_gemma":[0.06907863,0.00028195418,0.90529567,0.000423623,0.00009707505,0.0005320407,0.015827425,0.0026955893,0.005767996],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99698204,0.0009305872,0.00022323382,0.0008448088,0.0008600486,0.00015934916],"domain_scores_gemma":[0.9952449,0.0018020909,0.00031104023,0.0016157198,0.00086308655,0.00016300817],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0040769414,0.0015734694,0.0010948386,0.0017591204,0.0008926041,0.0013541778,0.00190898,0.0010233441,0.012826913],"category_scores_gemma":[0.013590873,0.0013787006,0.0013418231,0.0015939722,0.00048497668,0.001581643,0.002523136,0.0023231162,0.010784705],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0016990964,0.00033965323,0.0068563605,0.0014832729,0.00061602046,0.0005409815,0.0004629427,0.01126341,0.24198298,0.012558378,0.1061079,0.61608905],"study_design_scores_gemma":[0.0003901819,0.00045863408,0.01277116,0.00018855551,0.00017303388,0.0022293786,0.0002533822,0.23899432,0.5594818,0.040369328,0.14437799,0.00031223363],"about_ca_topic_score_codex":0.0006524893,"about_ca_topic_score_gemma":0.0020473045,"teacher_disagreement_score":0.012826913,"about_ca_system_score_codex":0.00057278347,"about_ca_system_score_gemma":0.00088883884,"threshold_uncertainty_score":0.042910337},"labels":[],"label_agreement":null},{"id":"W4241029082","doi":"10.1093/bioinformatics/btr601","title":"Automated workflows for accurate mass-based putative metabolite identification in LC/MS-derived metabolomic datasets","year":2011,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Metabolomics and Mass Spectrometry Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Discovery Centre; University of Alberta","funders":"Biotechnology and Biological Sciences Research Council; Directorate for Biological Sciences; National Institute for Health and Care Research; Cancer Research UK; Wellcome Trust","keywords":"Metabolomics; Workflow; Identification (biology); Metabolite; Computer science; Computational biology; Chromatography; Chemistry; Data mining; Biology; Biochemistry; Database","score_opus":0.030480199805048347,"score_gpt":0.28736586061543684,"score_spread":0.2568856608103885,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4241029082","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.017717773,0.00021980985,0.8164541,0.0003109023,0.00007722,0.0016720028,0.011839569,0.1507196,0.0009889896],"genre_scores_gemma":[0.03933207,0.00017964188,0.9326546,0.00021748488,0.000046182908,0.0024114398,0.020356579,0.0037233436,0.0010786575],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99398065,0.0011851195,0.0012818603,0.0014373368,0.0017394506,0.00037557856],"domain_scores_gemma":[0.9862037,0.0050995946,0.0018458845,0.0032840925,0.0029837908,0.0005830007],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.013067587,0.0022098643,0.0015967906,0.00404945,0.00150632,0.004331944,0.002772124,0.001079322,0.006885528],"category_scores_gemma":[0.019002581,0.0010958007,0.0017780978,0.0028190452,0.0010737476,0.0023261993,0.0026690757,0.0017088782,0.006476848],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0070419963,0.001240475,0.01871466,0.002660448,0.0006112801,0.0015966386,0.001667524,0.01786762,0.25465795,0.010305803,0.07014169,0.61349386],"study_design_scores_gemma":[0.0016266046,0.00069776986,0.030107308,0.00048572695,0.00022601847,0.0013420166,0.0005115164,0.27802876,0.58108944,0.026011478,0.07901656,0.0008568339],"about_ca_topic_score_codex":0.0036435556,"about_ca_topic_score_gemma":0.0033922056,"teacher_disagreement_score":0.013067587,"about_ca_system_score_codex":0.0018176455,"about_ca_system_score_gemma":0.0046536895,"threshold_uncertainty_score":0.069108844},"labels":[],"label_agreement":null},{"id":"W4243450674","doi":"10.1093/bioinformatics/btab790","title":"Theory of local k-mer selection with applications to long-read alignment","year":2021,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":46,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"The Scarborough Hospital; University of Toronto","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Selection (genetic algorithm); Computer science; Algorithm; Artificial intelligence","score_opus":0.009314820077001981,"score_gpt":0.22851076394630696,"score_spread":0.21919594386930497,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4243450674","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0039834846,0.0008883373,0.98944914,0.00040778084,0.000091370064,0.000050343937,0.000091333735,0.00035082857,0.0046874676],"genre_scores_gemma":[0.26054636,0.0038152349,0.71724385,0.001334388,0.0010024513,0.001332886,0.00074704655,0.0018827572,0.012095031],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9969982,0.0012243787,0.00017295603,0.00041850613,0.001005514,0.00018051443],"domain_scores_gemma":[0.98084015,0.015254576,0.00093604904,0.0011241134,0.0014900238,0.00035506074],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004457685,0.0011540288,0.0016315806,0.0022598107,0.0014890614,0.002382686,0.003330448,0.0022383237,0.010011519],"category_scores_gemma":[0.024088966,0.0009958512,0.001773486,0.0029783626,0.0029253073,0.005165155,0.0030202873,0.0038922653,0.0025686107],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000055893728,0.00006303051,0.0011347943,0.00061242026,0.000052703454,0.00021774236,0.00026020553,0.2725487,0.0020107299,0.68274546,0.0059470395,0.034351308],"study_design_scores_gemma":[0.000011351052,0.000024425688,0.00017029705,0.000056831355,0.000009518955,0.00008067434,0.000025132329,0.6924789,0.00065458583,0.30275986,0.0037078904,0.000020554138],"about_ca_topic_score_codex":0.002237194,"about_ca_topic_score_gemma":0.0019977563,"teacher_disagreement_score":0.010011519,"about_ca_system_score_codex":0.002352322,"about_ca_system_score_gemma":0.0019396759,"threshold_uncertainty_score":0.03349191},"labels":[],"label_agreement":null},{"id":"W4244443336","doi":"10.1093/bioinformatics/btt670","title":"ISCB/SPRINGER series in computational biology","year":2013,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Computational Physics and Python Applications","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Series (stratigraphy); Computer science; Computational biology; Biology","score_opus":0.01424726321169884,"score_gpt":0.24695382801820068,"score_spread":0.23270656480650184,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4244443336","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.001203286,0.12155012,0.07967751,0.013789559,0.04106798,0.0005228424,0.023217825,0.014882165,0.7040887],"genre_scores_gemma":[0.009473898,0.1243245,0.07360521,0.006746522,0.011908848,0.0011737419,0.041301493,0.013494263,0.71797156],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99652183,0.00040556752,0.0003456345,0.0007359145,0.0018212505,0.00016978795],"domain_scores_gemma":[0.9942624,0.0016120347,0.00039287715,0.0013888163,0.00169306,0.00065077114],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0024904844,0.002334557,0.0027377633,0.0052613737,0.0010711659,0.012539079,0.0026351484,0.00307754,0.32657772],"category_scores_gemma":[0.011817001,0.0012662507,0.0016886089,0.010761603,0.0011785907,0.007703028,0.003880388,0.0065317173,0.3446293],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000029208211,0.00003261663,0.00014102155,0.00076688366,0.000043266056,0.000058644073,0.00005466542,0.000891398,0.0005132173,0.020982644,0.77667534,0.19981113],"study_design_scores_gemma":[0.0000082257075,0.000008192309,0.0001966749,0.00035419053,0.000009321969,0.00012294391,0.00003109582,0.0006682187,0.00015344161,0.011858576,0.98657167,0.000017460134],"about_ca_topic_score_codex":0.0019280629,"about_ca_topic_score_gemma":0.001650307,"teacher_disagreement_score":0.32657772,"about_ca_system_score_codex":0.0016325884,"about_ca_system_score_gemma":0.003895069,"threshold_uncertainty_score":0.96055514},"labels":[],"label_agreement":null},{"id":"W4246220398","doi":"10.1093/bioinformatics/btt046","title":"A system for exact and approximate genetic linkage analysis of SNP data in large pedigrees","year":2013,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of Ottawa","funders":"","keywords":"Pedigree chart; Linkage (software); SNP; Computer science; Genetics; Genetic linkage; Genetic data; Computational biology; Biology; Data mining; Single-nucleotide polymorphism; Genotype; Population; Medicine; Gene","score_opus":0.02390971813834013,"score_gpt":0.27855223092193776,"score_spread":0.2546425127835976,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4246220398","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0009147746,0.0003372249,0.93701696,0.0002308483,0.00013920957,0.00018204795,0.0033131905,0.056722634,0.0011430207],"genre_scores_gemma":[0.019846292,0.00067941856,0.9562284,0.00026797247,0.00018344345,0.0014296724,0.011931857,0.0054653347,0.003967746],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9928226,0.002719493,0.0011482204,0.0011645206,0.0019775538,0.00016759096],"domain_scores_gemma":[0.98537076,0.007989544,0.00067057007,0.003928942,0.0016723963,0.00036781546],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.011954692,0.0018519299,0.0029731197,0.007563434,0.001859332,0.005530133,0.0042659263,0.0021724787,0.03906915],"category_scores_gemma":[0.053168867,0.0025296635,0.0026404057,0.009503835,0.0009728951,0.0035828585,0.005971059,0.0031340672,0.031242033],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0015186765,0.00019823133,0.0071885614,0.001319266,0.0012221503,0.000987834,0.0004993344,0.04745421,0.0043975217,0.042024147,0.21360236,0.6795876],"study_design_scores_gemma":[0.0009760124,0.00017033615,0.0038131464,0.00032007886,0.0003398946,0.0015918057,0.00013191914,0.6461904,0.0057496442,0.16730164,0.17314833,0.00026681303],"about_ca_topic_score_codex":0.005371437,"about_ca_topic_score_gemma":0.007794335,"teacher_disagreement_score":0.03906915,"about_ca_system_score_codex":0.0011688025,"about_ca_system_score_gemma":0.00442352,"threshold_uncertainty_score":0.13069934},"labels":[],"label_agreement":null},{"id":"W4250891669","doi":"10.1093/bioinformatics/btn204","title":"<i>Editorial</i>","year":2008,"lang":"he","type":"article","venue":"Bioinformatics","topic":"Earthquake and Disaster Impact Studies","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Computer science","score_opus":0.03853453417813308,"score_gpt":0.286133681234586,"score_spread":0.24759914705645292,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4250891669","genre_codex":"editorial","genre_gemma":"editorial","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"editorial","genre_consensus":"editorial","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000051807157,0.0022595788,0.00020500175,0.008987812,0.9820694,0.00008446562,0.00020279009,0.00023983068,0.0058993567],"genre_scores_gemma":[0.00050324,0.0038679817,0.00025249316,0.010908709,0.9543231,0.000098246936,0.0005399744,0.00023719762,0.029269058],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.992777,0.00071032275,0.00083298807,0.0008489674,0.0044233506,0.00040739882],"domain_scores_gemma":[0.95253325,0.005533344,0.0029456478,0.0023380748,0.0300292,0.006620462],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0066681164,0.003096382,0.0031751222,0.0055459943,0.0023040595,0.0104910415,0.0040744985,0.0060459976,0.12299614],"category_scores_gemma":[0.030734811,0.0009294642,0.0027129464,0.0024761118,0.0018913948,0.0043696226,0.0022174348,0.006906033,0.13485391],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00002213874,0.0000067395054,0.000016825314,0.00022128662,0.0000061958053,0.00004456311,0.000006266239,0.000008691626,0.00010157399,0.000071895185,0.992361,0.007132893],"study_design_scores_gemma":[0.000027304526,0.00004398068,0.0002151146,0.00030765764,0.000015454447,0.00020657049,0.000035135083,0.000049448903,0.00020341677,0.00030228935,0.9985783,0.000015366299],"about_ca_topic_score_codex":0.00064081425,"about_ca_topic_score_gemma":0.0010545031,"teacher_disagreement_score":0.87700385,"about_ca_system_score_codex":0.0024060523,"about_ca_system_score_gemma":0.0029606929,"threshold_uncertainty_score":0.41146308},"labels":[],"label_agreement":null},{"id":"W4280528542","doi":"10.1093/bioinformatics/btac304","title":"deepSimDEF: deep neural embeddings of gene products and gene ontology terms for functional analysis of genes","year":2022,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal; Mila - Quebec Artificial Intelligence Institute; University of Ottawa; Dalhousie University; Montreal Heart Institute","funders":"Fonds de Recherche du Québec - Santé; Institut de Valorisation des Données; Natural Sciences and Engineering Research Council of Canada; Canada Research Chairs","keywords":"Gene ontology; Gene; Computational biology; Computer science; Genetics; Biology; Artificial intelligence; Gene expression","score_opus":0.01238335273723722,"score_gpt":0.22635082858919084,"score_spread":0.2139674758519536,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4280528542","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.15009074,0.0024916343,0.78923064,0.0014197589,0.00029139954,0.00015855193,0.02168735,0.028701577,0.0059284233],"genre_scores_gemma":[0.6488056,0.0010115389,0.29416072,0.0007975091,0.00009163168,0.00028041564,0.047226317,0.00090727396,0.0067188847],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99974924,0.0000360571,0.0000132316645,0.00010230105,0.00007079007,0.000028373752],"domain_scores_gemma":[0.99961025,0.00013534709,0.00004664757,0.00008389127,0.000087954024,0.000036021425],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00050653407,0.0011224855,0.00046767318,0.0008159022,0.00025536894,0.00067601935,0.0015899613,0.00078966847,0.0037290705],"category_scores_gemma":[0.0016707019,0.0003007895,0.0006856857,0.000801995,0.0004997339,0.0011716947,0.0010049317,0.0015764604,0.0013117103],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007791524,0.00048912433,0.014200495,0.00089222775,0.00031369732,0.00030019286,0.00016598807,0.34310147,0.053904977,0.025394503,0.06470017,0.49575806],"study_design_scores_gemma":[0.000034511748,0.0000762968,0.0010486094,0.000027662398,0.000019190646,0.00007676339,0.000018543516,0.9662497,0.011057339,0.016432995,0.004941796,0.000016569818],"about_ca_topic_score_codex":0.005350661,"about_ca_topic_score_gemma":0.008200385,"teacher_disagreement_score":0.005350661,"about_ca_system_score_codex":0.00091866957,"about_ca_system_score_gemma":0.0008521086,"threshold_uncertainty_score":0.012475014},"labels":[],"label_agreement":null},{"id":"W4280553929","doi":"10.1093/bioinformatics/btac327","title":"Assessing and assuring interoperability of a genomics file format","year":2022,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Vector Institute; University of Toronto; Princess Margaret Cancer Centre; University Health Network","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Computer science; Software; Suite; Interoperability; Test suite; File format; Fuzz testing; Software engineering; Correctness; Test case; Database; Programming language; Operating system; Machine learning","score_opus":0.017293559913787802,"score_gpt":0.24105137407473362,"score_spread":0.2237578141609458,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4280553929","genre_codex":"empirical","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.5729061,0.0008175409,0.3624023,0.0011653785,0.00019807552,0.0012105228,0.008121166,0.047175158,0.006003806],"genre_scores_gemma":[0.6984926,0.0002446608,0.2799014,0.00044671868,0.00006203261,0.0006550416,0.013324359,0.0060024327,0.0008707938],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.97061205,0.0066949134,0.004820011,0.004689508,0.011945934,0.0012375715],"domain_scores_gemma":[0.8130282,0.098334104,0.019008614,0.03031795,0.03649567,0.002815364],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.04170804,0.0014925053,0.00077181746,0.0040708375,0.001630319,0.0041703396,0.0029854877,0.0017912886,0.0031796426],"category_scores_gemma":[0.1789893,0.0009455204,0.0012671201,0.0023581896,0.002461728,0.0060271947,0.0048743812,0.0018503908,0.0020439462],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.004112156,0.0012925345,0.44139645,0.002969343,0.0006558807,0.0013891505,0.005897249,0.039530326,0.19026402,0.015564122,0.021888247,0.2750405],"study_design_scores_gemma":[0.00021996097,0.0017210742,0.12769176,0.0007825449,0.00048379673,0.0015341574,0.001134407,0.15961964,0.6586074,0.016991377,0.030716717,0.0004970801],"about_ca_topic_score_codex":0.0037011695,"about_ca_topic_score_gemma":0.0023836168,"teacher_disagreement_score":0.04170804,"about_ca_system_score_codex":0.0016481552,"about_ca_system_score_gemma":0.003016674,"threshold_uncertainty_score":0.22057581},"labels":[],"label_agreement":null},{"id":"W4281636170","doi":"10.1093/bioinformatics/btac345","title":"Deep structure integrative representation of multi-omics data for cancer subtyping","year":2022,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":18,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"National Natural Science Foundation of China","keywords":"Subtyping; Computer science; Subspace topology; Omics; Representation (politics); External Data Representation; Cluster analysis; Data mining; Machine learning; Nonlinear dimensionality reduction; Artificial intelligence; Data science; Bioinformatics; Biology; Dimensionality reduction","score_opus":0.029297918903950445,"score_gpt":0.30043188300237283,"score_spread":0.2711339640984224,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4281636170","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.08844022,0.0021780864,0.8933716,0.0013198232,0.000102356986,0.00014723769,0.0060270983,0.005689939,0.0027237092],"genre_scores_gemma":[0.6088511,0.0015043691,0.3610024,0.00062502426,0.00010783041,0.0003452897,0.024075009,0.00033094874,0.003158123],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99943954,0.00013204507,0.000029010656,0.00015984572,0.00016942897,0.000070195725],"domain_scores_gemma":[0.9993326,0.00018280194,0.00008933575,0.00018164785,0.00015022789,0.00006338975],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011755185,0.0012001873,0.00087015016,0.0013040892,0.0003983712,0.0010711439,0.0013728825,0.0008570077,0.002276385],"category_scores_gemma":[0.0026605295,0.00026697555,0.0013355546,0.0016266899,0.000494765,0.001584237,0.0019234795,0.0016003837,0.00089519285],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00057381904,0.00046850165,0.015471441,0.0006608473,0.00059828593,0.00027499913,0.00028389462,0.37677923,0.04136333,0.022931388,0.028378176,0.51221603],"study_design_scores_gemma":[0.000017969323,0.000059009282,0.0018295685,0.000028792087,0.000057911664,0.00007403773,0.000045872006,0.9683769,0.007597763,0.018337023,0.0035544247,0.000020743008],"about_ca_topic_score_codex":0.0065689213,"about_ca_topic_score_gemma":0.0079164505,"teacher_disagreement_score":0.0065689213,"about_ca_system_score_codex":0.0010760644,"about_ca_system_score_gemma":0.0016212551,"threshold_uncertainty_score":0.013061345},"labels":[],"label_agreement":null},{"id":"W4281909951","doi":"10.1093/bioinformatics/btac355","title":"MAFFIN: metabolomics sample normalization using maximal density fold change with high-quality metabolic features and corrected signal intensities","year":2022,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Metabolomics and Mass Spectrometry Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":24,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"Natural Sciences and Engineering Research Council of Canada; Canada Foundation for Innovation","keywords":"Normalization (sociology); Computer science; Fold (higher-order function); Statistics; Mathematics","score_opus":0.025487399599166252,"score_gpt":0.24640956952106405,"score_spread":0.22092216992189778,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4281909951","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.012493656,0.001395235,0.9132091,0.0003183533,0.0003042715,0.00047033906,0.0077658775,0.06158493,0.0024582297],"genre_scores_gemma":[0.055687614,0.0005984493,0.9090949,0.000265461,0.00010060392,0.0024271328,0.01329353,0.015728857,0.0028033555],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9944759,0.0016718704,0.0004095331,0.0018866507,0.0013496445,0.00020638644],"domain_scores_gemma":[0.9924493,0.0038882606,0.0010063369,0.0012433004,0.0012781644,0.00013460401],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.011576375,0.003645661,0.0022844817,0.0037503345,0.0015385939,0.0030624485,0.0039428067,0.0017175032,0.012460516],"category_scores_gemma":[0.0343176,0.0016430238,0.002514435,0.0032121758,0.0013266015,0.002573644,0.0024395927,0.003041361,0.008532084],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0018203488,0.00036479402,0.037334956,0.0042000376,0.002358937,0.0006896658,0.0013933426,0.045902703,0.103644535,0.016922483,0.1458641,0.6395041],"study_design_scores_gemma":[0.00034917545,0.00071532716,0.055625238,0.0006975771,0.0006460788,0.0016575183,0.00027615335,0.4805678,0.19430885,0.031982046,0.23208584,0.0010884493],"about_ca_topic_score_codex":0.004564764,"about_ca_topic_score_gemma":0.005389729,"teacher_disagreement_score":0.012460516,"about_ca_system_score_codex":0.001698517,"about_ca_system_score_gemma":0.0021092794,"threshold_uncertainty_score":0.061222434},"labels":[],"label_agreement":null},{"id":"W4281964006","doi":"10.1093/bioinformatics/btac383","title":"Looking at the BiG picture: incorporating bipartite graphs in drug response prediction","year":2022,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Computational Drug Discovery Methods","field":"Computer Science","cited_by":30,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; Mila - Quebec Artificial Intelligence Institute","funders":"Natural Sciences and Engineering Research Council of Canada; Institut de Valorisation des Données; Merck Canada","keywords":"Bipartite graph; Drug response; Computer science; Leverage (statistics); Pharmacogenomics; Python (programming language); Drug; Graph; Big data; Machine learning; Artificial intelligence; Computational biology; Theoretical computer science; Data mining; Bioinformatics; Biology","score_opus":0.018610616020080822,"score_gpt":0.2560658220336284,"score_spread":0.23745520601354755,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4281964006","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.11178188,0.0032666347,0.8471019,0.0073434263,0.0002039152,0.00020444738,0.012853911,0.00897859,0.008265323],"genre_scores_gemma":[0.7735682,0.00196405,0.20335443,0.0023061498,0.00016950314,0.00027875035,0.013273987,0.0006100626,0.004474772],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99945027,0.00023597664,0.00002185865,0.0001448975,0.00009554717,0.00005150118],"domain_scores_gemma":[0.99770963,0.001608093,0.00018942264,0.00016733748,0.00016574115,0.00015966418],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013688698,0.00091157085,0.00079190684,0.001348463,0.00039565892,0.000954488,0.0014309522,0.0012497754,0.004013818],"category_scores_gemma":[0.0062539447,0.0005386832,0.0009514531,0.0012630858,0.0006619993,0.001699672,0.0012263971,0.0016167731,0.0009309159],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00038544586,0.00024158685,0.008075159,0.00049379026,0.00021391267,0.00027785214,0.000058428628,0.8499793,0.0034837879,0.015520984,0.018270096,0.102999724],"study_design_scores_gemma":[0.000018790384,0.000025191068,0.0006164936,0.000022308453,0.000028564391,0.000039201987,0.000007806701,0.9687885,0.00095600606,0.027552368,0.0019329763,0.000011760831],"about_ca_topic_score_codex":0.010577887,"about_ca_topic_score_gemma":0.016748393,"teacher_disagreement_score":0.010577887,"about_ca_system_score_codex":0.0012897422,"about_ca_system_score_gemma":0.0016284442,"threshold_uncertainty_score":0.021032631},"labels":[],"label_agreement":null},{"id":"W4282834074","doi":"10.1093/bioinformatics/btac391","title":"SPEAR: Systematic ProtEin AnnotatoR","year":2022,"lang":"en","type":"article","venue":"Bioinformatics","topic":"vaccines and immunoinformatics approaches","field":"Biochemistry, Genetics and Molecular Biology","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"Réseau Québécois de Recherche sur les Médicaments","keywords":"Spear; Python (programming language); Computer science; Annotation; Documentation; Visualization; Computational biology; Data mining; Biology; Programming language; Artificial intelligence; Archaeology","score_opus":0.011090178579655609,"score_gpt":0.21049638583135297,"score_spread":0.19940620725169736,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4282834074","genre_codex":"software","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0012496775,0.0010609147,0.1845962,0.0008783967,0.000871079,0.0006306547,0.21879031,0.5853475,0.0065753344],"genre_scores_gemma":[0.008715995,0.0012705388,0.25835294,0.0014802587,0.0002428733,0.0021027564,0.5681527,0.15213837,0.007543543],"study_design_codex":"not_applicable","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9937125,0.001211362,0.001011166,0.0019252991,0.0016043768,0.00053521065],"domain_scores_gemma":[0.990613,0.003040489,0.0011059397,0.0024539926,0.0020391853,0.00074736663],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.010518649,0.0052374103,0.0035013051,0.005802298,0.0021767633,0.00505448,0.00586406,0.002401217,0.079353176],"category_scores_gemma":[0.020410746,0.0032079588,0.0037635665,0.005038343,0.0014736812,0.0059996983,0.0075412183,0.0069019687,0.15047531],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007990164,0.00008433047,0.0013721267,0.004930392,0.00035747245,0.00038795755,0.0003580704,0.0012559352,0.01078595,0.0057817698,0.9290538,0.04483317],"study_design_scores_gemma":[0.00037211017,0.00012626637,0.0018000527,0.0012277049,0.00017251524,0.0007206695,0.00011833471,0.011399174,0.015747176,0.01660163,0.9514378,0.00027656334],"about_ca_topic_score_codex":0.0020110167,"about_ca_topic_score_gemma":0.0024776373,"teacher_disagreement_score":0.079353176,"about_ca_system_score_codex":0.001380007,"about_ca_system_score_gemma":0.006538193,"threshold_uncertainty_score":0.26546282},"labels":[],"label_agreement":null},{"id":"W4283659942","doi":"10.1093/bioinformatics/btac420","title":"Improving candidate Biosynthetic Gene Clusters in fungi through reinforcement learning","year":2022,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Concordia University; Université du Québec à Montréal","funders":"Fonds de recherche du Québec – Nature et technologies; Natural Sciences and Engineering Research Council of Canada; Université du Québec à Montréal","keywords":"Candidate gene; Aspergillus nidulans; Computer science; Task (project management); Reinforcement learning; Genome; Computational biology; Gene; Artificial intelligence; Biology; Genetics; Engineering","score_opus":0.007802561118249228,"score_gpt":0.21366434891964242,"score_spread":0.2058617878013932,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4283659942","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.41372007,0.0019608096,0.5675875,0.0008442676,0.0001209362,0.00016361669,0.0015355275,0.010564396,0.003502851],"genre_scores_gemma":[0.77199095,0.00028537563,0.22367144,0.0002919972,0.000032085874,0.00009128855,0.0021047832,0.00027534203,0.0012567132],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9994641,0.00012776976,0.00002310435,0.00023339286,0.00009925593,0.000052411902],"domain_scores_gemma":[0.9981812,0.0010360666,0.00016705596,0.00019260177,0.00030437007,0.00011870594],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00152741,0.0010043747,0.00076320465,0.00056434056,0.00044888924,0.0007222955,0.0011120685,0.000759878,0.0016841295],"category_scores_gemma":[0.0048554954,0.00029478085,0.00061232114,0.00044652683,0.00056298054,0.00072948384,0.0009210005,0.0010375531,0.0007396427],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007600203,0.00030080622,0.017093766,0.00041109786,0.00012095757,0.00013340043,0.000102559454,0.74435574,0.02691329,0.0018335488,0.003909271,0.20406547],"study_design_scores_gemma":[0.000027053748,0.000052944783,0.0007675628,0.000015247937,0.000017726468,0.000021413834,0.000009659945,0.9898637,0.007141343,0.0015088478,0.00056614587,0.000008308807],"about_ca_topic_score_codex":0.0064389165,"about_ca_topic_score_gemma":0.00668074,"teacher_disagreement_score":0.0064389165,"about_ca_system_score_codex":0.0010027899,"about_ca_system_score_gemma":0.0015025216,"threshold_uncertainty_score":0.012802839},"labels":[],"label_agreement":null},{"id":"W4283716699","doi":"10.1093/bioinformatics/btac429","title":"RAPPPID: towards generalizable protein interaction prediction with AWD-LSTM twin networks","year":2022,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":35,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; Mila - Quebec Artificial Intelligence Institute","funders":"Compute Canada; Natural Sciences and Engineering Research Council of Canada; McGill University","keywords":"Computer science; Regularization (linguistics); Machine learning; Artificial intelligence; Source code; Code (set theory); Training set; Set (abstract data type); Data mining","score_opus":0.008924453457807548,"score_gpt":0.21018700338334148,"score_spread":0.20126254992553394,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4283716699","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.04841448,0.0009500531,0.92698395,0.00063255057,0.0001689826,0.0001285445,0.0025419227,0.017920936,0.0022585117],"genre_scores_gemma":[0.43845123,0.0006283835,0.53938925,0.0009198909,0.000121256475,0.0005043685,0.010573535,0.001202591,0.008209421],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9995989,0.00007944469,0.000016852278,0.00017475607,0.00008568975,0.00004445916],"domain_scores_gemma":[0.99935836,0.00028283743,0.000060060756,0.00012453223,0.000105894454,0.00006838403],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012264111,0.0015987486,0.001027967,0.0007147337,0.0004110689,0.0010795317,0.0031830932,0.0017647705,0.003471207],"category_scores_gemma":[0.0035304152,0.0009008757,0.0010281692,0.000809017,0.0006598642,0.0021731516,0.002769174,0.0031641645,0.0020020572],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00055690494,0.0002517359,0.0028093567,0.00040041038,0.00033619054,0.00030765127,0.0001534998,0.69652677,0.015045338,0.008558291,0.023405734,0.25164807],"study_design_scores_gemma":[0.00001187815,0.000013745636,0.00008829739,0.000004681724,0.0000068193167,0.000016783817,0.000005014582,0.99492306,0.0011138108,0.0033229345,0.00048824016,0.0000048090656],"about_ca_topic_score_codex":0.005374288,"about_ca_topic_score_gemma":0.008227117,"teacher_disagreement_score":0.005374288,"about_ca_system_score_codex":0.001021728,"about_ca_system_score_gemma":0.0011837977,"threshold_uncertainty_score":0.011612296},"labels":[],"label_agreement":null},{"id":"W4288429604","doi":"10.1093/bioinformatics/btac497","title":"BioCaster in 2021: automatic disease outbreaks detection from global news media","year":2022,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Data-Driven Disease Surveillance","field":"Medicine","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université Laval; McGill University","funders":"Economic and Social Research Council","keywords":"Computer science; Upgrade; Outbreak; Infectious disease (medical specialty); Dashboard; Machine translation; Visualization; Artificial neural network; Ontology; Data science; Cloud computing; Semantics (computer science); World Wide Web; Artificial intelligence; Disease; Medicine; Programming language","score_opus":0.01273902401072915,"score_gpt":0.24822163088394172,"score_spread":0.23548260687321257,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4288429604","genre_codex":"dataset","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.038256127,0.0028746964,0.028500749,0.0071860286,0.0012810883,0.0007977949,0.7482047,0.14999256,0.022906197],"genre_scores_gemma":[0.04885078,0.00090415403,0.085390374,0.0011114223,0.00040617224,0.0005415199,0.85165554,0.003938824,0.0072012823],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9982967,0.0003605964,0.00014840087,0.00050141924,0.0005244045,0.00016845395],"domain_scores_gemma":[0.9927428,0.0019985868,0.00067838083,0.0013109045,0.0023804698,0.0008888637],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0029551394,0.0014930313,0.00094429066,0.0073569617,0.0013229129,0.003477512,0.0020402914,0.0012140272,0.019762846],"category_scores_gemma":[0.012975638,0.00059119845,0.0011126879,0.0061223633,0.0004980227,0.0019297666,0.0026889571,0.0011170591,0.019880014],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00062192953,0.00010109638,0.012708754,0.0010805655,0.00021769865,0.0003174894,0.000250545,0.0015323961,0.0043201274,0.0014916072,0.9087945,0.06856316],"study_design_scores_gemma":[0.0009646087,0.00024290067,0.059580814,0.00064738933,0.00046229316,0.0006140946,0.00082857173,0.073517986,0.023939312,0.010886183,0.82807225,0.00024349117],"about_ca_topic_score_codex":0.0511282,"about_ca_topic_score_gemma":0.08726595,"teacher_disagreement_score":0.0511282,"about_ca_system_score_codex":0.0023027183,"about_ca_system_score_gemma":0.0030206104,"threshold_uncertainty_score":0.101661205},"labels":[],"label_agreement":null},{"id":"W4292961172","doi":"10.1093/bioinformatics/btac564","title":"ntHash2: recursive spaced seed hashing for nucleotide sequences","year":2022,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Algorithms and Data Compression","field":"Computer Science","cited_by":22,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canada's Michael Smith Genome Sciences Centre; University of British Columbia","funders":"National Institutes of Health","keywords":"Hash function; Computer science; Dynamic perfect hashing; Universal hashing; Sequence (biology); Linear hashing; Algorithm; Hash table; Theoretical computer science; Perfect hash function; Double hashing; Biology; Genetics; Programming language","score_opus":0.022581972130463614,"score_gpt":0.25195325326204365,"score_spread":0.22937128113158003,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4292961172","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.030408397,0.00085474347,0.9366002,0.00021866671,0.00030201455,0.00026509236,0.002947528,0.024628434,0.0037750888],"genre_scores_gemma":[0.21274559,0.00035026605,0.7678595,0.00021578114,0.00010773331,0.0004405181,0.008409044,0.0021054603,0.0077660847],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9988795,0.00017886465,0.00010832066,0.00024280767,0.0005035585,0.00008686992],"domain_scores_gemma":[0.9982431,0.0004753008,0.00013888856,0.0006269772,0.0003962083,0.000119645636],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011736227,0.00079421065,0.0007444144,0.0009913014,0.0006988385,0.0010551,0.0019822773,0.00078646815,0.00946551],"category_scores_gemma":[0.0062016323,0.0005283648,0.000631374,0.0014721763,0.00081626466,0.0020252457,0.0023827767,0.00092418387,0.007543378],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.002676787,0.00018449883,0.007167232,0.0011847613,0.00015096345,0.00042157678,0.0007402276,0.038655747,0.13678755,0.041584693,0.07070463,0.6997413],"study_design_scores_gemma":[0.00040041868,0.0006469808,0.0029357204,0.00014613521,0.000069172056,0.0013916076,0.00026307398,0.61319715,0.23794708,0.05439419,0.08838964,0.00021887789],"about_ca_topic_score_codex":0.0014817521,"about_ca_topic_score_gemma":0.0023990853,"teacher_disagreement_score":0.00946551,"about_ca_system_score_codex":0.0006495165,"about_ca_system_score_gemma":0.0014081937,"threshold_uncertainty_score":0.031665325},"labels":[],"label_agreement":null},{"id":"W4294631394","doi":"10.1093/bioinformatics/btac599","title":"CCPLS reveals cell-type-specific spatial dependence of transcriptomes in single cells","year":2022,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Single-cell and spatial transcriptomics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Institute of Genetics; Japan Society for the Promotion of Science; Research Organization of Information and Systems","keywords":"Cell type; Transcriptome; Cell; Interpretability; Context (archaeology); Computational biology; Biology; Gene expression; Computer science; Gene; Genetics; Artificial intelligence","score_opus":0.018084635621990055,"score_gpt":0.21369034047721464,"score_spread":0.19560570485522458,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4294631394","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.271086,0.0015541486,0.70605206,0.0005008537,0.0001079647,0.00007629198,0.01165567,0.006642962,0.002324064],"genre_scores_gemma":[0.8505404,0.0009206699,0.13249902,0.0003049215,0.00010166352,0.0002749264,0.012961994,0.0010604084,0.0013359895],"study_design_codex":"simulation_or_modeling","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9994259,0.00013785737,0.00002436715,0.00024023908,0.000102977705,0.00006869443],"domain_scores_gemma":[0.99789965,0.0011843565,0.00031916978,0.00027250047,0.00022811299,0.00009613468],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014510618,0.00079659175,0.00097108184,0.001012182,0.0004668534,0.0011148257,0.0007621664,0.0006471404,0.0031117497],"category_scores_gemma":[0.0043548085,0.00036086002,0.0014952908,0.0015070682,0.0008082415,0.00085699913,0.0009480733,0.0011378673,0.0012617962],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011699434,0.00019151346,0.092525855,0.001998961,0.0006599428,0.00078345556,0.0010017946,0.3658018,0.34833804,0.015131857,0.013407425,0.1589894],"study_design_scores_gemma":[0.000037219306,0.00015909757,0.04422329,0.000075108095,0.00014667984,0.00035770575,0.00029458583,0.8790965,0.0436914,0.023880666,0.007922763,0.00011502047],"about_ca_topic_score_codex":0.002490945,"about_ca_topic_score_gemma":0.0026546977,"teacher_disagreement_score":0.0031117497,"about_ca_system_score_codex":0.000507903,"about_ca_system_score_gemma":0.0011182736,"threshold_uncertainty_score":0.010409832},"labels":[],"label_agreement":null},{"id":"W4295681693","doi":"10.1093/bioinformatics/btac628","title":"SQuAPP—simple quantitative analysis of proteins and PTMs","year":2022,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Advanced Proteomics Techniques and Applications","field":"Chemistry","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"BC Children's Hospital; University of British Columbia","funders":"Michael Cuccione Foundation; Natural Sciences and Engineering Research Council of Canada; Michael Smith Health Research BC","keywords":"Computer science; Workflow; Proteomics; Proteome; Pipeline (software); Visualization; Software; Quantitative proteomics; Data science; Data mining; Computational biology; Bioinformatics; Database; Chemistry; Biology; Programming language","score_opus":0.020229492025920257,"score_gpt":0.29578866180857577,"score_spread":0.2755591697826555,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4295681693","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.033987265,0.001637088,0.69551075,0.0004939744,0.00033773071,0.00043449766,0.026494076,0.23554872,0.005555956],"genre_scores_gemma":[0.12555707,0.0013460828,0.7886683,0.0010120091,0.00018882663,0.0017386909,0.045644514,0.027576897,0.008267571],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99790835,0.0001983756,0.00013397163,0.0006307521,0.000990515,0.0001381902],"domain_scores_gemma":[0.9977748,0.0007252391,0.00029089738,0.00039345975,0.0006418675,0.00017376838],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002806118,0.002151887,0.0014932469,0.00197033,0.00083152024,0.0022775366,0.0025835298,0.0009106038,0.017705025],"category_scores_gemma":[0.005193722,0.0013085464,0.0017056088,0.0015024737,0.00066392863,0.0018434594,0.002517515,0.002619863,0.011414207],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0033270952,0.00031877303,0.009252448,0.0037354953,0.0010714817,0.00093485176,0.00039119218,0.0082789995,0.6213981,0.011464983,0.12678024,0.21304637],"study_design_scores_gemma":[0.00030102878,0.0005794623,0.012449773,0.0003021927,0.0002665007,0.0014596301,0.000118507785,0.16312358,0.6688359,0.016931921,0.13515207,0.0004793939],"about_ca_topic_score_codex":0.0019373584,"about_ca_topic_score_gemma":0.0024261593,"teacher_disagreement_score":0.017705025,"about_ca_system_score_codex":0.0006925671,"about_ca_system_score_gemma":0.0013077828,"threshold_uncertainty_score":0.059229195},"labels":[],"label_agreement":null},{"id":"W4296780664","doi":"10.1093/bioinformatics/btac478","title":"GNN-SubNet: disease subnetwork detection with explainable graph neural networks","year":2022,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":68,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Austrian Science Fund; European Commission","keywords":"Subnetwork; Subnet; Interpretability; Computer science; Python (programming language); Graph; Machine learning; Artificial intelligence; Data mining; Theoretical computer science; Computer network","score_opus":0.005619933380862117,"score_gpt":0.188813731602062,"score_spread":0.18319379822119988,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4296780664","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.032639027,0.0008560882,0.8737591,0.0015379594,0.00028363918,0.0002759311,0.021743266,0.06519765,0.003707375],"genre_scores_gemma":[0.37891915,0.0008233497,0.55593103,0.000868445,0.00017879458,0.0007700063,0.05228948,0.003483922,0.0067358897],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9997348,0.00007014718,0.000014915881,0.000098289944,0.000052953004,0.000028867138],"domain_scores_gemma":[0.9993818,0.0003249033,0.00006405955,0.00011193845,0.0000712135,0.00004604779],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00079185405,0.0014594822,0.0006519479,0.0017336946,0.00037007369,0.0008955051,0.001721899,0.0012198921,0.010173258],"category_scores_gemma":[0.004438291,0.00050173304,0.0012819378,0.0007305556,0.00041686982,0.0009971339,0.0014546951,0.001224648,0.002334495],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00058944203,0.00024982385,0.017019564,0.0012494075,0.0007341512,0.0008717372,0.0002485399,0.49445155,0.0077488082,0.024636548,0.11685314,0.33534724],"study_design_scores_gemma":[0.000043660737,0.000026411359,0.0009250205,0.000041951214,0.00004201187,0.00012973844,0.000014375884,0.96536183,0.0020146035,0.024012337,0.0073704347,0.000017693892],"about_ca_topic_score_codex":0.0070354473,"about_ca_topic_score_gemma":0.012103436,"teacher_disagreement_score":0.010173258,"about_ca_system_score_codex":0.0009179084,"about_ca_system_score_gemma":0.001223609,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W4298125100","doi":"10.1093/bioinformatics/btac655","title":"Integrating phylogenetic and functional data in microbiome studies","year":2022,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gut microbiota and health","field":"Biochemistry, Genetics and Molecular Biology","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Dalhousie University; McGill University","funders":"Natural Sciences and Engineering Research Council of Canada; Mitacs; Tel Aviv University; Israel Science Foundation","keywords":"Phylogenetic tree; Microbiome; Computational biology; Metagenomics; Biology; Computer science; Evolutionary biology; Data science; Bioinformatics; Genetics; Gene","score_opus":0.04460253469201658,"score_gpt":0.29736789798655766,"score_spread":0.2527653632945411,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4298125100","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.13163152,0.012487015,0.76326907,0.0070895813,0.0006370254,0.00036547988,0.0614434,0.0152495075,0.007827406],"genre_scores_gemma":[0.2804045,0.004259488,0.67572945,0.001918671,0.0005791318,0.00048698374,0.032597043,0.003118984,0.0009057685],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.98863935,0.0068936604,0.00054648134,0.0019725254,0.0016341268,0.0003137975],"domain_scores_gemma":[0.9575362,0.029216014,0.0033482432,0.004628386,0.0032669124,0.0020042278],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01927925,0.0016759759,0.0018886952,0.009504028,0.001469867,0.0040257024,0.002300594,0.0019353116,0.0055766455],"category_scores_gemma":[0.06910492,0.0014062136,0.001977646,0.01186727,0.0018310734,0.004411818,0.007831227,0.0033292656,0.0029973143],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0027259872,0.0005794828,0.30193856,0.010805931,0.0047653774,0.0013798075,0.0025307473,0.101043135,0.06880355,0.055261895,0.052394837,0.39777058],"study_design_scores_gemma":[0.000295781,0.00047480222,0.13683827,0.0024170533,0.0012989868,0.0018704787,0.0011832013,0.2842386,0.02879355,0.40859303,0.13344438,0.0005518517],"about_ca_topic_score_codex":0.0027313824,"about_ca_topic_score_gemma":0.0041869055,"teacher_disagreement_score":0.01927925,"about_ca_system_score_codex":0.0010400238,"about_ca_system_score_gemma":0.0023249106,"threshold_uncertainty_score":0.101959646},"labels":[],"label_agreement":null},{"id":"W4304757181","doi":"10.1093/bioinformatics/btac649","title":"METAbolomics data Balancing with Over-sampling Algorithms (META-BOA): an online resource for addressing class imbalance","year":2022,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Imbalanced Data Classification Techniques","field":"Computer Science","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"National Research Council Canada; University of Ottawa","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Computer science; Overfitting; Sampling (signal processing); Class (philosophy); Data mining; Machine learning; Sample (material); Visualization; Random forest; Resource (disambiguation); Artificial intelligence; Algorithm; Artificial neural network","score_opus":0.1940735828464191,"score_gpt":0.3426043390083074,"score_spread":0.1485307561618883,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4304757181","genre_codex":"software","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0060848617,0.0013232352,0.28779203,0.0015014451,0.0005911322,0.0006767624,0.093683966,0.60044587,0.007900686],"genre_scores_gemma":[0.057655,0.0013906166,0.63524884,0.00299027,0.00044915514,0.0037364508,0.18239088,0.10597929,0.010159506],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9971943,0.0005250673,0.00025270175,0.0007130894,0.0011396555,0.00017518166],"domain_scores_gemma":[0.98992157,0.0050721057,0.0007400602,0.002121118,0.0015961095,0.00054914475],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0071681584,0.0026970035,0.002076896,0.0043576853,0.0013826473,0.0034668518,0.004394562,0.0021464976,0.06572302],"category_scores_gemma":[0.023960428,0.0015857092,0.003008183,0.0037239492,0.0005981429,0.0038082367,0.00528297,0.0029951718,0.040904053],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0020637636,0.00043032074,0.00885072,0.0026085505,0.00066207506,0.00043320184,0.00044387177,0.0076895203,0.013267263,0.0066166706,0.7250973,0.2318368],"study_design_scores_gemma":[0.0014995143,0.00042088883,0.018410262,0.0012735561,0.00039482964,0.001036697,0.00028377367,0.25995266,0.060303073,0.07375675,0.5819941,0.0006740035],"about_ca_topic_score_codex":0.0028846476,"about_ca_topic_score_gemma":0.005453579,"teacher_disagreement_score":0.06572302,"about_ca_system_score_codex":0.00097375014,"about_ca_system_score_gemma":0.0026642403,"threshold_uncertainty_score":0.21986538},"labels":[],"label_agreement":null},{"id":"W4307091043","doi":"10.1093/bioinformatics/btac692","title":"Defining the extent of gene function using ROC curvature","year":2022,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"U.S. National Library of Medicine; National Institute of Mental Health; National Institutes of Health","keywords":"Genome; Biology; Computational biology; Robustness (evolution); Gene; Generalizability theory; Context (archaeology); Computer science; Function (biology); Genetics; Mathematics","score_opus":0.009905996297909917,"score_gpt":0.21944269434356733,"score_spread":0.20953669804565742,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4307091043","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.48021266,0.0028056437,0.49474388,0.0011147533,0.00009543313,0.00024082254,0.005552921,0.007569475,0.007664445],"genre_scores_gemma":[0.95144576,0.0003870946,0.042917777,0.0001929589,0.00006147143,0.00016584044,0.0036749055,0.00051949493,0.00063460955],"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9904545,0.0031682705,0.00070207374,0.0023534375,0.002627611,0.0006941399],"domain_scores_gemma":[0.9098131,0.06203613,0.007863374,0.010719531,0.0077677737,0.0018000877],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.014824057,0.0022267883,0.001729337,0.0072790715,0.0005775418,0.004127587,0.0016127442,0.0019698434,0.0020247253],"category_scores_gemma":[0.079720214,0.0003885588,0.001551005,0.0044929124,0.0031561237,0.0027348304,0.002426284,0.002223424,0.001294119],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0019394084,0.0003538138,0.21750906,0.0012859179,0.00069820025,0.0005384941,0.0012806774,0.40973476,0.032993194,0.026145859,0.0076130326,0.2999076],"study_design_scores_gemma":[0.000050609367,0.0005997753,0.086781666,0.00016729748,0.00012120774,0.00071457186,0.00038437505,0.83141303,0.02762068,0.046940282,0.004995123,0.0002113842],"about_ca_topic_score_codex":0.0034498791,"about_ca_topic_score_gemma":0.0011254449,"teacher_disagreement_score":0.014824057,"about_ca_system_score_codex":0.0018345366,"about_ca_system_score_gemma":0.00093809457,"threshold_uncertainty_score":0.07839805},"labels":[],"label_agreement":null},{"id":"W4310594072","doi":"10.1093/bioinformatics/btac779","title":"DeepCellEss: cell line-specific essential protein prediction with attention-based interpretable deep learning","year":2022,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Machine Learning in Bioinformatics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":30,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Saskatchewan","funders":"National Natural Science Foundation of China","keywords":"Interpretability; Computer science; Artificial intelligence; Benchmark (surveying); Source code; Machine learning; Convolutional neural network; Deep learning; Code (set theory); Data mining","score_opus":0.004123358410521841,"score_gpt":0.19876093933838215,"score_spread":0.19463758092786032,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4310594072","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.1673307,0.0028527824,0.7769747,0.0013199758,0.00029381635,0.00020840281,0.011296205,0.034953106,0.004770443],"genre_scores_gemma":[0.68593884,0.0011476822,0.26834756,0.001154377,0.00017358013,0.0004318097,0.033982623,0.0009967057,0.007826785],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9997403,0.000039220267,0.000012566551,0.00010631374,0.000063531166,0.000038069204],"domain_scores_gemma":[0.9994259,0.00026432806,0.00006789464,0.00008714611,0.00009962511,0.000055189],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007513819,0.0015097916,0.00086760847,0.0009237393,0.00032764397,0.00095661107,0.0020518475,0.0013720858,0.0029090315],"category_scores_gemma":[0.0019847457,0.0004843093,0.0011246111,0.00063851377,0.00049423776,0.0012969231,0.0011740223,0.0019162329,0.0011337013],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005398793,0.0003935287,0.010956239,0.00059830584,0.00037079782,0.0004815718,0.0001232481,0.708298,0.027735606,0.012273495,0.0395223,0.19870704],"study_design_scores_gemma":[0.000013758375,0.000025070884,0.00026225983,0.000007858321,0.000011405506,0.000025256555,0.000004522912,0.99170154,0.0030681614,0.0040633725,0.0008097254,0.0000070678666],"about_ca_topic_score_codex":0.0054447358,"about_ca_topic_score_gemma":0.008265897,"teacher_disagreement_score":0.0054447358,"about_ca_system_score_codex":0.0012710909,"about_ca_system_score_gemma":0.0011041284,"threshold_uncertainty_score":0.010826111},"labels":[],"label_agreement":null},{"id":"W4311575706","doi":"10.1093/bioinformatics/btac809","title":"Efficient computation of contributional diversity metrics from microbiome data with <i>FuncDiv</i>","year":2022,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gut microbiota and health","field":"Biochemistry, Genetics and Molecular Biology","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; Dalhousie University; McGill Genome Centre","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"License; Metagenomics; Computer science; Microbiome; ENCODE; Diversity (politics); Function (biology); Alpha diversity; Data mining; Computational biology; Data science; Biology; Bioinformatics; Biodiversity; Genetics; Gene; Ecology","score_opus":0.01919680741520582,"score_gpt":0.24245384039659265,"score_spread":0.22325703298138683,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4311575706","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.062362246,0.0010511295,0.6962617,0.00091830356,0.00027179765,0.00021950535,0.032288536,0.19896685,0.007659926],"genre_scores_gemma":[0.16184522,0.000496334,0.7588304,0.0005350183,0.00015458008,0.00059809984,0.059253566,0.015649587,0.002637056],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9987268,0.00026013106,0.00011676731,0.00040246648,0.0003473065,0.00014654467],"domain_scores_gemma":[0.9965605,0.0016362201,0.00033213894,0.00067506864,0.0005822819,0.00021377853],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0026342762,0.0026446546,0.0013287106,0.0028207933,0.0010168416,0.003126844,0.002469438,0.0009369115,0.009161694],"category_scores_gemma":[0.014655465,0.00087484234,0.002231926,0.002620809,0.0010886462,0.0030580899,0.0032305247,0.0015533216,0.007882582],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0017030236,0.0004223101,0.056369457,0.0030626112,0.0011346022,0.0008476697,0.0015207074,0.1315078,0.061070897,0.03437393,0.1879996,0.51998746],"study_design_scores_gemma":[0.00024061374,0.00024062222,0.0122783715,0.00024534937,0.00012804224,0.00077868206,0.00042109485,0.7882537,0.04575679,0.08263841,0.06877641,0.00024184659],"about_ca_topic_score_codex":0.0035448105,"about_ca_topic_score_gemma":0.0054948847,"teacher_disagreement_score":0.009161694,"about_ca_system_score_codex":0.0013284783,"about_ca_system_score_gemma":0.002336298,"threshold_uncertainty_score":0.030648887},"labels":[],"label_agreement":null},{"id":"W4312018781","doi":"10.1093/bioinformatics/btac813","title":"Integrative chromatin domain annotation through graph embedding of Hi-C data","year":2022,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"Compute Canada","keywords":"Annotation; Chromatin; Computational biology; Genome; Embedding; Domain (mathematical analysis); Pairwise comparison; Epigenomics; Genomics; Computer science; Graph; Biology; Genetics; Theoretical computer science; Artificial intelligence; Gene; DNA methylation; Gene expression; Mathematics","score_opus":0.01531051330598558,"score_gpt":0.26775388303336095,"score_spread":0.25244336972737536,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4312018781","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.09287519,0.00066280743,0.8545838,0.0005521004,0.00008996673,0.00019764133,0.027834127,0.020393694,0.002810758],"genre_scores_gemma":[0.41547585,0.00041113337,0.490829,0.0003488548,0.000062760904,0.0002732105,0.08772738,0.001746643,0.0031251216],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9995764,0.00007694944,0.000013992449,0.0002207825,0.000070302674,0.00004157573],"domain_scores_gemma":[0.99904555,0.00040547864,0.00008103218,0.00023388746,0.00016833932,0.000065790664],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005153239,0.00105707,0.00056624226,0.0021754815,0.00043388415,0.00092483574,0.0015917162,0.001206577,0.0028019873],"category_scores_gemma":[0.0024705494,0.0004334061,0.0009953419,0.0017154113,0.00046417964,0.0012091283,0.001173322,0.0012607191,0.0016094],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001009564,0.00051575457,0.025425207,0.0016263304,0.0004842075,0.0007830253,0.0006056763,0.45755428,0.10572253,0.024749126,0.05935949,0.3221648],"study_design_scores_gemma":[0.000022062544,0.000042199023,0.0034045905,0.000034783017,0.000043922868,0.00012434497,0.00007339783,0.95789933,0.011102916,0.018809844,0.008408268,0.000034246143],"about_ca_topic_score_codex":0.0070530376,"about_ca_topic_score_gemma":0.012982387,"teacher_disagreement_score":0.0070530376,"about_ca_system_score_codex":0.00073779933,"about_ca_system_score_gemma":0.0006269535,"threshold_uncertainty_score":0.01402396},"labels":[],"label_agreement":null},{"id":"W4316494517","doi":"10.1093/bioinformatics/btad031","title":"Cytoscape.js 2023 update: a graph theory library for visualization and analysis","year":2023,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":165,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Sinai Health System; Princess Margaret Cancer Centre; Lunenfeld-Tanenbaum Research Institute; University Health Network; University of Toronto","funders":"National Center for Research Resources; National Institute of General Medical Sciences; National Cancer Institute; National Human Genome Research Institute; National Institutes of Health","keywords":"Computer science; JavaScript; Visualization; Documentation; Source code; Graph; Software; Graph drawing; Web application; Programming language; World Wide Web; Information retrieval; Data mining; Theoretical computer science","score_opus":0.007757637779909554,"score_gpt":0.23830022338613036,"score_spread":0.2305425856062208,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4316494517","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0014450615,0.001978029,0.16146375,0.0015415186,0.0022605534,0.00042062713,0.31492892,0.501112,0.014849505],"genre_scores_gemma":[0.012968658,0.0028533305,0.21001492,0.0029188092,0.0009852875,0.0020867807,0.5600313,0.18475811,0.023382785],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9975769,0.00032784045,0.00022946199,0.00046802658,0.0012055706,0.00019216145],"domain_scores_gemma":[0.99402654,0.0021082317,0.00043793005,0.0009966553,0.0018552023,0.0005753529],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002750324,0.0024466119,0.0024685992,0.0071470914,0.0019030309,0.004284323,0.0068347254,0.0022163983,0.13409476],"category_scores_gemma":[0.01872673,0.0027205157,0.0020900827,0.005073053,0.0005181217,0.0052224984,0.0034023423,0.005376002,0.116041034],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00009563343,0.00003227421,0.0003231577,0.0015234168,0.00018891663,0.00014415367,0.00010952702,0.00078077224,0.004361635,0.0037148204,0.9449568,0.04376899],"study_design_scores_gemma":[0.0001628849,0.000028497541,0.0010060787,0.00030920235,0.00016525348,0.0005522254,0.00004217701,0.0060726292,0.0130346455,0.016551578,0.9618647,0.00021009333],"about_ca_topic_score_codex":0.00871795,"about_ca_topic_score_gemma":0.011927817,"teacher_disagreement_score":0.13409476,"about_ca_system_score_codex":0.0010730671,"about_ca_system_score_gemma":0.0042901733,"threshold_uncertainty_score":0.44859165},"labels":[],"label_agreement":null},{"id":"W4316926479","doi":"10.1093/bioinformatics/btad032","title":"JBrowse Jupyter: a Python interface to JBrowse 2","year":2023,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ontario Institute for Cancer Research","funders":"National Institute of General Medical Sciences; National Human Genome Research Institute; National Institutes of Health","keywords":"Python (programming language); Computer science; Programming language; Interface (matter); Operating system","score_opus":0.014844535896914578,"score_gpt":0.26097935467307,"score_spread":0.2461348187761554,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4316926479","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0017596743,0.00044111337,0.21280058,0.00056251365,0.00037084642,0.0003008559,0.036645997,0.73847383,0.008644619],"genre_scores_gemma":[0.018844731,0.0007655347,0.27858022,0.001516489,0.00025271994,0.0020728677,0.101420395,0.57855767,0.01798945],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99729234,0.0005020823,0.00026081316,0.0005880159,0.0009216063,0.0004351513],"domain_scores_gemma":[0.99556667,0.0016635983,0.00037944727,0.000877021,0.00094220863,0.00057103444],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005602944,0.003033731,0.002327728,0.0026964904,0.0016799257,0.0040292637,0.007174361,0.0017198666,0.12087108],"category_scores_gemma":[0.01410287,0.003128768,0.0027215756,0.0023449643,0.0016850589,0.0055070384,0.0067916205,0.0063095135,0.14697856],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009477812,0.0001584072,0.0019617623,0.0017552893,0.0002493529,0.00036391517,0.0006307758,0.002683213,0.012949569,0.007875519,0.90070444,0.06971985],"study_design_scores_gemma":[0.0006309617,0.00010818495,0.0039553,0.0005991089,0.00012119098,0.00062425097,0.00014227601,0.029123126,0.030311493,0.029186567,0.9046851,0.00051250024],"about_ca_topic_score_codex":0.004035129,"about_ca_topic_score_gemma":0.0033108692,"teacher_disagreement_score":0.12087108,"about_ca_system_score_codex":0.0009224095,"about_ca_system_score_gemma":0.003433947,"threshold_uncertainty_score":0.40435404},"labels":[],"label_agreement":null},{"id":"W4317212859","doi":"10.1093/bioinformatics/btad036","title":"HelixGAN a deep-learning methodology for conditional <i>de novo</i> design of α-helix structures","year":2023,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Chemical Synthesis and Analysis","field":"Biochemistry, Genetics and Molecular Biology","cited_by":22,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"Canadian Institutes of Health Research","keywords":"Computational biology; Peptide; Synthetic biology; Peptidomimetic; Computer science; Protein design; Helix (gastropod); Deep learning; Protein structure; Chemistry; Biology; Artificial intelligence; Biochemistry","score_opus":0.044930952105651935,"score_gpt":0.3029639355541699,"score_spread":0.25803298344851794,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4317212859","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.012702422,0.00038020435,0.9833723,0.000239274,0.0000442552,0.0000405361,0.00013373284,0.0014316345,0.0016556672],"genre_scores_gemma":[0.52200043,0.0005616358,0.46601754,0.00081009185,0.00009222543,0.0003161664,0.0012046855,0.0005241591,0.008473048],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99977666,0.00008719723,0.000008230429,0.000053938933,0.00004655169,0.0000273974],"domain_scores_gemma":[0.99941707,0.0003978973,0.000048546448,0.000043089534,0.000061780804,0.000031690644],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00088040327,0.00082983583,0.0005087032,0.00030499717,0.00018099272,0.00040076656,0.000988319,0.0009157887,0.0029655444],"category_scores_gemma":[0.0015197037,0.00047251445,0.0005682633,0.00020745026,0.00064874627,0.0005591857,0.00079520384,0.0015485664,0.00064295874],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00009468211,0.000044456203,0.00048761626,0.000103815044,0.00005531345,0.00007573153,0.000028720608,0.9203507,0.005159762,0.009109333,0.0031323982,0.06135746],"study_design_scores_gemma":[0.000004405207,0.000019515774,0.000023006365,0.000004196917,0.0000026308408,0.000008424344,0.0000015355279,0.9963831,0.0009137954,0.0022207473,0.00041682817,0.000001885528],"about_ca_topic_score_codex":0.0017780681,"about_ca_topic_score_gemma":0.0026627912,"teacher_disagreement_score":0.0029655444,"about_ca_system_score_codex":0.000624996,"about_ca_system_score_gemma":0.0007289526,"threshold_uncertainty_score":0.009920716},"labels":[],"label_agreement":null},{"id":"W4317869953","doi":"10.1093/bioinformatics/btad055","title":"Binding peptide generation for MHC Class I proteins with deep reinforcement learning","year":2023,"lang":"en","type":"article","venue":"Bioinformatics","topic":"vaccines and immunoinformatics approaches","field":"Biochemistry, Genetics and Molecular Biology","cited_by":16,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"National Research Council Canada; University of Ottawa","funders":"National Science Foundation","keywords":"Reinforcement learning; Computational biology; Class (philosophy); Computer science; Peptide; MHC class I; Artificial intelligence; Major histocompatibility complex; Chemistry; Biology; Biochemistry; Gene","score_opus":0.02448110864336796,"score_gpt":0.23952572503999564,"score_spread":0.21504461639662767,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4317869953","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.175636,0.00075072254,0.8114438,0.0005621526,0.00014074397,0.0001852253,0.00043464836,0.005290924,0.005555864],"genre_scores_gemma":[0.77020425,0.00020676425,0.22487539,0.0004211812,0.000035921123,0.00027135757,0.0007131336,0.00030298522,0.0029689965],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99975604,0.000052327014,0.0000114939185,0.00007074364,0.00006500952,0.00004444105],"domain_scores_gemma":[0.9995105,0.00025507112,0.00005948415,0.000048012356,0.000077756544,0.00004915689],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007596832,0.0007415313,0.00054039154,0.00028214406,0.00024489712,0.0004773644,0.0010360922,0.0008777196,0.0025353911],"category_scores_gemma":[0.0015461162,0.00037351777,0.00051093905,0.00021495474,0.00054768776,0.00048424336,0.00072044897,0.0011585442,0.0005699663],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001424536,0.00016315325,0.0019225471,0.00015797197,0.00004343662,0.0001473919,0.000039618004,0.91842806,0.019022724,0.0047996165,0.002303068,0.052829996],"study_design_scores_gemma":[0.000011396924,0.00002146529,0.000067489476,0.0000031401346,0.0000035534808,0.000015770289,0.000001867691,0.9953766,0.0028661783,0.0012784153,0.00035135305,0.0000027795015],"about_ca_topic_score_codex":0.0018986333,"about_ca_topic_score_gemma":0.0016906464,"teacher_disagreement_score":0.0025353911,"about_ca_system_score_codex":0.0009033228,"about_ca_system_score_gemma":0.00082880026,"threshold_uncertainty_score":0.008481741},"labels":[],"label_agreement":null},{"id":"W4318193482","doi":"10.1093/bioinformatics/btad057","title":"How to optimally sample a sequence for rapid analysis","year":2023,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"Japan Science and Technology Agency; U.S. National Library of Medicine; Natural Sciences and Engineering Research Council of Canada; National Institutes of Health","keywords":"Sequence (biology); Computer science; Source code; Sampling (signal processing); Sample (material); Sequence analysis; Sequence alignment; Algorithm; Biology; Peptide sequence; Genetics; DNA; Programming language","score_opus":0.03516069343064127,"score_gpt":0.2718266799308762,"score_spread":0.23666598650023496,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4318193482","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.021920172,0.00078594784,0.970036,0.0017370069,0.0001264463,0.00020295792,0.00052798935,0.003208767,0.0014546866],"genre_scores_gemma":[0.0984957,0.0005964594,0.89619094,0.00073752855,0.00014943472,0.0003374115,0.0014187589,0.001279649,0.0007941325],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.992568,0.0032947809,0.00057478185,0.0014618221,0.0017771848,0.00032332362],"domain_scores_gemma":[0.9772719,0.016361136,0.0011095823,0.0021170853,0.0024519956,0.00068834086],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.006303701,0.0015321362,0.0017332211,0.0015921609,0.0012227352,0.0020312741,0.0017900545,0.0018163311,0.0050877696],"category_scores_gemma":[0.043951135,0.0011998707,0.0012013138,0.0017961151,0.002290186,0.005163869,0.0022279709,0.0027194137,0.005046855],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013404044,0.0004385092,0.009683511,0.002659439,0.00029365357,0.00081628456,0.0012160058,0.18746394,0.09930953,0.056853913,0.03512733,0.6047975],"study_design_scores_gemma":[0.00018090288,0.0003434753,0.0018935504,0.00041408255,0.00013814913,0.0010535105,0.00057328155,0.6426622,0.059858058,0.274561,0.018193614,0.00012811285],"about_ca_topic_score_codex":0.0015049558,"about_ca_topic_score_gemma":0.0018182591,"teacher_disagreement_score":0.006303701,"about_ca_system_score_codex":0.00085196656,"about_ca_system_score_gemma":0.0028803977,"threshold_uncertainty_score":0.033337533},"labels":[],"label_agreement":null},{"id":"W4318391569","doi":"10.1093/bioinformatics/btad063","title":"Efficient penalized generalized linear mixed models for variable selection and genetic risk prediction in high-dimensional data","year":2023,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Montréal; McGill University","funders":"Natural Sciences and Engineering Research Council of Canada; Medical Research Council; Compute Canada","keywords":"Generalized linear mixed model; Lasso (programming language); Feature selection; Genome-wide association study; Curse of dimensionality; Linear model; Generalized linear model; Computer science; Mixed model; Binary number; Model selection; Covariance; Mathematics; Statistics; Artificial intelligence; Biology; Single-nucleotide polymorphism","score_opus":0.025862531734247348,"score_gpt":0.2652881851989897,"score_spread":0.23942565346474237,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4318391569","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0031168833,0.0003404706,0.9950305,0.0004912842,0.000041064264,0.00005468376,0.00026442064,0.00042017215,0.00024055256],"genre_scores_gemma":[0.14881703,0.00081316975,0.8434657,0.0006391472,0.0002696604,0.0010728857,0.002034193,0.00047349313,0.0024147746],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9934568,0.0052349716,0.00018567966,0.0005148163,0.00045174625,0.00015607371],"domain_scores_gemma":[0.9804874,0.015984667,0.0012717255,0.0010852622,0.0009252,0.00024575624],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.009463115,0.001613466,0.0018333461,0.00092313316,0.00074430904,0.0016230094,0.003324598,0.00208818,0.0044309455],"category_scores_gemma":[0.031389844,0.0012474612,0.001856777,0.0016466782,0.0015513004,0.0014220272,0.0021068216,0.00375327,0.0018509255],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0004639035,0.00012868545,0.005150511,0.00052488584,0.00040032924,0.0003929345,0.00021351686,0.79609483,0.001634784,0.069676295,0.010755517,0.11456393],"study_design_scores_gemma":[0.000028215429,0.000019315436,0.00019423489,0.000028737431,0.000017151626,0.000025460537,0.000006761378,0.9788248,0.00021398567,0.019503599,0.0011256511,0.00001215051],"about_ca_topic_score_codex":0.0065852725,"about_ca_topic_score_gemma":0.008166076,"teacher_disagreement_score":0.009463115,"about_ca_system_score_codex":0.0010491213,"about_ca_system_score_gemma":0.0024676109,"threshold_uncertainty_score":0.050046325},"labels":[],"label_agreement":null},{"id":"W4320817603","doi":"10.1093/bioinformatics/btad087","title":"Generalizations of the genomic rank distance to indels","year":2023,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genome Rearrangement Algorithms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Natural Sciences and Engineering Research Council of Canada; Foreign, Commonwealth and Development Office; Fundação de Amparo à Pesquisa do Estado de São Paulo; Medical Research Council; Alfred P. Sloan Foundation","keywords":"Indel; Rank (graph theory); Computer science; Genetics; Biology; Computational biology; Mathematics; Combinatorics; Single-nucleotide polymorphism; Gene; Genotype","score_opus":0.013477248210070252,"score_gpt":0.23895492266185675,"score_spread":0.2254776744517865,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4320817603","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.039478216,0.0012023342,0.94844866,0.00044837568,0.0001804944,0.000100632955,0.0020582522,0.0023636653,0.005719473],"genre_scores_gemma":[0.28354815,0.001387035,0.7027293,0.0005032216,0.0003606916,0.000263797,0.0050178943,0.0011640313,0.0050257966],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99731284,0.00065212004,0.00019915622,0.00083451264,0.0008483085,0.00015312243],"domain_scores_gemma":[0.98942494,0.005399158,0.0010714758,0.0020435876,0.001672362,0.000388449],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002381563,0.000837906,0.00089406,0.003195466,0.0007779451,0.0022023176,0.0025237796,0.0012505678,0.0055150166],"category_scores_gemma":[0.020339474,0.00045136787,0.0010492941,0.004795299,0.001433625,0.0039933287,0.0020770826,0.0023493695,0.002034663],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00047903156,0.00020447238,0.017595839,0.001230722,0.00023392035,0.0006940574,0.00054256653,0.31931975,0.011433061,0.28015622,0.017254692,0.35085565],"study_design_scores_gemma":[0.000070770555,0.00017863583,0.0045947996,0.00012146082,0.000063597574,0.0011580245,0.00017005103,0.5143614,0.009308556,0.44281834,0.02702911,0.0001253384],"about_ca_topic_score_codex":0.002635528,"about_ca_topic_score_gemma":0.0026305916,"teacher_disagreement_score":0.0055150166,"about_ca_system_score_codex":0.0011533019,"about_ca_system_score_gemma":0.001085632,"threshold_uncertainty_score":0.018449545},"labels":[],"label_agreement":null},{"id":"W4321611764","doi":"10.1093/bioinformatics/btad099","title":"CLAIRE: contrastive learning-based batch correction framework for better balance between batch mixing and preservation of cellular heterogeneity","year":2023,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Single-cell and spatial transcriptomics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":12,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Saskatchewan","funders":"National Natural Science Foundation of China","keywords":"Computer science; Benchmark (surveying); Source code; Code (set theory); Artificial intelligence; Machine learning; Programming language","score_opus":0.02165348175230125,"score_gpt":0.2564710270575074,"score_spread":0.23481754530520615,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4321611764","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.010416177,0.00034565935,0.9840232,0.00017523849,0.00006505372,0.00011823568,0.00017449404,0.0038190682,0.0008628782],"genre_scores_gemma":[0.17926143,0.00023157043,0.81146985,0.0007301904,0.00013314799,0.0004396543,0.0015370899,0.0011800291,0.0050169746],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9983884,0.00035276808,0.000086520464,0.00057616265,0.0004541397,0.00014202413],"domain_scores_gemma":[0.9973036,0.0010414681,0.00026639813,0.000437713,0.0007816999,0.00016913541],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004068292,0.001671148,0.0016685502,0.0012406104,0.0009384862,0.0013285334,0.005221047,0.0020398244,0.0032103185],"category_scores_gemma":[0.008591389,0.0008227034,0.001437062,0.00083223206,0.0013103549,0.002516948,0.00239183,0.0031194831,0.0012911661],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008767078,0.0004344681,0.0061046295,0.00040357545,0.00028526012,0.00043897866,0.00038219997,0.36841172,0.05575033,0.021550829,0.015501056,0.52986026],"study_design_scores_gemma":[0.000028525816,0.000055448927,0.00025135526,0.000009430574,0.000019920308,0.00006283468,0.000012321237,0.98504746,0.008239406,0.004329579,0.0019239392,0.000019757674],"about_ca_topic_score_codex":0.008764486,"about_ca_topic_score_gemma":0.0142536,"teacher_disagreement_score":0.008764486,"about_ca_system_score_codex":0.0018227068,"about_ca_system_score_gemma":0.0025613422,"threshold_uncertainty_score":0.021515429},"labels":[],"label_agreement":null},{"id":"W4323353510","doi":"10.1093/bioinformatics/btad118","title":"NDEx IQuery: a multi-method network gene set analysis leveraging the Network Data Exchange","year":2023,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":42,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"National Cancer Institute; National Human Genome Research Institute; Advanced Research Projects Agency; Defense Advanced Research Projects Agency; National Institutes of Health","keywords":"Computer science; Java; Context (archaeology); Set (abstract data type); Biological network; JavaScript; Task (project management); Network analysis; Variety (cybernetics); Software; Gene regulatory network; Data mining; Computational biology; World Wide Web; Artificial intelligence; Programming language; Gene; Biology","score_opus":0.05488798261072136,"score_gpt":0.306567996007621,"score_spread":0.25168001339689966,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4323353510","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.017493296,0.0005743601,0.7997083,0.00061471114,0.00030164616,0.0007065242,0.042612236,0.13084589,0.007143008],"genre_scores_gemma":[0.082374364,0.0003798361,0.82563204,0.00033356395,0.00007746713,0.0018668546,0.06291463,0.020095693,0.006325524],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9985399,0.00036321566,0.00009493031,0.00038972072,0.0005237301,0.000088497785],"domain_scores_gemma":[0.99794894,0.0011445195,0.00011999918,0.00041457187,0.00027996278,0.000092079594],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00405056,0.0016388772,0.0012803405,0.003481531,0.0013403267,0.0025416005,0.0033864581,0.00090118445,0.030599497],"category_scores_gemma":[0.008277518,0.0009469024,0.0021951683,0.002121729,0.00064337614,0.0022983514,0.0037170718,0.0018328783,0.004806086],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0023882943,0.00054093276,0.018221295,0.0035021654,0.0020571523,0.0012791313,0.0012456585,0.12670393,0.05174916,0.07095201,0.32073534,0.4006249],"study_design_scores_gemma":[0.00056916854,0.00017142408,0.0059346794,0.00020176783,0.00022438873,0.00044801587,0.0002896587,0.74916226,0.027408073,0.05643274,0.15895829,0.00019956328],"about_ca_topic_score_codex":0.0086209625,"about_ca_topic_score_gemma":0.011919845,"teacher_disagreement_score":0.030599497,"about_ca_system_score_codex":0.0014591549,"about_ca_system_score_gemma":0.002113141,"threshold_uncertainty_score":0.10236555},"labels":[],"label_agreement":null},{"id":"W4323810601","doi":"10.1093/bioinformatics/btad126","title":"RNAget: an API to securely retrieve RNA quantifications","year":2023,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ontario Genomics; McGill University; Ontario Institute for Cancer Research","funders":"National Human Genome Research Institute; European Regional Development Fund; Medical Research Council; National Institutes of Health; Generalitat de Catalunya; Ministerio de Ciencia e Innovación; Centres de Recerca de Catalunya; Canadian Institutes of Health Research; Genome Canada","keywords":"Computer science; Genomics; Slicing; DNA microarray; Computational biology; Data mining; World Wide Web; Genome; Biology; Genetics; Gene; Gene expression","score_opus":0.034816919898544706,"score_gpt":0.3055031289160492,"score_spread":0.2706862090175045,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4323810601","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0013195989,0.00034557632,0.21163315,0.00020889642,0.00014672909,0.00033981042,0.043891612,0.7381565,0.0039581433],"genre_scores_gemma":[0.04727376,0.0009948888,0.26242477,0.003041983,0.0003132952,0.0046831677,0.3882916,0.2744236,0.018552994],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9965988,0.0005089751,0.00047787942,0.0007449206,0.0012809918,0.00038854062],"domain_scores_gemma":[0.9950059,0.0019295634,0.00038839603,0.0014825267,0.00088485196,0.00030883998],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0047744336,0.0041440725,0.002079256,0.0033853115,0.00093725923,0.0041606436,0.0042582126,0.0018571094,0.05541604],"category_scores_gemma":[0.010205297,0.0025022144,0.002041044,0.0020653177,0.0009420602,0.0043094815,0.006599903,0.0033677632,0.0708766],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0024375997,0.00024993916,0.0042847847,0.0023611474,0.00049119553,0.00073980406,0.0005852566,0.0035887763,0.044872586,0.010429826,0.78707635,0.14288266],"study_design_scores_gemma":[0.00086569425,0.00022543824,0.0074389367,0.00059891964,0.00020189465,0.0014592406,0.0002359474,0.05870196,0.14067626,0.038929332,0.7499163,0.00075013033],"about_ca_topic_score_codex":0.002067064,"about_ca_topic_score_gemma":0.0022382108,"teacher_disagreement_score":0.05541604,"about_ca_system_score_codex":0.0011636823,"about_ca_system_score_gemma":0.001420194,"threshold_uncertainty_score":0.18538517},"labels":[],"label_agreement":null},{"id":"W4327682542","doi":"10.1093/bioinformatics/btad136","title":"GENLIB: new function to simulate haplotype transmission in large complex genealogies","year":2023,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Hôpital Maisonneuve-Rosemont; Université du Québec à Chicoutimi; University of Ottawa","funders":"Natural Sciences and Engineering Research Council of Canada; Canada Foundation for Innovation; Canadian Institutes of Health Research; University of Ottawa; Canarie","keywords":"Haplotype; R package; Computer science; Function (biology); Transmission (telecommunications); Proband; Genealogy; Theoretical computer science; Biology; Mutation; Genetics; Programming language; History; Telecommunications; Gene","score_opus":0.036914413292160185,"score_gpt":0.29490424534431,"score_spread":0.2579898320521498,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4327682542","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.017532837,0.0013580301,0.6239697,0.0011139495,0.00064256816,0.00021824564,0.089245856,0.25794035,0.007978417],"genre_scores_gemma":[0.12908329,0.0014242604,0.5477868,0.0030340326,0.0004119511,0.002800338,0.1412714,0.15863131,0.015556743],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9992237,0.00036299136,0.000056244982,0.0001481947,0.00014289812,0.000065907414],"domain_scores_gemma":[0.99764204,0.0015716018,0.00013554556,0.00033261545,0.00021270444,0.000105422936],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002554646,0.0015886584,0.0014825482,0.0012444899,0.0006458474,0.001800908,0.0034504265,0.0014473364,0.049319766],"category_scores_gemma":[0.011614538,0.001467669,0.0024645058,0.0013841051,0.0006319631,0.0013502171,0.0021593906,0.0020267013,0.019442543],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00077353924,0.00014284113,0.022565285,0.0018030062,0.0016913584,0.00065410126,0.0006284326,0.1760501,0.0035530208,0.03121275,0.6645316,0.09639407],"study_design_scores_gemma":[0.0009025523,0.000118813936,0.004706174,0.0002884688,0.0003341237,0.0009691528,0.00009322833,0.589754,0.006104998,0.07587102,0.32063752,0.00021987813],"about_ca_topic_score_codex":0.006765838,"about_ca_topic_score_gemma":0.0077578346,"teacher_disagreement_score":0.049319766,"about_ca_system_score_codex":0.00090258813,"about_ca_system_score_gemma":0.002050237,"threshold_uncertainty_score":0.16499108},"labels":[],"label_agreement":null},{"id":"W4328048262","doi":"10.1093/bioinformatics/btad139","title":"The hidden factor: accounting for covariate effects in power and sample size computation for a binary trait","year":2023,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Public Health Ontario; University of Toronto","funders":"Natural Sciences and Engineering Research Council of Canada; University of Toronto","keywords":"Covariate; Biobank; Sample size determination; Computer science; Trait; Statistics; Sample (material); Logistic regression; Binary number; Data mining; Replication (statistics); Econometrics; Mathematics; Biology; Bioinformatics","score_opus":0.019299697030622898,"score_gpt":0.2859291187542457,"score_spread":0.2666294217236228,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4328048262","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0024049706,0.00034425914,0.99537355,0.0005725559,0.000076315955,0.0002964678,0.0001488245,0.00029026205,0.0004929112],"genre_scores_gemma":[0.09412636,0.00040962748,0.90115404,0.00057520403,0.00025004352,0.00202078,0.00027882666,0.0002935807,0.000891513],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.95683384,0.035091065,0.0012452246,0.0036355879,0.002687453,0.00050689734],"domain_scores_gemma":[0.8271141,0.14828022,0.005083746,0.015184248,0.0032372985,0.0011003293],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.055008546,0.0013761624,0.0017667573,0.0012505745,0.0008171511,0.002135136,0.0042766873,0.0032964214,0.0066259773],"category_scores_gemma":[0.21977001,0.0010600005,0.00182394,0.002259082,0.003417924,0.0033246456,0.0039483444,0.003995097,0.0013044998],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.002063396,0.00025602983,0.026460761,0.0021757493,0.0010772712,0.00072176155,0.0018958075,0.14482573,0.006609365,0.19475418,0.018430555,0.6007294],"study_design_scores_gemma":[0.0009975444,0.0007690391,0.008363786,0.00084968493,0.00048913853,0.0007516213,0.00013998646,0.57157093,0.006089504,0.38729268,0.022506244,0.0001797081],"about_ca_topic_score_codex":0.0029152648,"about_ca_topic_score_gemma":0.002691979,"teacher_disagreement_score":0.055008546,"about_ca_system_score_codex":0.0012087086,"about_ca_system_score_gemma":0.0041317465,"threshold_uncertainty_score":0.29091644},"labels":[],"label_agreement":null},{"id":"W4362657842","doi":"10.1093/bioinformatics/btad177","title":"MARSY: a multitask deep-learning framework for prediction of drug combination synergy scores","year":2023,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Computational Drug Discovery Methods","field":"Computer Science","cited_by":48,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Mila - Quebec Artificial Intelligence Institute; McGill University","funders":"Natural Sciences and Engineering Research Council of Canada; Institut de Valorisation des Données; Merck Canada; Canada First Research Excellence Fund; Compute Canada; Government of Canada; McGill University","keywords":"Python (programming language); Machine learning; Computer science; Artificial intelligence; Multi-task learning; Deep learning; Drug; Encoder; Drug repositioning; Task (project management); Medicine; Pharmacology","score_opus":0.022969270557317737,"score_gpt":0.29107115244868986,"score_spread":0.26810188189137213,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4362657842","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.10004012,0.004287696,0.8596884,0.0022524574,0.00031924117,0.00028818767,0.009762442,0.02068458,0.0026768816],"genre_scores_gemma":[0.6925762,0.0013684345,0.27846235,0.0016768896,0.00025205224,0.0009093523,0.017029386,0.00067198096,0.007053354],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9995553,0.00012740481,0.000026958545,0.00013079731,0.000089659035,0.00006990914],"domain_scores_gemma":[0.99927324,0.0004022008,0.00007967871,0.0000534178,0.000117481904,0.00007399746],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016223945,0.0018347882,0.0013602911,0.00081886,0.0002975179,0.0009325218,0.0025830115,0.0019894124,0.003315259],"category_scores_gemma":[0.003737957,0.0008593357,0.0015201841,0.0005831523,0.0005668557,0.0012342659,0.0014093186,0.0026878384,0.0010135391],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00038139077,0.00024963223,0.003194179,0.00022722922,0.0002391214,0.00015113765,0.000035444362,0.90570384,0.002476969,0.0026506567,0.011938035,0.07275241],"study_design_scores_gemma":[0.000013577627,0.000032469274,0.000094093026,0.000006600209,0.000008365291,0.0000073383753,0.0000019723939,0.997326,0.0004711112,0.0016434735,0.0003893735,0.000005657347],"about_ca_topic_score_codex":0.009812407,"about_ca_topic_score_gemma":0.013555631,"teacher_disagreement_score":0.009812407,"about_ca_system_score_codex":0.001197332,"about_ca_system_score_gemma":0.002099591,"threshold_uncertainty_score":0.019510627},"labels":[],"label_agreement":null},{"id":"W4365444089","doi":"10.1093/bioinformatics/btad189","title":"Structure-aware protein self-supervised learning","year":2023,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Machine Learning in Bioinformatics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":48,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; Mila - Quebec Artificial Intelligence Institute","funders":"","keywords":"Computer science; Artificial intelligence; Machine learning; Pairwise comparison; Leverage (statistics); Graph; Supervised learning; Artificial neural network; Theoretical computer science","score_opus":0.007098345358430243,"score_gpt":0.2367498723285093,"score_spread":0.22965152697007907,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4365444089","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.074355766,0.00059010985,0.91825646,0.00043828116,0.00006928784,0.00007344979,0.0002702661,0.003747143,0.002199213],"genre_scores_gemma":[0.750258,0.00033890494,0.24216157,0.00043732097,0.00012207654,0.00017244932,0.0016074713,0.00037444982,0.0045276685],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9993436,0.00016439345,0.000029131237,0.00022896325,0.00018007474,0.000053731143],"domain_scores_gemma":[0.9984054,0.00045080707,0.00024305536,0.0003287736,0.00049107894,0.00008094852],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00089864456,0.00076322793,0.0007851162,0.0004489708,0.0002841303,0.00042443868,0.00199065,0.0010114411,0.0012600302],"category_scores_gemma":[0.0022641767,0.00030590544,0.0005681875,0.0005305381,0.0008367299,0.0013330373,0.0009621685,0.0011380645,0.00070921576],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00016914975,0.0002204179,0.0017866379,0.00022794542,0.00006830124,0.00010915861,0.00008368435,0.78937906,0.015596426,0.005103167,0.0062157176,0.18104038],"study_design_scores_gemma":[0.0000036248066,0.000012733293,0.00009492494,0.0000019051397,0.0000022130714,0.000011782977,0.0000022920244,0.9960372,0.0019469653,0.0016507724,0.00023306892,0.000002660644],"about_ca_topic_score_codex":0.0019038011,"about_ca_topic_score_gemma":0.0023127187,"teacher_disagreement_score":0.00199065,"about_ca_system_score_codex":0.00070495193,"about_ca_system_score_gemma":0.0007445875,"threshold_uncertainty_score":0.0051148534},"labels":[],"label_agreement":null},{"id":"W4366351318","doi":"10.1093/bioinformatics/btad255","title":"ConsAlign: simultaneous RNA structural aligner based on rich transfer learning and thermodynamic ensemble model of alignment scoring","year":2023,"lang":"en","type":"article","venue":"Bioinformatics","topic":"RNA and protein synthesis mechanisms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Institute of Genetics","keywords":"Computer science; Ensemble learning; Artificial intelligence; Software; Machine learning; Programming language","score_opus":0.01284406087212039,"score_gpt":0.23595095961090834,"score_spread":0.22310689873878795,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4366351318","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.03056417,0.0004998457,0.841572,0.00033803456,0.00022580213,0.00022855315,0.0025838593,0.12040986,0.0035779492],"genre_scores_gemma":[0.25222066,0.00036898608,0.7158695,0.00047901797,0.00018691397,0.0006795964,0.011377348,0.013021028,0.005796853],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9990336,0.0002547241,0.0000464443,0.00030236717,0.00029136136,0.00007164884],"domain_scores_gemma":[0.9988361,0.00049740827,0.000098677745,0.00024204227,0.0002183646,0.00010741688],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0019203878,0.0015885865,0.0016434579,0.0011880152,0.0011562449,0.0011680329,0.0028527796,0.0016412156,0.01159216],"category_scores_gemma":[0.005093836,0.0009694826,0.0012612178,0.0011009014,0.0006566829,0.0025775393,0.002016506,0.0029708927,0.005852748],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011016399,0.0005720789,0.009070553,0.0010768964,0.0005573848,0.0006547049,0.00034212324,0.33408356,0.038548417,0.030252056,0.121611476,0.4621291],"study_design_scores_gemma":[0.000052566993,0.00006804855,0.00038854164,0.000013740685,0.000014489933,0.00010563182,0.000015157396,0.9803115,0.0067628096,0.007933044,0.0043041366,0.000030286086],"about_ca_topic_score_codex":0.0026783478,"about_ca_topic_score_gemma":0.004712839,"teacher_disagreement_score":0.01159216,"about_ca_system_score_codex":0.0008521582,"about_ca_system_score_gemma":0.0017673125,"threshold_uncertainty_score":0.038779616},"labels":[],"label_agreement":null},{"id":"W4366549780","doi":"10.1093/bioinformatics/btad180","title":"The DynaSig-ML Python package: automated learning of biomolecular dynamics–function relationships","year":2023,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Protein Structure and Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal; Institute for Research in Immunology and Cancer","funders":"Natural Sciences and Engineering Research Council of Canada; Canadian Institutes of Health Research; Compute Canada; Genome Canada","keywords":"Python (programming language); Computer science; R package; Programming language; Artificial intelligence","score_opus":0.00876049196145396,"score_gpt":0.23710940974201583,"score_spread":0.22834891778056188,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4366549780","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0060489275,0.0004290821,0.4489043,0.00055953703,0.00022347772,0.00021772848,0.039310463,0.49836722,0.005939272],"genre_scores_gemma":[0.088505365,0.0010739323,0.63440585,0.0016813477,0.00018757871,0.0024903344,0.07492243,0.18294312,0.013789944],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99919564,0.00016604419,0.000054581433,0.00020126924,0.0002814918,0.00010098395],"domain_scores_gemma":[0.9986528,0.00068219035,0.00012312029,0.00026050577,0.00016714749,0.00011420644],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00188714,0.0021850348,0.0019089553,0.0012797066,0.00089732476,0.0019583232,0.0036233065,0.0011876549,0.053746305],"category_scores_gemma":[0.005682231,0.0015256831,0.0022496297,0.0010055868,0.00095814344,0.002247058,0.003258953,0.0041823452,0.03103147],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005586158,0.00028016572,0.005684472,0.0033911292,0.0007846105,0.0006060711,0.0004333665,0.06508208,0.015928295,0.034522388,0.7272296,0.14549926],"study_design_scores_gemma":[0.00039506174,0.00009736841,0.0032748193,0.000220617,0.00008939978,0.0004659162,0.00007252711,0.70772886,0.019708892,0.06675479,0.20092568,0.0002661697],"about_ca_topic_score_codex":0.0035528871,"about_ca_topic_score_gemma":0.0061569763,"teacher_disagreement_score":0.053746305,"about_ca_system_score_codex":0.0010019495,"about_ca_system_score_gemma":0.0030784488,"threshold_uncertainty_score":0.17979932},"labels":[],"label_agreement":null},{"id":"W4366721767","doi":"10.1093/bioinformatics/btad276","title":"spongEffects: ceRNA modules offer patient-specific insights into the miRNA regulatory landscape","year":2023,"lang":"en","type":"article","venue":"Bioinformatics","topic":"MicroRNA in disease regulation","field":"Biochemistry, Genetics and Molecular Biology","cited_by":14,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Genome British Columbia","funders":"Bundesministerium für Bildung und Forschung; Deutsche Forschungsgemeinschaft; European Commission","keywords":"microRNA; Computational biology; Computer science; Competing endogenous RNA; Data science; Biology; Gene; Genetics","score_opus":0.007984792714195431,"score_gpt":0.21192679396232234,"score_spread":0.2039420012481269,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4366721767","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.18695882,0.0026066827,0.67556113,0.002140257,0.00038911423,0.0003020522,0.06417423,0.05745191,0.010415799],"genre_scores_gemma":[0.7250168,0.001028769,0.22227685,0.0010756676,0.00021347543,0.00053465593,0.039342515,0.005201723,0.0053096036],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9994923,0.00014306212,0.000034392357,0.00019828956,0.00009016656,0.00004166491],"domain_scores_gemma":[0.99812704,0.0012098594,0.00022737385,0.00019191057,0.000120811485,0.00012309116],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014866005,0.0009838889,0.00078860996,0.0014232425,0.0004141742,0.0013643547,0.00079362176,0.00079376163,0.012750883],"category_scores_gemma":[0.008402983,0.0005205463,0.0013235814,0.000798205,0.00050220196,0.00076493504,0.0011461016,0.0009546112,0.0031166612],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00449167,0.00039383845,0.27362797,0.0035010965,0.0024517255,0.0018087903,0.00070640136,0.114667945,0.107013114,0.028186604,0.12602982,0.337121],"study_design_scores_gemma":[0.00057180616,0.00075331645,0.06445842,0.00042763838,0.0009978579,0.0030547823,0.00020205122,0.6457755,0.095635295,0.102903426,0.08496616,0.00025377652],"about_ca_topic_score_codex":0.0014419917,"about_ca_topic_score_gemma":0.003156817,"teacher_disagreement_score":0.012750883,"about_ca_system_score_codex":0.0004023446,"about_ca_system_score_gemma":0.00105548,"threshold_uncertainty_score":0.042656004},"labels":[],"label_agreement":null},{"id":"W4366992370","doi":"10.1093/bioinformatics/btad163","title":"Correction to: Continuous chromatin state feature annotation of the human epigenome","year":2023,"lang":"en","type":"erratum","venue":"Bioinformatics","topic":"Machine Learning in Bioinformatics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Natural Sciences and Engineering Research Council of Canada; Michael Smith Health Research BC; Compute Canada; Genome Canada","keywords":"Epigenome; Chromatin; Feature (linguistics); Annotation; Computer science; State (computer science); Computational biology; Artificial intelligence; Biology; Genetics; DNA methylation; Programming language; DNA; Gene","score_opus":0.006974353029127545,"score_gpt":0.2547317770100916,"score_spread":0.24775742398096404,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4366992370","genre_codex":"editorial","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00027641415,0.00038924738,0.0016929223,0.038907684,0.9513853,0.000029388897,0.0036784904,0.0010003456,0.0026401056],"genre_scores_gemma":[0.03294525,0.004792263,0.019513778,0.11174198,0.39156368,0.00048287783,0.013687528,0.0071579795,0.41811466],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9973673,0.00038527747,0.0005118238,0.00044728955,0.0010129038,0.00027539238],"domain_scores_gemma":[0.9716015,0.008070424,0.0011430888,0.002525711,0.014754261,0.0019050373],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0025485707,0.0019471924,0.0020303687,0.0037222959,0.0028970425,0.0033759868,0.0032128417,0.0054797037,0.11828151],"category_scores_gemma":[0.053896952,0.0013370902,0.0013070462,0.002383221,0.002381503,0.0019222128,0.0023656208,0.009276669,0.05697017],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000062512576,0.000008858224,0.000097724354,0.000098418604,0.000013485348,0.0003562316,0.000019759009,0.000057579564,0.00009090167,0.0005964261,0.993457,0.005141134],"study_design_scores_gemma":[0.00011490882,0.00003173353,0.0017270268,0.00023178507,0.00004962205,0.0015018276,0.000093734605,0.00063271495,0.0009484124,0.003583711,0.99102306,0.00006140018],"about_ca_topic_score_codex":0.015922964,"about_ca_topic_score_gemma":0.017952653,"teacher_disagreement_score":0.11828151,"about_ca_system_score_codex":0.00275539,"about_ca_system_score_gemma":0.0038283116,"threshold_uncertainty_score":0.3956911},"labels":[],"label_agreement":null},{"id":"W4367836399","doi":"10.1093/bioinformatics/btad210","title":"Signed Distance Correlation (SiDCo): an online implementation of distance correlation and partial distance correlation for data-driven network analysis","year":2023,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Metabolomics and Mass Spectrometry Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":15,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Ottawa; National Research Council Canada","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Pearson product-moment correlation coefficient; Distance correlation; Partial correlation; Correlation; Covariance; Covariance and correlation; Linear correlation; Software; Computer science; Correlation coefficient; Distance measures; Data mining; Linear model; Mathematics; Statistics; Artificial intelligence; Random variable","score_opus":0.027772411800048722,"score_gpt":0.30760081987860377,"score_spread":0.279828408078555,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4367836399","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0036064417,0.00038952281,0.88692755,0.0002792285,0.00026871767,0.00026339685,0.011152174,0.09374968,0.0033632296],"genre_scores_gemma":[0.05775129,0.00064193574,0.8863341,0.0005223486,0.00012860172,0.0015588012,0.03018224,0.016611356,0.0062693167],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99653095,0.00089730206,0.00034993934,0.00068077986,0.0013499768,0.00019105902],"domain_scores_gemma":[0.99115527,0.004372304,0.00096867426,0.0012862557,0.0018560416,0.00036140304],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0050074253,0.0024077639,0.001343543,0.0033034475,0.0009722641,0.0032429027,0.0025128806,0.0014311547,0.036823384],"category_scores_gemma":[0.024530424,0.0011852962,0.0020529588,0.0033761135,0.00082272926,0.0034414923,0.0034024897,0.002640348,0.014062524],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009586817,0.00039153453,0.014913322,0.003673319,0.0009992579,0.00092083437,0.00089861243,0.054296616,0.023380231,0.07934293,0.30782014,0.51240456],"study_design_scores_gemma":[0.00030046742,0.000331729,0.006798025,0.00056382286,0.00019671788,0.0012862039,0.00027377182,0.552731,0.03859706,0.14314507,0.2553212,0.0004548867],"about_ca_topic_score_codex":0.003467817,"about_ca_topic_score_gemma":0.0059972303,"teacher_disagreement_score":0.036823384,"about_ca_system_score_codex":0.0010355283,"about_ca_system_score_gemma":0.0028078305,"threshold_uncertainty_score":0.12318653},"labels":[],"label_agreement":null},{"id":"W4375955770","doi":"10.1093/bioinformatics/btad309","title":"Genome mining for anti-CRISPR operons using machine learning","year":2023,"lang":"en","type":"article","venue":"Bioinformatics","topic":"CRISPR and Genetic Engineering","field":"Biochemistry, Genetics and Molecular Biology","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Institutes of Health; National Institute of General Medical Sciences; National Institute of Allergy and Infectious Diseases; University of Nebraska-Lincoln; University of Toronto; U.S. Department of Agriculture","keywords":"CRISPR; Genome; Operon; Computational biology; Context (archaeology); Computer science; Python (programming language); Gene; Biology; Genetics; Programming language","score_opus":0.02191280019472707,"score_gpt":0.3098302355894656,"score_spread":0.2879174353947385,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4375955770","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.3276218,0.004069872,0.5794847,0.0016372134,0.00015504217,0.0005220044,0.027504237,0.054463457,0.004541609],"genre_scores_gemma":[0.4249014,0.00065711595,0.5305088,0.00081032905,0.00006042097,0.0004528998,0.040646676,0.0008979295,0.0010643883],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9988285,0.00018539664,0.00012338306,0.00048822386,0.00027191185,0.000102592574],"domain_scores_gemma":[0.9969977,0.0018002625,0.00047387194,0.00018586068,0.00043990265,0.00010246627],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014783512,0.0014756935,0.001387012,0.003932244,0.00070738816,0.00118716,0.0018505809,0.0013160958,0.0019054555],"category_scores_gemma":[0.005646828,0.00060515886,0.0019191643,0.0023259039,0.00041587002,0.00085328677,0.000937642,0.0012413622,0.0012633643],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013269931,0.0013261165,0.107686415,0.0040695444,0.0009419658,0.0018044197,0.00048760424,0.21081993,0.088967755,0.0048558866,0.036440708,0.5412726],"study_design_scores_gemma":[0.00008854276,0.00021788945,0.007629079,0.00014166605,0.00017638881,0.0005420463,0.00014479877,0.9365047,0.036808588,0.008148106,0.0095456755,0.000052459698],"about_ca_topic_score_codex":0.002728114,"about_ca_topic_score_gemma":0.0028498715,"teacher_disagreement_score":0.003932244,"about_ca_system_score_codex":0.0007765738,"about_ca_system_score_gemma":0.001296531,"threshold_uncertainty_score":0.007818341},"labels":[],"label_agreement":null},{"id":"W4377086801","doi":"10.1093/bioinformatics/btad328","title":"GIL: a python package for designing custom indexing primers","year":2023,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Algorithms and Data Compression","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"Natural Sciences and Engineering Research Council of Canada; University of British Columbia; Western Canada Research Grid; Stem Cell Network; Canadian Institutes of Health Research; Compute Canada","keywords":"Python (programming language); Search engine indexing; Programming language; Computer science; R package; Information retrieval","score_opus":0.03511654673081748,"score_gpt":0.27567387795689735,"score_spread":0.24055733122607986,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4377086801","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0027940895,0.00045917023,0.38707474,0.0003529647,0.00037481976,0.0005444239,0.07480764,0.5257096,0.007882605],"genre_scores_gemma":[0.02003827,0.00057997956,0.6616293,0.0018648304,0.00017121629,0.0037127275,0.10013275,0.19496727,0.016903637],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9976471,0.0003665561,0.00030384306,0.00069950084,0.00070119195,0.00028181108],"domain_scores_gemma":[0.997265,0.0012945719,0.00028017463,0.0004997454,0.0004749397,0.00018557766],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003295685,0.0025025597,0.0018788644,0.001977601,0.0013047627,0.002957051,0.0037199764,0.0011011903,0.10738149],"category_scores_gemma":[0.0098484745,0.0024744926,0.0019220219,0.0017726287,0.0009301091,0.002436275,0.0030222652,0.0036384265,0.08335165],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00101522,0.0001792398,0.0033286023,0.0036746748,0.00037066633,0.00039484704,0.0006104719,0.0060386886,0.04110224,0.013795565,0.7296231,0.19986655],"study_design_scores_gemma":[0.00039197085,0.00017650147,0.0032033448,0.00045435995,0.00017787596,0.00077126065,0.00016160366,0.04582868,0.090352476,0.03811494,0.81998676,0.00038021637],"about_ca_topic_score_codex":0.0022429724,"about_ca_topic_score_gemma":0.0034498512,"teacher_disagreement_score":0.10738149,"about_ca_system_score_codex":0.0011500753,"about_ca_system_score_gemma":0.0030387219,"threshold_uncertainty_score":0.35922688},"labels":[],"label_agreement":null},{"id":"W4378782144","doi":"10.1093/bioinformatics/btad353","title":"MRGCN: cancer subtyping with multi-reconstruction graph convolutional network using full and partial multi-omics dataset","year":2023,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":32,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"Xi'an Municipal Bureau of Science and Technology; Education Department of Shaanxi Province; Xi'an Polytechnic University; National Natural Science Foundation of China","keywords":"Subtyping; Omics; Computer science; Graph; Rank (graph theory); Computational biology; Data mining; Bioinformatics; Biology; Theoretical computer science; Mathematics","score_opus":0.02777623596967806,"score_gpt":0.264301826644408,"score_spread":0.23652559067472992,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4378782144","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.1766173,0.0051271953,0.6865561,0.0029089747,0.0005149185,0.000650029,0.07350388,0.0477761,0.0063454853],"genre_scores_gemma":[0.404412,0.0017590867,0.3996061,0.0012167125,0.00015747307,0.0007983197,0.18157138,0.0016685852,0.008810451],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9995708,0.000070932234,0.000017038046,0.00020372249,0.00007824215,0.00005919346],"domain_scores_gemma":[0.99958223,0.00011379487,0.000044518492,0.0001323028,0.000082878294,0.00004413617],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010275318,0.0018655999,0.0009415119,0.001809787,0.00070177065,0.0009731469,0.002203865,0.0014680407,0.0024516236],"category_scores_gemma":[0.002616281,0.00063186913,0.0017455716,0.0017685225,0.00048561595,0.0010485501,0.0014780492,0.0017198592,0.001128317],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008518718,0.00043420919,0.029309148,0.0010286644,0.0015131242,0.00087292085,0.00029803085,0.37971658,0.019407697,0.010827407,0.15740056,0.39833978],"study_design_scores_gemma":[0.00007564183,0.000053590076,0.0028549503,0.000039352533,0.0001282849,0.00014409723,0.00003701842,0.97467333,0.003278451,0.010230748,0.008442147,0.00004248682],"about_ca_topic_score_codex":0.051389653,"about_ca_topic_score_gemma":0.07527999,"teacher_disagreement_score":0.051389653,"about_ca_system_score_codex":0.0017920347,"about_ca_system_score_gemma":0.002466584,"threshold_uncertainty_score":0.10218108},"labels":[],"label_agreement":null},{"id":"W4380869524","doi":"10.1093/bioinformatics/btad381","title":"GRETTA: an R package for mapping <i>in silico</i> genetic interaction and essentiality networks","year":2023,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canada's Michael Smith Genome Sciences Centre; Genome British Columbia; University of British Columbia","funders":"University of British Columbia; University of Alabama at Birmingham; Canada Research Chairs; Broad Institute; University of Alabama; Canadian Institutes of Health Research; Canada's Michael Smith Genome Sciences Centre","keywords":"DECIPHER; In silico; Computer science; MIT License; Documentation; Computational biology; Source code; Biology; Genetics; Gene; Software; Programming language","score_opus":0.017506506091304393,"score_gpt":0.26022353994634445,"score_spread":0.24271703385504007,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4380869524","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.002966682,0.0013895783,0.32056713,0.001041522,0.0006343251,0.00031119198,0.1704985,0.49568903,0.006902],"genre_scores_gemma":[0.039376054,0.0017540712,0.37226734,0.0020752146,0.00034770858,0.003020831,0.23872651,0.32974255,0.012689756],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9974968,0.00086345605,0.00021303643,0.0007327344,0.00048130393,0.00021261466],"domain_scores_gemma":[0.990928,0.0061932253,0.0007260296,0.0011448157,0.00068069785,0.00032721565],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004414839,0.0038010927,0.0030293511,0.0026478386,0.0010076282,0.0040506218,0.0041185003,0.0016280869,0.10408506],"category_scores_gemma":[0.021846207,0.0020707548,0.0043977825,0.0021467877,0.0011077013,0.0024011917,0.0036673432,0.0048721544,0.08599849],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00059395557,0.000068088244,0.0056265276,0.004475769,0.0016941219,0.0006018305,0.00044513866,0.013395277,0.005769528,0.01670819,0.8892944,0.061327316],"study_design_scores_gemma":[0.00067335967,0.00018757221,0.0072729364,0.0008636841,0.0008449765,0.0012524512,0.0001721788,0.07225652,0.009864274,0.09361846,0.8126004,0.00039314345],"about_ca_topic_score_codex":0.0041882438,"about_ca_topic_score_gemma":0.0054261405,"teacher_disagreement_score":0.10408506,"about_ca_system_score_codex":0.00090835424,"about_ca_system_score_gemma":0.0036433756,"threshold_uncertainty_score":0.3481992},"labels":[],"label_agreement":null},{"id":"W4380871602","doi":"10.1093/bioinformatics/btad390","title":"Interpretable deep learning architectures for improving drug response prediction performance: myth or reality?","year":2023,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Computational Drug Discovery Methods","field":"Computer Science","cited_by":30,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Mila - Quebec Artificial Intelligence Institute; McGill University","funders":"Natural Sciences and Engineering Research Council of Canada; Institut de Valorisation des Données; Merck Canada; Canada First Research Excellence Fund; Government of Canada; Génome Québec; Compute Canada","keywords":"Interpretability; Computer science; Generalizability theory; Machine learning; Artificial intelligence; Baseline (sea); Random forest; Deep learning; Overfitting; Perceptron; Predictive modelling; Data mining; Artificial neural network; Statistics; Mathematics","score_opus":0.02475005689958606,"score_gpt":0.29803671581054364,"score_spread":0.2732866589109576,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4380871602","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.19995032,0.023406858,0.70082635,0.03402026,0.0012747874,0.00033180116,0.011867143,0.01646332,0.0118591795],"genre_scores_gemma":[0.7945546,0.007183098,0.18182859,0.002926288,0.00044566466,0.00025085895,0.008670107,0.00042959064,0.0037111181],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.998648,0.0005957063,0.00009786674,0.0002582991,0.00030088326,0.00009935238],"domain_scores_gemma":[0.99394786,0.0035051682,0.00047815114,0.000912226,0.0009516928,0.00020495245],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005071693,0.001974602,0.0007660084,0.00070464873,0.0002972324,0.0018000323,0.0022917413,0.0017186162,0.004152864],"category_scores_gemma":[0.015892873,0.0004651106,0.00071416347,0.00080223684,0.0010574935,0.0035171795,0.0015379296,0.0042697894,0.0016721094],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009329826,0.0005678304,0.012835109,0.0015996706,0.0003925985,0.0002744586,0.000246138,0.60171497,0.010133014,0.017107494,0.039275616,0.31492013],"study_design_scores_gemma":[0.000067796194,0.0002012263,0.001560022,0.0002764894,0.00009720891,0.0000929453,0.000055594308,0.9504724,0.007776568,0.03326858,0.006080751,0.000050408962],"about_ca_topic_score_codex":0.004522501,"about_ca_topic_score_gemma":0.0061433534,"teacher_disagreement_score":0.005071693,"about_ca_system_score_codex":0.0012507174,"about_ca_system_score_gemma":0.0016400348,"threshold_uncertainty_score":0.02682203},"labels":[],"label_agreement":null},{"id":"W4382631279","doi":"10.1093/bioinformatics/btad253","title":"Robust reconstruction of single-cell RNA-seq data with iterative gene weight updates","year":2023,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Single-cell and spatial transcriptomics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Defense Advanced Research Projects Agency; Azrieli Foundation; Israel Science Foundation; European Commission; U.S. Department of Energy; National Science Foundation","keywords":"RNA-Seq; Computer science; Computational biology; Gene; Algorithm; Gene expression; Data mining; Biology; Transcriptome; Genetics","score_opus":0.03678924505138624,"score_gpt":0.2210204430184372,"score_spread":0.18423119796705095,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4382631279","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.019676108,0.00009332005,0.9769372,0.00014578714,0.000025857287,0.000040118655,0.00021143234,0.0024536396,0.00041663545],"genre_scores_gemma":[0.09619861,0.0001032936,0.8999949,0.0001428203,0.00002146414,0.00018557365,0.0014762474,0.0007906615,0.0010863419],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9993899,0.00016772516,0.000032593427,0.0001663201,0.00019218306,0.000051268198],"domain_scores_gemma":[0.9980573,0.00108561,0.0001457773,0.00025230923,0.00037416167,0.00008478353],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0019600284,0.0012749692,0.0009666834,0.00092832383,0.0005721811,0.0010601999,0.0018170816,0.0013140272,0.0021633962],"category_scores_gemma":[0.007056511,0.00083188206,0.0011591315,0.0011468383,0.0009882725,0.0010476665,0.0015218366,0.0018080858,0.0015925636],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003305457,0.00012362585,0.005018866,0.00033117647,0.00018823639,0.0002788417,0.00036895653,0.77406794,0.059776172,0.009554802,0.005762107,0.14419886],"study_design_scores_gemma":[0.000013494875,0.000014547338,0.00027951354,0.0000059870245,0.0000075517764,0.00003703403,0.000018790462,0.9860234,0.00846693,0.0042721853,0.00084881764,0.000011608308],"about_ca_topic_score_codex":0.005257445,"about_ca_topic_score_gemma":0.008282557,"teacher_disagreement_score":0.005257445,"about_ca_system_score_codex":0.0009603364,"about_ca_system_score_gemma":0.0016625734,"threshold_uncertainty_score":0.010453701},"labels":[],"label_agreement":null},{"id":"W4382631507","doi":"10.1093/bioinformatics/btad250","title":"PlasBin-flow: a flow-based MILP algorithm for plasmid contigs binning","year":2023,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"Natural Sciences and Engineering Research Council of Canada; Vedecká Grantová Agentúra MŠVVaŠ SR a SAV; European Commission; Genome Canada","keywords":"Contig; Plasmid; Genetics; Biology; Computer science; Computational biology; Sequence assembly; Genome; Algorithm; DNA; Gene","score_opus":0.01611820032584436,"score_gpt":0.24090628305546335,"score_spread":0.224788082729619,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4382631507","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.007458732,0.0003391424,0.97935504,0.00024403646,0.000089896195,0.00018421205,0.0012073994,0.00884126,0.0022802684],"genre_scores_gemma":[0.061416186,0.00019589091,0.9286713,0.00022061901,0.000057918773,0.00051533827,0.004707208,0.0020047955,0.0022107554],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9993963,0.00015122832,0.00003768631,0.0001827728,0.00014269726,0.00008928923],"domain_scores_gemma":[0.998292,0.00118014,0.00011279144,0.00010672454,0.00021230074,0.00009598904],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016504201,0.0024661194,0.0014080248,0.0016334062,0.0009415882,0.0017577083,0.001859654,0.0016523013,0.012920599],"category_scores_gemma":[0.004540192,0.0011388579,0.0017650393,0.0015754566,0.00088432746,0.0015560086,0.0019869239,0.0024820475,0.0028411285],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00022501094,0.00018500337,0.0016523385,0.0003958039,0.00009018314,0.00014792044,0.0001263311,0.823683,0.0033584775,0.009567188,0.016590757,0.14397794],"study_design_scores_gemma":[0.00004170712,0.000026877546,0.0000898013,0.000020806192,0.000008888872,0.000020439169,0.000020307147,0.9866933,0.0008708005,0.009198483,0.0030000862,0.000008511178],"about_ca_topic_score_codex":0.007395631,"about_ca_topic_score_gemma":0.009456426,"teacher_disagreement_score":0.012920599,"about_ca_system_score_codex":0.0011902744,"about_ca_system_score_gemma":0.002828695,"threshold_uncertainty_score":0.04322368},"labels":[],"label_agreement":null},{"id":"W4382631688","doi":"10.1093/bioinformatics/btad249","title":"PPAD: a deep learning architecture to predict progression of Alzheimer’s disease","year":2023,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Machine Learning in Healthcare","field":"Computer Science","cited_by":24,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Canadian Institutes of Health Research; National Institutes of Health; Genentech; IXICO; H. Lundbeck A/S; Servier; Eisai; National Institute of General Medical Sciences; Northern California Institute for Research and Education; University of North Texas; Pfizer; Novartis Pharmaceuticals Corporation; University of Southern California; Biogen; Eli Lilly and Company; Bristol-Myers Squibb; BioClinica; U.S. Department of Defense; Alzheimer's Disease Neuroimaging Initiative; Meso Scale Diagnostics; National Institute on Aging; Alzheimer's Association","keywords":"Autoencoder; Dementia; Recurrent neural network; Disease; Alzheimer's disease; Neuroimaging; Artificial intelligence; Deep learning; Cognitive impairment; Electronic health record; Medicine; Computer science; Machine learning; Gerontology; Artificial neural network; Psychiatry; Health care; Internal medicine","score_opus":0.022478761562064423,"score_gpt":0.31194538166756747,"score_spread":0.28946662010550306,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4382631688","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.24090214,0.009255182,0.7263327,0.00286396,0.000923246,0.00023514892,0.00462234,0.007879072,0.0069861584],"genre_scores_gemma":[0.89677566,0.0018662879,0.08945841,0.00068188703,0.00016098605,0.00018986348,0.0043489416,0.00006170723,0.0064562326],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9997676,0.000049377417,0.000018560466,0.000080460224,0.000043943997,0.000039960174],"domain_scores_gemma":[0.9995426,0.00016909985,0.000033835066,0.000031592324,0.00018718823,0.000035673584],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008947563,0.0011221177,0.0005447535,0.00060464046,0.00026868054,0.00052926404,0.0012633685,0.00095553976,0.0013247208],"category_scores_gemma":[0.0017926139,0.00036671592,0.0005446979,0.00046616525,0.00025485986,0.0008924136,0.0007480561,0.0018702246,0.00043763279],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0004609286,0.00045959553,0.015231145,0.00020818738,0.00027695502,0.00024459022,0.00006668743,0.63329816,0.0033254721,0.0019062536,0.014501971,0.33002],"study_design_scores_gemma":[0.000010001937,0.00004834746,0.00070332334,0.000011891819,0.000018308812,0.000022331486,0.0000046767886,0.997083,0.0005968489,0.0010211018,0.00047403024,0.0000061477276],"about_ca_topic_score_codex":0.016136384,"about_ca_topic_score_gemma":0.015626395,"teacher_disagreement_score":0.016136384,"about_ca_system_score_codex":0.00078661577,"about_ca_system_score_gemma":0.0010426085,"threshold_uncertainty_score":0.032084882},"labels":[],"label_agreement":null},{"id":"W4382631698","doi":"10.1093/bioinformatics/btad266","title":"Reference panel-guided super-resolution inference of Hi-C data","year":2023,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"Fonds de recherche du Québec – Nature et technologies; Institut de Valorisation des Données","keywords":"Computer science; Inference; Focus (optics); Data mining; DNA sequencing; Chromatin; Task (project management); Computational biology; Resolution (logic); Genome; Artificial intelligence; DNA; Biology; Gene; Genetics; Physics","score_opus":0.08873438610513366,"score_gpt":0.3032298105770997,"score_spread":0.214495424471966,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4382631698","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.090626076,0.0028976514,0.8489328,0.0009692041,0.00018334339,0.00019637168,0.023237707,0.028498992,0.004457829],"genre_scores_gemma":[0.36136934,0.0010666967,0.53523606,0.0014529116,0.00015610585,0.00049429433,0.09326652,0.002379177,0.0045789126],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9988306,0.00028206548,0.000046699904,0.00046746095,0.00023867008,0.0001345706],"domain_scores_gemma":[0.99596524,0.0017309103,0.00023276753,0.00091021776,0.0009783062,0.0001825143],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0039180247,0.0012618166,0.0011397022,0.0021028875,0.00090623536,0.0016543689,0.0039286823,0.0018925536,0.0051814923],"category_scores_gemma":[0.011637211,0.00078469544,0.0011675768,0.0015942901,0.0007707826,0.0017541234,0.0025146322,0.0022336715,0.0026963814],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0015842599,0.0005584027,0.059804693,0.0019734607,0.0010260153,0.00087305694,0.0005303343,0.37679127,0.08259871,0.015727034,0.12542619,0.33310655],"study_design_scores_gemma":[0.00006623067,0.000087805,0.009313574,0.000108961445,0.000116714094,0.00026796243,0.00008589952,0.93648905,0.0250754,0.014121269,0.014200028,0.000067105],"about_ca_topic_score_codex":0.015242722,"about_ca_topic_score_gemma":0.026975794,"teacher_disagreement_score":0.015242722,"about_ca_system_score_codex":0.0011675345,"about_ca_system_score_gemma":0.001808468,"threshold_uncertainty_score":0.030308008},"labels":[],"label_agreement":null},{"id":"W4382632303","doi":"10.1093/bioinformatics/btad242","title":"SpatialSort: a Bayesian model for clustering and cell population annotation of spatial proteomics data","year":2023,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Single-cell and spatial transcriptomics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia; Terry Fox Research Institute; BC Cancer Agency","funders":"Natural Sciences and Engineering Research Council of Canada; Canadian Institutes of Health Research; Cancer Research UK; BC Cancer Foundation; Terry Fox Research Institute; Michael Smith Health Research BC; V Foundation for Cancer Research","keywords":"Cluster analysis; Computer science; Annotation; Bayesian probability; Data mining; Spatial analysis; Source code; Profiling (computer programming); Population; Context (archaeology); Consensus clustering; Artificial intelligence; Machine learning; Fuzzy clustering; CURE data clustering algorithm; Geography; Mathematics; Statistics","score_opus":0.037640320235041845,"score_gpt":0.2650842334166531,"score_spread":0.22744391318161125,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4382632303","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0019710741,0.00012905955,0.9935777,0.00023109697,0.000029364048,0.00007601868,0.0011954407,0.002440589,0.00034959355],"genre_scores_gemma":[0.07036674,0.00041219092,0.91436493,0.00051085476,0.00014838322,0.0007958186,0.0082342215,0.0015876956,0.0035791283],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9973394,0.0010239435,0.00014343343,0.0007053644,0.0006253955,0.00016244862],"domain_scores_gemma":[0.995533,0.0024867111,0.0003140068,0.0005426974,0.00090485247,0.00021861332],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005677397,0.0013294333,0.0018943695,0.0021676836,0.0012957705,0.002581895,0.0055636796,0.0024394975,0.0060634157],"category_scores_gemma":[0.013100147,0.0017388891,0.0026101265,0.0029682359,0.0015763043,0.0024310274,0.0025997297,0.00334809,0.0041459072],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0004906097,0.00010761458,0.0031367415,0.00037847925,0.00030356503,0.0001691656,0.00038040092,0.7831138,0.005190328,0.04276578,0.026959691,0.13700388],"study_design_scores_gemma":[0.000026183494,0.0000140196,0.00022203012,0.0000181537,0.000012817359,0.00003544857,0.000017957387,0.97017044,0.0009699006,0.025082558,0.0034096185,0.000020849953],"about_ca_topic_score_codex":0.024013707,"about_ca_topic_score_gemma":0.040740605,"teacher_disagreement_score":0.024013707,"about_ca_system_score_codex":0.002923201,"about_ca_system_score_gemma":0.0034235823,"threshold_uncertainty_score":0.04774785},"labels":[],"label_agreement":null},{"id":"W4383186166","doi":"10.1093/bioinformatics/btad423","title":"MEDIPIPE: an automated and comprehensive pipeline for cfMeDIP-seq data quality control and analysis","year":2023,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Epigenetics and DNA Methylation","field":"Biochemistry, Genetics and Molecular Biology","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ontario Institute for Cancer Research; University of Toronto; Princess Margaret Cancer Centre; University Health Network","funders":"Ontario Institute for Cancer Research; Cancer Research Institute","keywords":"Computer science; MIT License; Pipeline (software); Software; Bioconductor; Data mining; Source code; Operating system; Biology","score_opus":0.05855120257903918,"score_gpt":0.37471489919629397,"score_spread":0.3161636966172548,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4383186166","genre_codex":"software","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.005772624,0.0007531327,0.40164733,0.00077644415,0.00038092758,0.00083504786,0.0682809,0.5184281,0.0031254506],"genre_scores_gemma":[0.03041006,0.0006745561,0.71638834,0.0015961643,0.0002081734,0.0038308436,0.15219766,0.08861941,0.006074793],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.996436,0.0005306236,0.0003907795,0.0012662548,0.0010326131,0.00034374557],"domain_scores_gemma":[0.9952413,0.0020154307,0.00044333722,0.0008718026,0.0009882738,0.00043985696],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008339934,0.0032331913,0.002224024,0.0034293756,0.0023130786,0.003741586,0.004097681,0.001783184,0.04008136],"category_scores_gemma":[0.012948074,0.0028693161,0.0027794153,0.002273109,0.001234428,0.0026989486,0.0045374106,0.005507341,0.029380104],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0029607024,0.00022861999,0.007218957,0.0037278186,0.001213658,0.0007238425,0.0013980386,0.006819497,0.12809849,0.0069301585,0.6617325,0.17894764],"study_design_scores_gemma":[0.0012804253,0.00041273318,0.019515358,0.00063325925,0.00054442603,0.0012338216,0.00039959737,0.15392986,0.23471706,0.03110093,0.55502874,0.0012037561],"about_ca_topic_score_codex":0.004347923,"about_ca_topic_score_gemma":0.007460725,"teacher_disagreement_score":0.04008136,"about_ca_system_score_codex":0.0015156218,"about_ca_system_score_gemma":0.0049344627,"threshold_uncertainty_score":0.13408548},"labels":[],"label_agreement":null},{"id":"W4385237450","doi":"10.1093/bioinformatics/btad457","title":"LegNet: a best-in-class deep learning model for short DNA regulatory regions","year":2023,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":46,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"Ministry of Science and Higher Education of the Russian Federation; Russian Science Foundation","keywords":"Computer science; Gene regulatory network; Python (programming language); Computational biology; Scripting language; Deep learning; Artificial intelligence; Gene; Data mining; Gene expression; Biology; Genetics; Programming language","score_opus":0.0159666890867824,"score_gpt":0.2420830472253247,"score_spread":0.2261163581385423,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4385237450","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.07698023,0.0018540517,0.88671803,0.0013533701,0.00030042254,0.00013465346,0.0074208267,0.016272368,0.0089661],"genre_scores_gemma":[0.5832608,0.0014681019,0.35721537,0.001462803,0.00014998621,0.00059937756,0.025746293,0.002396159,0.027701056],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9998553,0.000024991394,0.00000576604,0.000059809954,0.000032807144,0.000021381207],"domain_scores_gemma":[0.99979657,0.00010038961,0.00001759942,0.000023417057,0.000041557138,0.00002041251],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000548226,0.0010701714,0.0004892796,0.00047304737,0.00030563085,0.0007621443,0.0017682204,0.0012382357,0.0041018166],"category_scores_gemma":[0.0013745113,0.00044757134,0.0006211033,0.00048443835,0.0005183,0.0009903574,0.00072439655,0.0014946815,0.0015339361],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00037847913,0.00017376822,0.0018039766,0.00021858698,0.00010048363,0.00017161475,0.000075889824,0.8066053,0.014730285,0.017009791,0.026130015,0.13260187],"study_design_scores_gemma":[0.000013099748,0.00002437493,0.00009532201,0.000010541273,0.000007007271,0.000015611116,0.0000040100463,0.9878507,0.0028369692,0.006630015,0.002506143,0.0000062207387],"about_ca_topic_score_codex":0.008062228,"about_ca_topic_score_gemma":0.015767636,"teacher_disagreement_score":0.008062228,"about_ca_system_score_codex":0.0011010257,"about_ca_system_score_gemma":0.0010620478,"threshold_uncertainty_score":0.01603055},"labels":[],"label_agreement":null},{"id":"W4385453913","doi":"10.1093/bioinformatics/btad477","title":"<i>USNAP</i>: fast unique dense region detection and its application to lung cancer","year":2023,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Princess Margaret Cancer Centre; University of Toronto; Discovery Centre; University Health Network","funders":"Natural Sciences and Engineering Research Council of Canada; National Science Foundation","keywords":"Scalability; Computer science; Time complexity; Graph; Theoretical computer science; Dynamic programming; Algorithm","score_opus":0.008048395387785028,"score_gpt":0.24074019103440478,"score_spread":0.23269179564661974,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4385453913","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.039860863,0.0022466406,0.92666346,0.00412735,0.0006190998,0.00033178538,0.002971702,0.018397432,0.004781761],"genre_scores_gemma":[0.2456438,0.0012425013,0.7383824,0.0018243688,0.0005873203,0.00045868565,0.006924705,0.0012250365,0.0037112655],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9990502,0.00019242268,0.000055832086,0.0003273117,0.00029587382,0.000078378805],"domain_scores_gemma":[0.9975073,0.0012045274,0.00021743924,0.0003694497,0.00049648894,0.00020481556],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013977262,0.0010106305,0.0009017935,0.0017390341,0.0010480384,0.0010629729,0.0017839108,0.0012857695,0.001966744],"category_scores_gemma":[0.005796181,0.00044951978,0.0009164302,0.002182623,0.0012175489,0.0015812613,0.0018031999,0.0013871711,0.0010898946],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00047061,0.00022391322,0.011558992,0.0007662584,0.00036097752,0.0013620864,0.00057334686,0.21569665,0.03492149,0.02259763,0.17565753,0.53581053],"study_design_scores_gemma":[0.000042708045,0.00008953754,0.0022856626,0.000032307467,0.000037037582,0.0004812671,0.00010581618,0.9488181,0.013723028,0.018288914,0.016044619,0.000050966275],"about_ca_topic_score_codex":0.015947137,"about_ca_topic_score_gemma":0.016670018,"teacher_disagreement_score":0.015947137,"about_ca_system_score_codex":0.0009898476,"about_ca_system_score_gemma":0.001689586,"threshold_uncertainty_score":0.031708598},"labels":[],"label_agreement":null},{"id":"W4385897776","doi":"10.1093/bioinformatics/btad508","title":"<i>i</i>DeLUCS: a deep learning interactive tool for alignment-free clustering of DNA sequences","year":2023,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":15,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University; University of Waterloo","funders":"Natural Sciences and Engineering Research Council of Canada; Compute Canada","keywords":"Cluster analysis; Computer science; Metagenomics; DNA sequencing; Data mining; Artificial intelligence; Scalability; Genomics; Software; Set (abstract data type); Genome; Unsupervised learning; Identifier; Computational biology; Pattern recognition (psychology); Machine learning; DNA; Biology; Database; Genetics; Gene","score_opus":0.01482127274179808,"score_gpt":0.25078210115095595,"score_spread":0.23596082840915786,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4385897776","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0037561597,0.00018526561,0.49775916,0.00025002356,0.00013541493,0.00013310919,0.014535329,0.47915775,0.004087789],"genre_scores_gemma":[0.092682324,0.00041980756,0.7186749,0.0011084881,0.000109317916,0.0016419744,0.0834288,0.08844051,0.013493864],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9991491,0.0000998026,0.00007092537,0.00021988289,0.00037187655,0.00008853572],"domain_scores_gemma":[0.99868613,0.00047599702,0.000113652095,0.00028001974,0.00031105458,0.00013303354],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012714602,0.0018538206,0.00084372255,0.0015823187,0.00052036473,0.0013739726,0.0049391687,0.0013327573,0.054571614],"category_scores_gemma":[0.0043274527,0.0012935756,0.001330441,0.0012143662,0.0007362384,0.0018213841,0.0032574432,0.0024380195,0.017927911],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009372853,0.00017985317,0.0029594284,0.0015088932,0.00031483502,0.00055653916,0.00034010923,0.03368494,0.031643704,0.012528846,0.5914459,0.32389954],"study_design_scores_gemma":[0.00048614776,0.00013633643,0.0028666614,0.00031370597,0.00005882941,0.0005147214,0.000066079876,0.66265047,0.09557214,0.023140939,0.21390922,0.00028482603],"about_ca_topic_score_codex":0.005056207,"about_ca_topic_score_gemma":0.0063808067,"teacher_disagreement_score":0.054571614,"about_ca_system_score_codex":0.0011336235,"about_ca_system_score_gemma":0.0012798943,"threshold_uncertainty_score":0.1825602},"labels":[],"label_agreement":null},{"id":"W4385898657","doi":"10.1093/bioinformatics/btad498","title":"Adjusting for gene-specific covariates to improve RNA-seq analysis","year":2023,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Institute of Food and Agriculture; Genome Alberta; Genome Canada; U.S. Department of Agriculture","keywords":"Covariate; Estimator; Null (SQL); Type I and type II errors; Context (archaeology); Computer science; Null hypothesis; Variable (mathematics); Null model; Statistics; Multiple comparisons problem; Data mining; Mathematics; Biology","score_opus":0.026445728465342967,"score_gpt":0.28277570739909486,"score_spread":0.25632997893375187,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4385898657","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.015525033,0.00026592234,0.9795855,0.00044644834,0.00015638524,0.00010901812,0.0006446848,0.0028280453,0.0004388372],"genre_scores_gemma":[0.21069068,0.00021307051,0.7805694,0.0012947527,0.00027073454,0.00068314513,0.002555343,0.001960029,0.0017628738],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99078584,0.0050172852,0.00039034267,0.0022745233,0.0012414237,0.00029049884],"domain_scores_gemma":[0.9546925,0.032341145,0.0024918178,0.006580445,0.003082821,0.00081126945],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.027629625,0.0011426073,0.0013491779,0.0010809257,0.0011052492,0.002260332,0.0029551394,0.0015639442,0.004509326],"category_scores_gemma":[0.07125904,0.00066585455,0.0019946927,0.0014007177,0.0013166487,0.0018802698,0.0022867937,0.0032049182,0.001997451],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0019233578,0.0005692303,0.13405362,0.0020690835,0.0023197667,0.0006326673,0.0012814378,0.0939743,0.17025176,0.03547305,0.02299506,0.53445673],"study_design_scores_gemma":[0.00045013346,0.0007391586,0.050894175,0.0002499686,0.00077748904,0.0006929919,0.00022784596,0.7072288,0.086738616,0.11186065,0.03982859,0.00031162013],"about_ca_topic_score_codex":0.0017908571,"about_ca_topic_score_gemma":0.0039933342,"teacher_disagreement_score":0.027629625,"about_ca_system_score_codex":0.0009064871,"about_ca_system_score_gemma":0.0022305949,"threshold_uncertainty_score":0.1461212},"labels":[],"label_agreement":null},{"id":"W4386084463","doi":"10.1093/bioinformatics/btad523","title":"metGWAS 1.0: an R workflow for network-driven over-representation analysis between independent metabolomic and meta-genome-wide association studies","year":2023,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Metabolomics and Mass Spectrometry Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Toronto General Hospital; University of Toronto; Canadian Institute for Advanced Research","funders":"National Heart, Lung, and Blood Institute; Banting and Best Diabetes Centre, University of Toronto; Natural Sciences and Engineering Research Council of Canada; Diabetes Canada; Canadian Institutes of Health Research; U.S. Department of Veterans Affairs","keywords":"Metabolomics; Genome-wide association study; Computational biology; Workflow; Genetic association; Biology; Single-nucleotide polymorphism; Genomics; Computer science; Bioinformatics; Genome; Genetics; Gene; Database; Genotype","score_opus":0.049196344753754735,"score_gpt":0.3230295530492259,"score_spread":0.2738332082954712,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4386084463","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.004673054,0.000463341,0.5355865,0.0005680517,0.0002915502,0.00088689674,0.05781853,0.39773306,0.001979022],"genre_scores_gemma":[0.055998486,0.0006641745,0.7777324,0.00078813575,0.0001209765,0.005314007,0.076940924,0.07855859,0.0038823509],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99768186,0.000799771,0.00026030312,0.00065687305,0.00040668686,0.00019446448],"domain_scores_gemma":[0.99220216,0.004849901,0.00067136285,0.001046222,0.00088768173,0.00034265363],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008067191,0.004387118,0.0022271082,0.003488672,0.0010935065,0.0034488875,0.0039047995,0.0010552377,0.040151674],"category_scores_gemma":[0.021631205,0.0022291872,0.003881581,0.0019699147,0.0009908689,0.00175064,0.004203757,0.0033725344,0.019958599],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0043005385,0.00038809452,0.022814345,0.010074583,0.0064438605,0.0029614787,0.0021113083,0.061932113,0.03726322,0.03120127,0.59186524,0.22864388],"study_design_scores_gemma":[0.0022778919,0.00046558815,0.016701771,0.0011261051,0.001486704,0.001962936,0.00057752,0.3911212,0.044185992,0.12623212,0.41302383,0.00083836244],"about_ca_topic_score_codex":0.00470106,"about_ca_topic_score_gemma":0.006798481,"teacher_disagreement_score":0.040151674,"about_ca_system_score_codex":0.0012450111,"about_ca_system_score_gemma":0.004961301,"threshold_uncertainty_score":0.1343208},"labels":[],"label_agreement":null},{"id":"W4386554491","doi":"10.1093/bioinformatics/btad552","title":"μ- PBWT: a lightweight r-indexing of the PBWT for storing and querying UK Biobank data","year":2023,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Dalhousie University","funders":"National Human Genome Research Institute; European Commission; Natural Sciences and Engineering Research Council of Canada; Japan Society for the Promotion of Science; National Institutes of Health; National Science Foundation","keywords":"Computer science; Biobank; Search engine indexing; Leverage (statistics); Set (abstract data type); Data structure; Haplotype; Data mining; Index (typography); Computation; Theoretical computer science; Information retrieval; Algorithm; Artificial intelligence; Bioinformatics; Biology; Genotype; Genetics","score_opus":0.046232991827891444,"score_gpt":0.2960578176711589,"score_spread":0.24982482584326746,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4386554491","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.032100074,0.0013806986,0.8676173,0.0007611309,0.00034287444,0.0006611947,0.015038491,0.07718548,0.0049128155],"genre_scores_gemma":[0.11913596,0.00058307184,0.8385093,0.00032608636,0.0001460684,0.0007921009,0.033192854,0.0032407357,0.0040737484],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9973851,0.00028565052,0.00039192874,0.000547636,0.001189202,0.00020057928],"domain_scores_gemma":[0.99468416,0.0011584254,0.0004915956,0.0026868496,0.0007499763,0.00022897912],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015227543,0.001384109,0.0012240479,0.0030890165,0.00086390116,0.0026535832,0.004342877,0.0011522411,0.009690761],"category_scores_gemma":[0.013338219,0.0008574724,0.0013057628,0.0067439037,0.0011510119,0.00687886,0.004042896,0.001538793,0.0073380666],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014397528,0.00038600856,0.005047842,0.0013239479,0.000141541,0.00066211284,0.0009057513,0.022771103,0.051354513,0.035251886,0.12668495,0.75403064],"study_design_scores_gemma":[0.0006671652,0.0005528515,0.003874345,0.00024267574,0.00009863857,0.0014191105,0.0006673128,0.6908586,0.09367286,0.08824123,0.11943794,0.00026738818],"about_ca_topic_score_codex":0.00681375,"about_ca_topic_score_gemma":0.006203332,"teacher_disagreement_score":0.009690761,"about_ca_system_score_codex":0.0011291213,"about_ca_system_score_gemma":0.0028377562,"threshold_uncertainty_score":0.032418847},"labels":[],"label_agreement":null},{"id":"W4386979606","doi":"10.1093/bioinformatics/btad590","title":"Using a novel structure/function approach to select diverse swine major histocompatibility complex 1 alleles to predict epitopes for vaccine development","year":2023,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Animal Virus Infections Studies","field":"Agricultural and Biological Sciences","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Saskatchewan","funders":"Alliance de recherche numérique du Canada; Innovation Saskatchewan; Ministry of Agriculture - Saskatchewan; University of Saskatchewan","keywords":"Epitope; Biology; Major histocompatibility complex; Docking (animal); Computational biology; Homology modeling; Allele; Human leukocyte antigen; Genetics; Homology (biology); Amino acid; Antigen; Gene; Biochemistry; Medicine","score_opus":0.13127981398610575,"score_gpt":0.28298871093904376,"score_spread":0.15170889695293802,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4386979606","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.59527725,0.0003718207,0.39878085,0.00026584222,0.000035012796,0.00022353676,0.0002593981,0.0017577751,0.0030285958],"genre_scores_gemma":[0.7177989,0.00026904992,0.27955338,0.00015424311,0.000012918362,0.00019211305,0.0004919656,0.00010023228,0.0014271708],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9999213,0.000011691594,0.000004538819,0.000017266866,0.00003368266,0.0000114262275],"domain_scores_gemma":[0.9999386,0.000014252036,0.00001545603,0.000005551651,0.000015934647,0.000010085054],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00028337308,0.0008380049,0.00047895082,0.00047325148,0.00036024695,0.00044917027,0.0005138027,0.0005604038,0.0012804058],"category_scores_gemma":[0.0002465178,0.00031349866,0.00052298495,0.00026179067,0.00019706678,0.00039054616,0.00033162528,0.0005175822,0.00032998028],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00029578712,0.0007163283,0.005737185,0.0001591846,0.00015081675,0.00048271735,0.00006679547,0.4125284,0.5203238,0.005177088,0.0007456329,0.05361642],"study_design_scores_gemma":[0.000044888,0.00015482793,0.0007744576,0.0000041974877,0.000033106735,0.00014182995,0.000010755052,0.95693433,0.04009569,0.0009867969,0.0008081894,0.000010831386],"about_ca_topic_score_codex":0.0013611595,"about_ca_topic_score_gemma":0.0015532988,"teacher_disagreement_score":0.0013611595,"about_ca_system_score_codex":0.0005888057,"about_ca_system_score_gemma":0.0005603536,"threshold_uncertainty_score":0.0042833686},"labels":[],"label_agreement":null},{"id":"W4387300238","doi":"10.1093/bioinformatics/btad608","title":"Surfaces: a software to quantify and visualize interactions within and between proteins and ligands","year":2023,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Surface Chemistry and Catalysis","field":"Engineering","cited_by":20,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"Natural Sciences and Engineering Research Council of Canada; Compute Canada; Genome Canada","keywords":"Visualization; Computer science; Scripting language; Documentation; Software; Software visualization; Computational science; Surface (topology); Computer graphics (images); Data mining; Programming language; Software development; Component-based software engineering; Mathematics","score_opus":0.02404851307661481,"score_gpt":0.267226260133734,"score_spread":0.2431777470571192,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4387300238","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.015187209,0.0010286266,0.5433258,0.00055227167,0.00022694687,0.00022538808,0.028896121,0.40199658,0.0085611325],"genre_scores_gemma":[0.14521264,0.0024190666,0.628595,0.00077056285,0.0001742939,0.0024123073,0.086000204,0.11625068,0.018165186],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9992161,0.00012396001,0.000061990024,0.00012575333,0.00039143572,0.00008081987],"domain_scores_gemma":[0.999111,0.00045388014,0.00008769373,0.00009419296,0.0001853034,0.00006796885],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011551568,0.002057393,0.0012966502,0.0018612958,0.0007184863,0.001945303,0.0023004035,0.0010894088,0.035196237],"category_scores_gemma":[0.0030386036,0.001141644,0.0015071916,0.0011391812,0.0005270374,0.0017859193,0.0028196455,0.0020895586,0.0138788],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00089116377,0.00023276013,0.004579404,0.0032689155,0.00063022645,0.0005686166,0.00083582656,0.036194235,0.08052765,0.03231774,0.5867927,0.2531607],"study_design_scores_gemma":[0.000403753,0.00015806414,0.004626771,0.00034695587,0.00019006438,0.0006623819,0.00020110133,0.40350705,0.09819082,0.04488419,0.446523,0.00030584997],"about_ca_topic_score_codex":0.0027655673,"about_ca_topic_score_gemma":0.0019346247,"teacher_disagreement_score":0.035196237,"about_ca_system_score_codex":0.0005551404,"about_ca_system_score_gemma":0.0014077252,"threshold_uncertainty_score":0.117743194},"labels":[],"label_agreement":null},{"id":"W4387359355","doi":"10.1093/bioinformatics/btag207","title":"Segzoo: a turnkey system that summarizes genome annotations","year":2023,"lang":"en","type":"preprint","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Vector Institute; Princess Margaret Cancer Centre; University of Toronto; University Health Network","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Computer science; Python (programming language); Annotation; Turnkey; Upload; Visualization; MIT License; JSON; Genome; Source code; Unix; Download; Software; Data mining; Operating system; Database; Artificial intelligence; Biology","score_opus":0.03593523873103464,"score_gpt":0.2498236841964263,"score_spread":0.21388844546539168,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4387359355","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.006216215,0.001120621,0.08419174,0.0005584689,0.0006538048,0.0002576058,0.24178582,0.6488819,0.016333813],"genre_scores_gemma":[0.03797248,0.0010912815,0.15580952,0.0012103174,0.00020921776,0.0010479919,0.62511754,0.16291106,0.014630568],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.998589,0.000144787,0.00009429362,0.0006579006,0.00034080885,0.00017324489],"domain_scores_gemma":[0.9974336,0.00068661995,0.00032738718,0.00070302736,0.000621963,0.0002274431],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002099564,0.0041047647,0.0017437413,0.004872131,0.0015285689,0.0045624524,0.0033729174,0.0011692189,0.06480628],"category_scores_gemma":[0.006911676,0.0016788219,0.0024146666,0.0039040886,0.00080774527,0.0034340252,0.004867378,0.0017498703,0.05168053],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013446164,0.000047994163,0.005800578,0.0016596743,0.0003505675,0.00031470796,0.00054747035,0.0018654233,0.013376778,0.0041297437,0.9091449,0.061417658],"study_design_scores_gemma":[0.00037241276,0.00017137008,0.012349953,0.00048954156,0.00039011106,0.0005575714,0.00050369574,0.019334799,0.034012206,0.022236971,0.9091959,0.0003853915],"about_ca_topic_score_codex":0.009655733,"about_ca_topic_score_gemma":0.013104821,"teacher_disagreement_score":0.06480628,"about_ca_system_score_codex":0.0013981375,"about_ca_system_score_gemma":0.002825598,"threshold_uncertainty_score":0.2167986},"labels":[],"label_agreement":null},{"id":"W4387363064","doi":"10.1093/bioinformatics/btad613","title":"SBILib: a handle for protein modeling and engineering","year":2023,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Protein Structure and Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"BC Children's Hospital; University of British Columbia","funders":"Agencia Estatal de Investigación; Ministerio de Ciencia e Innovación","keywords":"Python (programming language); Computer science; Parsing; Structural bioinformatics; File format; Programming language; Simple (philosophy); Computational biology; Software engineering; Protein structure; Chemistry; Biology; Biochemistry","score_opus":0.008994534504336108,"score_gpt":0.2203856039355147,"score_spread":0.2113910694311786,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4387363064","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0023220235,0.00036340312,0.28827897,0.00050755957,0.00029735194,0.00020801871,0.037025556,0.6600639,0.010933221],"genre_scores_gemma":[0.035331115,0.0012386161,0.33562616,0.0018861262,0.0001805743,0.0020217597,0.192065,0.40625915,0.02539154],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99877757,0.0001624401,0.00011046983,0.00021362999,0.0005409898,0.00019480656],"domain_scores_gemma":[0.99866176,0.00037693247,0.0001176289,0.00035552238,0.00028413782,0.00020396995],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0024323019,0.0028726317,0.0018203172,0.0020773683,0.0013376896,0.0031900287,0.006736322,0.0016723011,0.088926256],"category_scores_gemma":[0.00477331,0.0024464466,0.0032580392,0.0019309048,0.00079406006,0.0034894112,0.005888807,0.0047126445,0.08063215],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00063870125,0.00020027981,0.0018418805,0.0016388078,0.00030232902,0.00048261406,0.00035184226,0.010348723,0.0103276195,0.0212932,0.8730425,0.07953138],"study_design_scores_gemma":[0.00045455692,0.00008248136,0.0017739807,0.00041913023,0.0001061308,0.0005776332,0.000089832785,0.11104115,0.021028154,0.040259458,0.8239128,0.0002546633],"about_ca_topic_score_codex":0.004744968,"about_ca_topic_score_gemma":0.0046968027,"teacher_disagreement_score":0.088926256,"about_ca_system_score_codex":0.0013959697,"about_ca_system_score_gemma":0.0022447414,"threshold_uncertainty_score":0.29748797},"labels":[],"label_agreement":null},{"id":"W4387796811","doi":"10.1093/bioinformatics/btad639","title":"Privacy-preserving federated genome-wide association studies via dynamic sampling","year":2023,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Privacy-Preserving Technologies in Data","field":"Computer Science","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Artificial Intelligence in Medicine (Canada)","funders":"U.S. National Library of Medicine; National Institute of General Medical Sciences; University of Texas Health Science Center at Houston; Cisco Systems; National Science Foundation; National Institutes of Health; Cancer Prevention and Research Institute of Texas","keywords":"Computer science; Robustness (evolution); Data mining; Source code; Genome-wide association study; Data sharing; Single-nucleotide polymorphism","score_opus":0.05433862897628468,"score_gpt":0.30878861326997387,"score_spread":0.2544499842936892,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4387796811","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.017215658,0.00025893806,0.9791574,0.00066150905,0.00003831905,0.00013588004,0.00041573105,0.001300295,0.000816141],"genre_scores_gemma":[0.6326769,0.00033632512,0.3625865,0.0005337249,0.00017947317,0.00049911125,0.0014396574,0.0001916312,0.0015566875],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.98197997,0.010180685,0.0009061649,0.002980603,0.0032439826,0.00070859655],"domain_scores_gemma":[0.93695253,0.03171707,0.0038604308,0.023501376,0.0028574043,0.0011111755],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.019615754,0.0008412925,0.0020288513,0.0015990379,0.0017552293,0.003723102,0.0039619203,0.0014581347,0.0020641487],"category_scores_gemma":[0.061439354,0.00082608877,0.0016986761,0.0029913594,0.0022888102,0.0042276992,0.0061229696,0.0022758564,0.001038317],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0029103945,0.00046830802,0.022438444,0.0005307265,0.0006107177,0.001389706,0.0013512853,0.46436003,0.014828294,0.15740094,0.00883237,0.32487884],"study_design_scores_gemma":[0.00017465497,0.00012327459,0.0011483657,0.000038946197,0.000084284475,0.0005364024,0.00012327223,0.7976,0.0071545704,0.1898046,0.0031762053,0.000035370667],"about_ca_topic_score_codex":0.0013695002,"about_ca_topic_score_gemma":0.0014766242,"teacher_disagreement_score":0.019615754,"about_ca_system_score_codex":0.0014239365,"about_ca_system_score_gemma":0.0040464457,"threshold_uncertainty_score":0.10373926},"labels":[],"label_agreement":null},{"id":"W4388556708","doi":"10.1093/bioinformatics/btad673","title":"hipFG: high-throughput harmonization and integration pipeline for functional genomics data","year":2023,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Weston Brain Institute; National Institute on Aging; Alzheimer’s Research UK; Alzheimer's Association","keywords":"Pipeline (software); Harmonization; Throughput; Computer science; Genomics; Functional genomics; Data integration; Computational biology; Data mining; Biology; Programming language; Genome; Operating system; Gene; Genetics","score_opus":0.05108167327523701,"score_gpt":0.28346035393168745,"score_spread":0.23237868065645045,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4388556708","genre_codex":"software","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0061854296,0.0004534682,0.3259123,0.00093500357,0.00033893326,0.0007179766,0.13502075,0.5262404,0.0041956934],"genre_scores_gemma":[0.042062555,0.00050971285,0.39048374,0.001445356,0.00018750358,0.002762354,0.47598633,0.08124099,0.0053213937],"study_design_codex":"not_applicable","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9956589,0.000638954,0.00044396758,0.0013341982,0.0015208493,0.00040316622],"domain_scores_gemma":[0.99399453,0.0017834641,0.00049448205,0.0019490859,0.0012428208,0.00053557963],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.009499087,0.0033091547,0.0019138814,0.004659533,0.0020284844,0.004124656,0.0045777257,0.0013564257,0.026472643],"category_scores_gemma":[0.015593366,0.0022355819,0.0031522932,0.004305152,0.0011240618,0.0034084595,0.007637373,0.0038640904,0.026530951],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0018130686,0.00022681143,0.009371953,0.0016157848,0.00070746784,0.000783959,0.0011398217,0.006209657,0.037119266,0.0073090573,0.7921094,0.14159377],"study_design_scores_gemma":[0.00094230566,0.0003247239,0.028192736,0.0005843531,0.0005112884,0.0016261494,0.0006140928,0.07931459,0.11667313,0.05133773,0.7189645,0.00091433263],"about_ca_topic_score_codex":0.0071523115,"about_ca_topic_score_gemma":0.006236257,"teacher_disagreement_score":0.026472643,"about_ca_system_score_codex":0.0016056256,"about_ca_system_score_gemma":0.0048562996,"threshold_uncertainty_score":0.088559866},"labels":[],"label_agreement":null},{"id":"W4388608496","doi":"10.1093/bioinformatics/btad684","title":"PiDeeL: metabolic pathway-informed deep learning model for survival analysis and pathological classification of gliomas","year":2023,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Glioma Diagnosis and Treatment","field":"Medicine","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"Bilim Akademisi; Centre National de la Recherche Scientifique","keywords":"Computer science; Recall; Artificial intelligence; Deep learning; Machine learning; Pathological; Index (typography); Concordance; Mechanism (biology); Pathology; Bioinformatics; Medicine; Biology; Psychology; Cognitive psychology","score_opus":0.06304541078609775,"score_gpt":0.3137776443325855,"score_spread":0.2507322335464877,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4388608496","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.16754016,0.0023595248,0.81501454,0.002061317,0.00020847599,0.00014483619,0.0025608796,0.0055898787,0.0045204284],"genre_scores_gemma":[0.910605,0.000593764,0.077554956,0.0008545425,0.00007626186,0.00027583507,0.0029013422,0.000093649054,0.0070445486],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99986887,0.000025428233,0.0000069396233,0.000043749482,0.000031279225,0.00002384114],"domain_scores_gemma":[0.9997912,0.000091093396,0.000025360145,0.000016322649,0.000056701334,0.000019290219],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00051249133,0.00080688327,0.00050895003,0.00041692032,0.00016784565,0.00059299136,0.0014717143,0.00105276,0.0015563426],"category_scores_gemma":[0.0012767573,0.00029369214,0.0005961188,0.00026856145,0.00027343738,0.00057402573,0.00079465465,0.0011847899,0.00039246565],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00025072723,0.0001757467,0.0043561365,0.00011451619,0.00010700888,0.00015833632,0.000029211473,0.88230884,0.0041807266,0.0015902668,0.004244333,0.1024841],"study_design_scores_gemma":[0.000008743662,0.000028621502,0.00022030495,0.0000045412253,0.000008108103,0.000014235933,0.0000020204925,0.9978243,0.00066096627,0.00095294893,0.00027051984,0.0000046735167],"about_ca_topic_score_codex":0.006274188,"about_ca_topic_score_gemma":0.0077187708,"teacher_disagreement_score":0.006274188,"about_ca_system_score_codex":0.00084285595,"about_ca_system_score_gemma":0.0011245646,"threshold_uncertainty_score":0.012475371},"labels":[],"label_agreement":null},{"id":"W4388706662","doi":"10.1093/bioinformatics/btad679","title":"ReGeNNe: genetic pathway-based deep neural network using canonical correlation regularizer for disease prediction","year":2023,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":12,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Princess Margaret Cancer Centre; Public Health Ontario; University of Toronto; University Health Network","funders":"","keywords":"Interpretability; Computer science; Artificial intelligence; Correlation; Machine learning; Robustness (evolution); Artificial neural network; Canonical correlation; Generalizability theory; Convolutional neural network; Novelty; Gene; Biology; Mathematics; Genetics; Statistics","score_opus":0.01467302902274496,"score_gpt":0.2301602004368411,"score_spread":0.21548717141409615,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4388706662","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.10831016,0.0018027512,0.8657306,0.0012852616,0.0001884218,0.00015145952,0.0036787577,0.014168235,0.0046843765],"genre_scores_gemma":[0.7210708,0.00083663495,0.2601956,0.0007016488,0.00009080751,0.0003029776,0.008221326,0.00042488496,0.00815531],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9998547,0.00002863156,0.0000070129477,0.0000516755,0.000037418813,0.000020557112],"domain_scores_gemma":[0.9997218,0.00010715469,0.00003277384,0.000038978887,0.000076326854,0.000022942646],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00056608923,0.00071313686,0.0005361634,0.0006252999,0.00018656344,0.0004541525,0.001175004,0.00070701813,0.002136209],"category_scores_gemma":[0.001447772,0.0002884686,0.00053982734,0.0004807381,0.00031873467,0.00064916746,0.0006419784,0.0010770295,0.0006380085],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00020668504,0.00014743683,0.005776163,0.00011537473,0.00014330381,0.00017608491,0.000034337074,0.8225325,0.0036995586,0.005193587,0.010745112,0.15122995],"study_design_scores_gemma":[0.000007575319,0.000010628565,0.00017235709,0.0000053761846,0.000006085918,0.000012794945,0.000001609311,0.99659747,0.00076473295,0.0019870102,0.000431038,0.0000033957526],"about_ca_topic_score_codex":0.014280565,"about_ca_topic_score_gemma":0.01822583,"teacher_disagreement_score":0.014280565,"about_ca_system_score_codex":0.00089677697,"about_ca_system_score_gemma":0.0014395062,"threshold_uncertainty_score":0.028394878},"labels":[],"label_agreement":null},{"id":"W4389452578","doi":"10.1093/bioinformatics/btad739","title":"Local Disordered Region Sampling (LDRS) for ensemble modeling of proteins with experimentally undetermined or low confidence prediction segments","year":2023,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Protein Structure and Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"College of Family Physicians of Canada; Hospital for Sick Children; University of Toronto","funders":"National Institutes of Health; National Institute of General Medical Sciences; Natural Sciences and Engineering Research Council of Canada; Canada Research Chairs","keywords":"Sampling (signal processing); Confidence interval; Computer science; Artificial intelligence; Computational biology; Pattern recognition (psychology); Statistics; Biology; Mathematics; Computer vision","score_opus":0.02603965739753545,"score_gpt":0.269981924206933,"score_spread":0.24394226680939757,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4389452578","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.02568083,0.00029189288,0.9125709,0.00022011052,0.00008829616,0.000113465816,0.005253876,0.05109181,0.0046888664],"genre_scores_gemma":[0.2728495,0.00051370164,0.68103415,0.00029938447,0.000080106336,0.0008159745,0.020894581,0.01610959,0.0074030426],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9995945,0.00011174691,0.00002621051,0.000083488725,0.00014641971,0.000037611513],"domain_scores_gemma":[0.99931204,0.00034878895,0.00006977978,0.00010905763,0.00010770991,0.000052614454],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013157085,0.0010383654,0.00079715066,0.0005598691,0.00048549817,0.00079646887,0.0023761143,0.0008257228,0.014111934],"category_scores_gemma":[0.0027306266,0.0005116818,0.0011357312,0.00048758864,0.0002956542,0.00096346374,0.0011083537,0.0010894373,0.003677988],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00056649564,0.00018783784,0.004959463,0.0006688798,0.00039944024,0.0003958656,0.00027526164,0.7785875,0.018591214,0.033433326,0.060327012,0.101607725],"study_design_scores_gemma":[0.000028995486,0.000022297645,0.00016310358,0.000012392685,0.000011327658,0.00003543168,0.000010047939,0.98449796,0.0033428292,0.0040283413,0.007836345,0.00001090885],"about_ca_topic_score_codex":0.004279665,"about_ca_topic_score_gemma":0.006521934,"teacher_disagreement_score":0.014111934,"about_ca_system_score_codex":0.00067260436,"about_ca_system_score_gemma":0.0012452151,"threshold_uncertainty_score":0.047209144},"labels":[],"label_agreement":null},{"id":"W4389452890","doi":"10.1093/bioinformatics/btad720","title":"NMJ Analyser: a novel method to quantify neuromuscular junction morphology in zebrafish","year":2023,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Zebrafish Biomedical Research Applications","field":"Biochemistry, Genetics and Molecular Biology","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal; Institut National de la Recherche Scientifique","funders":"Fonds de Recherche du Québec - Santé; Canadian Institutes of Health Research; Natural Sciences and Engineering Research Council of Canada; Fondation Brain Canada","keywords":"Neuromuscular junction; Zebrafish; Postsynaptic potential; Biology; Neuroscience; Neuromuscular transmission; Cell biology; Genetics; Receptor","score_opus":0.03480099033414182,"score_gpt":0.3513222281535572,"score_spread":0.3165212378194154,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4389452890","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.052623864,0.0014397829,0.76797074,0.00046648656,0.00033633332,0.0005042364,0.025398144,0.14142418,0.009836311],"genre_scores_gemma":[0.07033298,0.0007914833,0.890399,0.00040761483,0.0000664024,0.001646733,0.011606796,0.015770894,0.008978171],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9992072,0.000041582494,0.00006952556,0.00020945426,0.00041616615,0.000056073954],"domain_scores_gemma":[0.9991835,0.00020749158,0.00020938499,0.00010737905,0.00023208589,0.000060250615],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001211565,0.0011191053,0.0007787907,0.0042179585,0.00073023536,0.0008786316,0.0011840374,0.0010252539,0.014695609],"category_scores_gemma":[0.0017140679,0.0010199342,0.0008571514,0.0009016443,0.0004914344,0.0009876827,0.0012001495,0.001075741,0.0061774408],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002878604,0.000054853997,0.0076889535,0.0009212385,0.000112748625,0.0003385189,0.0003121836,0.002128696,0.8162583,0.0029762262,0.03328009,0.13564037],"study_design_scores_gemma":[0.00012646831,0.00023490361,0.05433407,0.00035741608,0.00023078741,0.0018319294,0.00015538445,0.105602235,0.71772593,0.0034348713,0.1155155,0.00045052374],"about_ca_topic_score_codex":0.003963433,"about_ca_topic_score_gemma":0.0071829553,"teacher_disagreement_score":0.014695609,"about_ca_system_score_codex":0.0008360577,"about_ca_system_score_gemma":0.001088638,"threshold_uncertainty_score":0.049161732},"labels":[],"label_agreement":null},{"id":"W4389453022","doi":"10.1093/bioinformatics/btad726","title":"invMap: a sensitive mapping tool for long noisy reads with inversion structural variants","year":2023,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Saskatchewan","funders":"Natural Science Basic Research Program of Shaanxi Province; Natural Sciences and Engineering Research Council of Canada; Education Department of Shaanxi Province; China Scholarship Council","keywords":"Chaining; Nanopore sequencing; Computer science; Inversion (geology); Benchmark (surveying); Software; Algorithm; Data mining; Artificial intelligence; Genome; Genetics; Biology; Cartography","score_opus":0.014794007641766662,"score_gpt":0.22759963580796688,"score_spread":0.21280562816620022,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4389453022","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.01879274,0.0013001388,0.75259334,0.00046454198,0.0003999446,0.00033946257,0.012417286,0.21025455,0.0034380585],"genre_scores_gemma":[0.10856555,0.0006393202,0.83393854,0.00093312754,0.0001835178,0.0013475934,0.026981978,0.022790272,0.0046200943],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9968736,0.00070467166,0.00025452994,0.0010441203,0.0009392813,0.00018378199],"domain_scores_gemma":[0.99364704,0.0037865278,0.00072362664,0.00096482266,0.0006615918,0.00021640495],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0047334433,0.0026891374,0.0014542035,0.0027956262,0.0015294753,0.0024396041,0.0033093274,0.0017851578,0.012538743],"category_scores_gemma":[0.019228945,0.0015862574,0.0019403233,0.002563102,0.0008520875,0.0027273563,0.004458185,0.002894603,0.0072048674],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.003080405,0.0003578313,0.019137394,0.005217016,0.0016685043,0.0018820264,0.0020836843,0.030825388,0.123215735,0.020508165,0.22786877,0.5641551],"study_design_scores_gemma":[0.00076374295,0.0006788583,0.014000553,0.000598871,0.0005121248,0.004165892,0.0006386635,0.4344646,0.25480178,0.054411978,0.23415282,0.0008101418],"about_ca_topic_score_codex":0.001077091,"about_ca_topic_score_gemma":0.002004862,"teacher_disagreement_score":0.012538743,"about_ca_system_score_codex":0.00061270205,"about_ca_system_score_gemma":0.0015093187,"threshold_uncertainty_score":0.041946232},"labels":[],"label_agreement":null},{"id":"W4389799054","doi":"10.1093/bioinformatics/btad750","title":"Rarity: discovering rare cell populations from single-cell imaging data","year":2023,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Single-cell and spatial transcriptomics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":12,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Engineering and Physical Sciences Research Council; Medical Research Council; University of Toronto; King's Health Partners; King's College London; UK Research and Innovation; European Commission; Alan Turing Institute; Wellcome Trust; Francis Crick Institute; Cancer Research UK","keywords":"Computer science; Computational biology; Data mining; Biology","score_opus":0.05919013795234787,"score_gpt":0.259301637795228,"score_spread":0.20011149984288015,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4389799054","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.027135717,0.0017977989,0.8690078,0.0010734583,0.0001759567,0.00046912633,0.026106698,0.070911005,0.0033224933],"genre_scores_gemma":[0.09785864,0.0010691207,0.8455761,0.00059455866,0.0001564335,0.0008405073,0.046644576,0.0049948967,0.0022652375],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9972664,0.0006255716,0.00021726219,0.0008204379,0.0008668751,0.00020332746],"domain_scores_gemma":[0.9920264,0.0037520584,0.00086879503,0.0017535904,0.0012157984,0.00038332597],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0067404457,0.0013189525,0.0016115843,0.0037705814,0.0012143095,0.0027900788,0.0049280585,0.0022274305,0.0043126317],"category_scores_gemma":[0.01980142,0.0009215008,0.0016982768,0.0024793586,0.0011692379,0.0023967582,0.0032698417,0.0025707067,0.0048505007],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0015435383,0.0003892226,0.04545334,0.004628789,0.0010104523,0.0011482483,0.001500732,0.14246462,0.14095129,0.031998906,0.22426848,0.40464243],"study_design_scores_gemma":[0.00020348775,0.00023789494,0.017630067,0.0003859229,0.00019409125,0.0017104489,0.00035280953,0.75531197,0.09387907,0.05411137,0.07560677,0.00037611075],"about_ca_topic_score_codex":0.0070455144,"about_ca_topic_score_gemma":0.009959134,"teacher_disagreement_score":0.0070455144,"about_ca_system_score_codex":0.0011782398,"about_ca_system_score_gemma":0.0027108868,"threshold_uncertainty_score":0.035647333},"labels":[],"label_agreement":null},{"id":"W4390345333","doi":"10.1093/bioinformatics/btad783","title":"ESQmodel: biologically informed evaluation of 2-D cell segmentation quality in multiplexed tissue images","year":2023,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Single-cell and spatial transcriptomics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canada's Michael Smith Genome Sciences Centre; University of British Columbia; BC Children's Hospital; BC Cancer Agency","funders":"Canadian Institutes of Health Research; BC Cancer Foundation; Natural Sciences and Engineering Research Council of Canada; Cancer Research UK; Michael Smith Health Research BC","keywords":"Computer science; Segmentation; Quality (philosophy); Computer vision; Biological tissue; Artificial intelligence; Multiplexing; Biomedical engineering; Medicine","score_opus":0.06053723972205262,"score_gpt":0.3398534103869885,"score_spread":0.2793161706649359,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4390345333","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.08561441,0.00092783367,0.86861575,0.0006282445,0.0001198014,0.00029778513,0.0071887462,0.03336691,0.0032404321],"genre_scores_gemma":[0.39870957,0.00044631425,0.5779971,0.0006296705,0.0000742997,0.0005315161,0.012774122,0.005766511,0.003070833],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99870944,0.00029675246,0.00007576809,0.00032601738,0.00050434255,0.000087704306],"domain_scores_gemma":[0.9963905,0.0022135852,0.00037689146,0.0003668955,0.0004976135,0.00015448622],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0046990993,0.0013646198,0.0010389077,0.0021359182,0.0007118864,0.0026530644,0.0019572573,0.002091336,0.0055421535],"category_scores_gemma":[0.011952716,0.0006008879,0.001488235,0.001037141,0.0009859288,0.0015524522,0.0020337163,0.0009858448,0.0018457428],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014088668,0.00030680042,0.026914524,0.0014812835,0.00072678586,0.00038782263,0.00062864064,0.658537,0.07504813,0.014358492,0.028535703,0.19166596],"study_design_scores_gemma":[0.000036488596,0.000099224504,0.0028247475,0.00005027442,0.000024482266,0.00012583133,0.000054413442,0.9718766,0.0153460335,0.006894154,0.0026210826,0.000046706715],"about_ca_topic_score_codex":0.004766458,"about_ca_topic_score_gemma":0.0073374338,"teacher_disagreement_score":0.0055421535,"about_ca_system_score_codex":0.0018153379,"about_ca_system_score_gemma":0.0014883513,"threshold_uncertainty_score":0.024851501},"labels":[],"label_agreement":null},{"id":"W4390793818","doi":"10.1093/bioinformatics/btad738","title":"Seq-InSite: sequence supersedes structure for protein interaction site prediction","year":2024,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Protein Structure and Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":27,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University; Western University","funders":"Natural Sciences and Engineering Research Council of Canada; Compute Canada","keywords":"Computer science; Sequence (biology); Source code; Matching (statistics); Code (set theory); Quality (philosophy); Data mining; Artificial intelligence; Machine learning; Programming language; Biology","score_opus":0.010736573104246281,"score_gpt":0.25635788674873794,"score_spread":0.24562131364449166,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4390793818","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.08294156,0.0016258141,0.48940888,0.0019217926,0.00086904084,0.00056641945,0.090868115,0.31732616,0.014472193],"genre_scores_gemma":[0.19081579,0.0010817449,0.5031477,0.0011088805,0.00022057445,0.001240015,0.26953736,0.025803763,0.0070441295],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9987747,0.00034275444,0.00007527866,0.00033716808,0.0003814746,0.00008867228],"domain_scores_gemma":[0.9978473,0.001185569,0.00022717167,0.00025615387,0.00027102444,0.00021283852],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0027925556,0.0023137014,0.0016235173,0.0016200166,0.0012739198,0.0016909175,0.003917534,0.0014021451,0.01319833],"category_scores_gemma":[0.007951347,0.0011555721,0.0019751037,0.0018982161,0.00087072037,0.002112783,0.0022971479,0.0035123911,0.011506739],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0040911604,0.0010924785,0.0271884,0.0037751482,0.0012134322,0.0008523109,0.0006215721,0.14004354,0.046275232,0.03013718,0.6075265,0.13718313],"study_design_scores_gemma":[0.0005176351,0.00043421233,0.0037293674,0.00020302669,0.00020403194,0.00053467,0.00014593372,0.85267675,0.031529035,0.02892398,0.080946796,0.00015455343],"about_ca_topic_score_codex":0.0055826455,"about_ca_topic_score_gemma":0.01183582,"teacher_disagreement_score":0.01319833,"about_ca_system_score_codex":0.00091074046,"about_ca_system_score_gemma":0.0030250452,"threshold_uncertainty_score":0.044152737},"labels":[],"label_agreement":null},{"id":"W4390946662","doi":"10.1093/bioinformatics/btae022","title":"Concurrent prediction of RNA secondary structures with pseudoknots and local 3D motifs in an integer programming framework","year":2024,"lang":"en","type":"article","venue":"Bioinformatics","topic":"RNA and protein synthesis mechanisms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Montréal","funders":"Fonds de Recherche du Québec - Santé; Natural Sciences and Engineering Research Council of Canada","keywords":"Integer programming; Computer science; Integer (computer science); Nucleic acid secondary structure; RNA; Computational biology; Theoretical computer science; Algorithm; Biology; Genetics; Programming language; Gene","score_opus":0.009611091094722876,"score_gpt":0.24205697556432612,"score_spread":0.23244588446960324,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4390946662","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.15498449,0.00024343235,0.83949506,0.00027466874,0.0000398723,0.00006389806,0.0004084543,0.0023290177,0.0021611662],"genre_scores_gemma":[0.50852394,0.0001247501,0.4873849,0.00015213125,0.00004690723,0.00018034548,0.0012545636,0.0003672473,0.0019652094],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99959916,0.0001215296,0.000017997389,0.00011356815,0.00008211365,0.00006562773],"domain_scores_gemma":[0.9986266,0.0009445842,0.00013617554,0.000054016924,0.00015283421,0.00008584152],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010146995,0.00094701524,0.00078158645,0.0004678228,0.00032484657,0.0009847902,0.0011194235,0.00078833464,0.0016936403],"category_scores_gemma":[0.0029109255,0.00050166895,0.0007771079,0.00047029826,0.0005561008,0.00086850714,0.00069152145,0.0012500146,0.0004244644],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00013213966,0.00007129854,0.0013449445,0.00005614832,0.00001895728,0.00006613498,0.000029383795,0.9732597,0.0018506094,0.0028111725,0.00056157424,0.01979794],"study_design_scores_gemma":[0.0000033759852,0.0000070241163,0.000035132874,0.0000014560894,0.0000013786819,0.0000038163175,0.0000034760856,0.99851173,0.00035969252,0.0009941962,0.00007753733,0.0000012110783],"about_ca_topic_score_codex":0.0045491294,"about_ca_topic_score_gemma":0.004624313,"teacher_disagreement_score":0.0045491294,"about_ca_system_score_codex":0.0006043976,"about_ca_system_score_gemma":0.0011638746,"threshold_uncertainty_score":0.009045303},"labels":[],"label_agreement":null},{"id":"W4391261909","doi":"10.1093/bioinformatics/btae051","title":"TKSM: highly modular, user-customizable, and scalable transcriptomic sequencing long-read simulator","year":2024,"lang":"en","type":"article","venue":"Bioinformatics","topic":"RNA modifications and cancer","field":"Biochemistry, Genetics and Molecular Biology","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia; Simon Fraser University","funders":"Natural Sciences and Engineering Research Council of Canada; Michael Smith Health Research BC","keywords":"Computer science; Modular design; Scalability; Software; Human–computer interaction; Operating system; Computer architecture","score_opus":0.012109639831769402,"score_gpt":0.2418908974974867,"score_spread":0.2297812576657173,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4391261909","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.052469607,0.0005812618,0.8097657,0.00048330027,0.00042682877,0.0004340234,0.008793229,0.11370224,0.013343859],"genre_scores_gemma":[0.4149503,0.0012452141,0.5245814,0.0009534667,0.00012434013,0.0023984397,0.023614641,0.020634713,0.011497435],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99955255,0.000097212374,0.000039733517,0.00009872121,0.00015065022,0.00006122215],"domain_scores_gemma":[0.9991672,0.00037862256,0.00005729712,0.00012279864,0.00016985222,0.000104232975],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001123245,0.0010953398,0.00071046175,0.00033916775,0.00045041408,0.00097043824,0.002720103,0.001352152,0.008926094],"category_scores_gemma":[0.0027866235,0.0007305465,0.0014409841,0.000495788,0.00051920017,0.00084284146,0.0010829628,0.0018243879,0.003672514],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007752174,0.00023170152,0.007475092,0.0011579636,0.00028820915,0.00041021436,0.0004002378,0.8345887,0.068326995,0.016719636,0.032085486,0.03754055],"study_design_scores_gemma":[0.000124554,0.00009830479,0.0006908049,0.000031725653,0.0000432968,0.0000902246,0.000028492115,0.9449239,0.022091128,0.0047051087,0.027114687,0.000057663725],"about_ca_topic_score_codex":0.0040777856,"about_ca_topic_score_gemma":0.003570029,"teacher_disagreement_score":0.008926094,"about_ca_system_score_codex":0.0009090101,"about_ca_system_score_gemma":0.0018170297,"threshold_uncertainty_score":0.029860735},"labels":[],"label_agreement":null},{"id":"W4391335182","doi":"10.1093/bioinformatics/btae264","title":"TA-RNN: an attention-based time-aware recurrent neural network architecture for electronic health records","year":2024,"lang":"en","type":"preprint","venue":"Bioinformatics","topic":"Machine Learning in Healthcare","field":"Computer Science","cited_by":20,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Institute of Biomedical Imaging and Bioengineering; Canadian Institutes of Health Research; National Institutes of Health; Genentech; IXICO; H. Lundbeck A/S; Servier; Eisai; National Institute of General Medical Sciences; Northern California Institute for Research and Education; University of North Texas; Pfizer; Novartis Pharmaceuticals Corporation; University of Southern California; Biogen; Eli Lilly and Company; Bristol-Myers Squibb; BioClinica; U.S. Department of Defense; Alzheimer's Disease Neuroimaging Initiative; Meso Scale Diagnostics; National Institute on Aging; Alzheimer's Association","keywords":"Recurrent neural network; Computer science; Architecture; Health records; Electronic health record; Artificial neural network; Artificial intelligence; History; Health care; Political science","score_opus":0.021397708566468953,"score_gpt":0.3124257444798212,"score_spread":0.29102803591335225,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4391335182","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.06978405,0.0028823419,0.9173939,0.0009726852,0.00030906918,0.000088263776,0.00068030116,0.0043070717,0.0035822028],"genre_scores_gemma":[0.832149,0.0015625404,0.15670033,0.0006183861,0.00015976503,0.00013663042,0.0016850486,0.00012597913,0.00686225],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9996878,0.00006679047,0.000026566195,0.000113425805,0.00006240488,0.000043019405],"domain_scores_gemma":[0.9995704,0.00015735153,0.000052819574,0.000036948244,0.00015970512,0.00002278583],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00077557645,0.0007685375,0.0005055476,0.00041534094,0.00023505131,0.00046245125,0.0015276108,0.0007292103,0.0010777459],"category_scores_gemma":[0.002043774,0.00032194902,0.0007111358,0.0004679577,0.00030269218,0.0009296604,0.0006065478,0.0012663256,0.0004627061],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00034288992,0.00021296085,0.0035550576,0.00017301652,0.00022833401,0.0002138279,0.00014447086,0.6217842,0.012827762,0.0042829234,0.0074406923,0.3487939],"study_design_scores_gemma":[0.0000047563576,0.000029322258,0.00026790155,0.0000062109775,0.000021343201,0.000019567868,0.000003967284,0.997315,0.0010128642,0.00091669784,0.0003964843,0.0000058365486],"about_ca_topic_score_codex":0.020280091,"about_ca_topic_score_gemma":0.022068463,"teacher_disagreement_score":0.020280091,"about_ca_system_score_codex":0.0008729417,"about_ca_system_score_gemma":0.0008763699,"threshold_uncertainty_score":0.040324092},"labels":[],"label_agreement":null},{"id":"W4391389830","doi":"10.1093/bioinformatics/btae056","title":"PERFUMES: pipeline to extract RNA functional motifs and exposed structures","year":2024,"lang":"en","type":"article","venue":"Bioinformatics","topic":"RNA and protein synthesis mechanisms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Montreal Clinical Research Institute; McGill University","funders":"Fonds de recherche du Québec – Nature et technologies; Natural Sciences and Engineering Research Council of Canada; Institut de Valorisation des Données; Compute Canada","keywords":"RNA; Computational biology; Structural motif; Computer science; Circular RNA; Pipeline (software); Python (programming language); Base pair; Biology; Gene; Genetics; Biochemistry; Programming language","score_opus":0.016611471485229144,"score_gpt":0.23920252842017242,"score_spread":0.22259105693494327,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4391389830","genre_codex":"software","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.025384318,0.0009346117,0.30977586,0.00074873934,0.00017518198,0.00040174767,0.12401558,0.5316604,0.0069036135],"genre_scores_gemma":[0.1620742,0.0007719622,0.45950764,0.001358842,0.0001179101,0.0014147278,0.31852782,0.048052873,0.008174077],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9994124,0.0000782799,0.00003491611,0.00023521346,0.00015264118,0.000086545384],"domain_scores_gemma":[0.9992028,0.0004484002,0.00008091527,0.00008876503,0.00011652043,0.00006273415],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001119647,0.0033179482,0.0011790516,0.0025887042,0.0010239157,0.0018037235,0.0021692323,0.0010288924,0.03521279],"category_scores_gemma":[0.0043122116,0.001200894,0.002265667,0.0013544752,0.0005443432,0.0016386181,0.0017301973,0.0014428494,0.014084734],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.002830604,0.0004159784,0.016222114,0.005268517,0.0008728346,0.0012503767,0.0007725437,0.0307445,0.08378682,0.016849214,0.59813875,0.24284773],"study_design_scores_gemma":[0.0009712809,0.00058933924,0.017371792,0.0003242502,0.00038024192,0.0020403422,0.0004577914,0.47290006,0.124375865,0.06454025,0.31544074,0.00060808513],"about_ca_topic_score_codex":0.0048052515,"about_ca_topic_score_gemma":0.009157067,"teacher_disagreement_score":0.03521279,"about_ca_system_score_codex":0.001169107,"about_ca_system_score_gemma":0.0018591611,"threshold_uncertainty_score":0.11779851},"labels":[],"label_agreement":null},{"id":"W4391390112","doi":"10.1093/bioinformatics/btae034","title":"GeNLP: a web tool for NLP-based exploration and prediction of microbial gene function","year":2024,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Machine Learning in Bioinformatics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Azrieli Foundation; Israel Science Foundation","keywords":"Computer science; MIT License; Context (archaeology); Function (biology); License; Source code; Interface (matter); Semantics (computer science); World Wide Web; Artificial intelligence; Web application; Natural language processing; Programming language; Biology; Genetics","score_opus":0.011154803564037847,"score_gpt":0.23325216432045523,"score_spread":0.2220973607564174,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4391390112","genre_codex":"software","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0094452705,0.0011843148,0.352767,0.0007317822,0.00021797564,0.00033164173,0.16972889,0.4572497,0.008343463],"genre_scores_gemma":[0.05827119,0.0013369323,0.52816325,0.00079915125,0.00014701438,0.0016324738,0.36890155,0.03256449,0.008183976],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9992101,0.0001607854,0.00007445182,0.00022112444,0.00028683772,0.000046614045],"domain_scores_gemma":[0.99810445,0.0012963045,0.00016421467,0.00014822267,0.00017912277,0.00010773365],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016790678,0.0021172352,0.0010265162,0.0035103823,0.0010378427,0.0018429612,0.0018882144,0.0013281937,0.05060483],"category_scores_gemma":[0.00481487,0.0008968825,0.001275424,0.0028446838,0.00051692815,0.0018759827,0.0025488252,0.0018187785,0.028662996],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011527052,0.00026709822,0.006147362,0.0050684214,0.000358128,0.001653835,0.0005541497,0.009317361,0.031382676,0.007864692,0.6832716,0.252962],"study_design_scores_gemma":[0.00083731697,0.00024905257,0.0083167525,0.0009234401,0.00020840995,0.002916162,0.0006110048,0.18426114,0.055186797,0.055989645,0.6901953,0.0003049396],"about_ca_topic_score_codex":0.0019016855,"about_ca_topic_score_gemma":0.0030842111,"teacher_disagreement_score":0.05060483,"about_ca_system_score_codex":0.00064398936,"about_ca_system_score_gemma":0.0013716507,"threshold_uncertainty_score":0.16929007},"labels":[],"label_agreement":null},{"id":"W4391897827","doi":"10.1093/bioinformatics/btae066","title":"VariantDetective: an accurate all-in-one pipeline for detecting consensus bacterial SNPs and SVs","year":2024,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canadian Food Inspection Agency","funders":"Canadian Food Inspection Agency","keywords":"MIT License; Single-nucleotide polymorphism; Source code; Computational biology; Benchmarking; Pipeline (software); SNP; Biology; Computer science; Genetics; Data mining; Software; Gene; Genotype; Programming language","score_opus":0.036520477259258666,"score_gpt":0.28541033089839296,"score_spread":0.2488898536391343,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4391897827","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.019577837,0.0014657899,0.5657493,0.00074371434,0.0006337637,0.00058312074,0.03767164,0.3695935,0.003981346],"genre_scores_gemma":[0.10876593,0.00068049954,0.7454042,0.001286423,0.00022836134,0.0009507646,0.098939754,0.037910644,0.0058334437],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9966132,0.000547398,0.00028920575,0.0013672211,0.0009003649,0.00028272733],"domain_scores_gemma":[0.9956566,0.0022076482,0.0003567899,0.00086502684,0.000631989,0.00028190404],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0045620645,0.0030772372,0.0017279341,0.0028192836,0.0016914295,0.0031660106,0.0044456543,0.0022185228,0.01616674],"category_scores_gemma":[0.017750802,0.0018405732,0.0023520407,0.0018207497,0.0008603893,0.0022618747,0.004225347,0.0026464108,0.013498701],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.004205602,0.0004597677,0.032453317,0.0030266189,0.0019705133,0.0019049247,0.0012941412,0.02063839,0.096051,0.010863083,0.37766033,0.44947228],"study_design_scores_gemma":[0.0014282028,0.0006965146,0.022668183,0.00050185533,0.00082967704,0.0046047918,0.00054071384,0.4568245,0.18718293,0.04976547,0.2738507,0.0011065082],"about_ca_topic_score_codex":0.004970082,"about_ca_topic_score_gemma":0.0088815745,"teacher_disagreement_score":0.01616674,"about_ca_system_score_codex":0.0008768849,"about_ca_system_score_gemma":0.002751589,"threshold_uncertainty_score":0.05408317},"labels":[],"label_agreement":null},{"id":"W4392103664","doi":"10.1093/bioinformatics/btae067","title":"Phenotype prediction from single-cell RNA-seq data using attention-based neural networks","year":2024,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Single-cell and spatial transcriptomics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":27,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"BC Cancer Agency; University of British Columbia; Simon Fraser University","funders":"Natural Sciences and Engineering Research Council of Canada; Canadian Institutes of Health Research; Cancer Research Society","keywords":"Artificial neural network; Phenotype; Computer science; RNA-Seq; Artificial intelligence; Computational biology; Deep neural networks; Software; Machine learning; Data mining; Pattern recognition (psychology); Biology; Gene; Genetics; Transcriptome; Gene expression","score_opus":0.03938044047685513,"score_gpt":0.2431787984064393,"score_spread":0.20379835792958417,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4392103664","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.5175125,0.002573175,0.46548647,0.0010101228,0.00022136043,0.00021360462,0.0029990952,0.0069298414,0.0030538505],"genre_scores_gemma":[0.9158055,0.0004104519,0.07578191,0.00061294134,0.00012328304,0.0001896737,0.004509474,0.00010498008,0.0024618332],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99978393,0.000034065986,0.000011307494,0.00010789959,0.000029285444,0.000033472723],"domain_scores_gemma":[0.9993994,0.00031733292,0.0000714928,0.000041098345,0.00012539065,0.00004536024],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006129263,0.0013157837,0.0006881226,0.0007072867,0.00024033767,0.00052471436,0.0008862016,0.0008316721,0.00080181495],"category_scores_gemma":[0.0013829567,0.00032988875,0.00083376997,0.0004439555,0.00033660445,0.0005386999,0.0005048727,0.0009594032,0.0003453171],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00085215893,0.00038791815,0.03729312,0.00029247729,0.00031083106,0.00047741804,0.00012879327,0.6279231,0.055651706,0.0009983777,0.00664742,0.2690367],"study_design_scores_gemma":[0.000007432691,0.000036374928,0.0018844844,0.0000048515076,0.000018743574,0.000021062635,0.0000073386195,0.9943297,0.0028440324,0.0006626256,0.00017688013,0.000006391897],"about_ca_topic_score_codex":0.008244507,"about_ca_topic_score_gemma":0.009541488,"teacher_disagreement_score":0.008244507,"about_ca_system_score_codex":0.00063260004,"about_ca_system_score_gemma":0.00051734026,"threshold_uncertainty_score":0.016393065},"labels":[],"label_agreement":null},{"id":"W4392563655","doi":"10.1093/bioinformatics/btae136","title":"EpiVar Browser: advanced exploration of epigenomics data under controlled access","year":2024,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Epigenetics and DNA Methylation","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"National Institute of General Medical Sciences; Genome Canada; Canadian Institutes of Health Research; Alliance de recherche numérique du Canada; Ministry of Education, Culture, Sports, Science and Technology; National Institute of Diabetes and Digestive and Kidney Diseases; McGill University","keywords":"Computer science; Epigenomics; Software; Data exploration; World Wide Web; Programming language; Data mining; Biology; Visualization","score_opus":0.06283079410161505,"score_gpt":0.339123684800046,"score_spread":0.276292890698431,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4392563655","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.012480749,0.0012082226,0.24846093,0.0019675326,0.00037494727,0.00064300536,0.27970034,0.43171424,0.023450034],"genre_scores_gemma":[0.08237298,0.001384247,0.35334176,0.0013986432,0.0002111112,0.0021832553,0.44502622,0.10313754,0.010944203],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9983846,0.0004662275,0.0001716114,0.00040437616,0.00044981725,0.00012345922],"domain_scores_gemma":[0.9936466,0.0033736245,0.00033328676,0.0012818122,0.0008550069,0.0005096796],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.00778722,0.0016291079,0.0010897586,0.0035746626,0.0009729433,0.0036491149,0.0029915087,0.0012925066,0.05783517],"category_scores_gemma":[0.015530001,0.001325004,0.0012961933,0.0032847188,0.00057267625,0.0036556064,0.00588687,0.0023805255,0.019315442],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0015456482,0.00018591616,0.013460905,0.0021314756,0.0005332272,0.0010373904,0.0020400863,0.0045572063,0.012483107,0.02518,0.8532459,0.08359899],"study_design_scores_gemma":[0.00067303714,0.00013481027,0.015328205,0.0010286876,0.0002776334,0.0012586837,0.0009082986,0.051606886,0.021857288,0.054792155,0.85177666,0.00035770598],"about_ca_topic_score_codex":0.0074563865,"about_ca_topic_score_gemma":0.012599824,"teacher_disagreement_score":0.9970085,"about_ca_system_score_codex":0.0007312472,"about_ca_system_score_gemma":0.0027014536,"threshold_uncertainty_score":0.19347787},"labels":[],"label_agreement":null},{"id":"W4392564168","doi":"10.1093/bioinformatics/btae135","title":"iNGNN-DTI: prediction of drug–target interaction with interpretable nested graph neural network and pretrained molecule models","year":2024,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Computational Drug Discovery Methods","field":"Computer Science","cited_by":45,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Children’s Health Research Institute; Lawson Health Research Institute; University of Toronto; University of Manitoba; Western University","funders":"Canadian Institutes of Health Research; Natural Sciences and Engineering Research Council of Canada; Canada Research Chairs","keywords":"Discriminative model; Computer science; Machine learning; Artificial intelligence; Graph; ENCODE; Generalization; Artificial neural network; Feature (linguistics); Benchmark (surveying); Interaction network; Set (abstract data type); Pattern recognition (psychology); Theoretical computer science","score_opus":0.017053019806671173,"score_gpt":0.2544432196605,"score_spread":0.2373901998538288,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4392564168","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.3106443,0.005089175,0.6475235,0.0023059223,0.00044100927,0.00030019233,0.0067312173,0.015623419,0.011341226],"genre_scores_gemma":[0.84444594,0.00086630293,0.13591021,0.00073501497,0.00009592397,0.00025220113,0.01051018,0.0003433234,0.0068408013],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99975747,0.000042764448,0.000011077829,0.00009916792,0.00005339814,0.000036156565],"domain_scores_gemma":[0.99948114,0.00023409637,0.00007669581,0.00007492365,0.000087505294,0.000045711506],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00061699096,0.001663628,0.00095881236,0.0008983668,0.00031220657,0.0006669546,0.0019425008,0.0016605377,0.0023880044],"category_scores_gemma":[0.0021225458,0.00053007156,0.0012497844,0.00080799503,0.00048896956,0.0013564089,0.0006806166,0.0020732335,0.00070986047],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00013793453,0.00014449818,0.002960264,0.00011070727,0.000107498025,0.00011870578,0.00001757614,0.9395982,0.0021627238,0.0022835792,0.005775445,0.04658283],"study_design_scores_gemma":[0.000003901725,0.000014549492,0.00012134786,0.0000025824836,0.000005644941,0.000008044501,0.000001143516,0.9982167,0.00038358607,0.0010522746,0.0001873231,0.0000028329052],"about_ca_topic_score_codex":0.018149847,"about_ca_topic_score_gemma":0.027132863,"teacher_disagreement_score":0.018149847,"about_ca_system_score_codex":0.0019109106,"about_ca_system_score_gemma":0.0012374977,"threshold_uncertainty_score":0.036088407},"labels":[],"label_agreement":null},{"id":"W4394582173","doi":"10.1093/bioinformatics/btae190","title":"Improving the performance of supervised deep learning for regulatory genomics using phylogenetic augmentation","year":2024,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":12,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"Natural Sciences and Engineering Research Council of Canada; Canada Research Chairs","keywords":"Genomics; Phylogenetic tree; Computer science; Deep learning; Source code; Set (abstract data type); Artificial intelligence; Sequence (biology); Function (biology); Machine learning; Code (set theory); Computational genomics; Functional genomics; Biological data; Computational biology; Genome; Biology; Bioinformatics; Genetics; Gene","score_opus":0.008751119482387123,"score_gpt":0.22010876798226825,"score_spread":0.21135764849988112,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4394582173","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.28673258,0.0018482214,0.6811692,0.0023289507,0.00026298536,0.0001661156,0.0018111155,0.019769968,0.005910767],"genre_scores_gemma":[0.76231486,0.00036799922,0.22752075,0.0007719986,0.000096708725,0.00023094034,0.0055675576,0.000512671,0.0026165838],"study_design_codex":"simulation_or_modeling","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9991049,0.00035126266,0.000055163284,0.0002267359,0.0001826493,0.00007924294],"domain_scores_gemma":[0.99490803,0.0031705368,0.00025444123,0.0007803787,0.0007184675,0.00016819303],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0029527391,0.0016594407,0.0008388522,0.0005907014,0.00046104586,0.00087134575,0.002009815,0.0014705537,0.0026306666],"category_scores_gemma":[0.010949333,0.000554782,0.0008976229,0.000617919,0.0008609536,0.0020308546,0.0014985407,0.0028072202,0.0013491708],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00060144026,0.00054024684,0.0071776477,0.00028515104,0.00015269339,0.00013294132,0.00013719901,0.75150484,0.017249824,0.004515399,0.008062038,0.20964056],"study_design_scores_gemma":[0.000010034403,0.000028184535,0.00022657786,0.000009211889,0.000006344354,0.000008683736,0.000006371767,0.9936923,0.003173374,0.0025026998,0.00033152555,0.0000046677974],"about_ca_topic_score_codex":0.0047549605,"about_ca_topic_score_gemma":0.005436358,"teacher_disagreement_score":0.0047549605,"about_ca_system_score_codex":0.0011757072,"about_ca_system_score_gemma":0.0016433556,"threshold_uncertainty_score":0.015615761},"labels":[],"label_agreement":null},{"id":"W4394675060","doi":"10.1093/bioinformatics/btae192","title":"VirusPredictor: XGBoost-based software to predict virus-related sequences in human data","year":2024,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Centre Hospitalier Universitaire Sainte-Justine","funders":"National Institute of Allergy and Infectious Diseases; National Institutes of Health; Open Medicine Foundation; U.S. Department of Defense","keywords":"Python (programming language); Contig; Genome; Metagenomics; Biology; Sanger sequencing; Computational biology; Software; Genetics; DNA sequencing; Genomics; Human genome; Computer science; Gene","score_opus":0.023752742637846082,"score_gpt":0.274237705772119,"score_spread":0.2504849631342729,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4394675060","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.065780565,0.0019574584,0.5753656,0.0004434877,0.0004062424,0.0006663376,0.024392674,0.32763997,0.0033475985],"genre_scores_gemma":[0.16748302,0.00077997573,0.77286124,0.0006970678,0.00012445937,0.0014706299,0.0399255,0.011799518,0.0048586614],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9994331,0.0001419159,0.000050310697,0.00019629102,0.00012940039,0.000048954167],"domain_scores_gemma":[0.9992812,0.00046886844,0.00006757786,0.0000584512,0.000087423396,0.00003651063],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016720925,0.0017037968,0.000888413,0.001427835,0.00047796522,0.0007304971,0.0016897508,0.0012072478,0.008169247],"category_scores_gemma":[0.003099554,0.0009246044,0.0014586409,0.0010240113,0.000338104,0.0007460208,0.0007178762,0.0016462761,0.0041586366],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0023724877,0.0006118831,0.022534555,0.0025509517,0.0016228991,0.0006022916,0.0004751324,0.26801458,0.028285291,0.004701976,0.17461407,0.49361384],"study_design_scores_gemma":[0.000116127136,0.00012273356,0.0025553359,0.00005366365,0.000052573934,0.000100856734,0.000029641476,0.97324806,0.008764538,0.0026867061,0.012234893,0.00003497663],"about_ca_topic_score_codex":0.0051780273,"about_ca_topic_score_gemma":0.0063274195,"teacher_disagreement_score":0.008169247,"about_ca_system_score_codex":0.00064417475,"about_ca_system_score_gemma":0.0012403759,"threshold_uncertainty_score":0.02732879},"labels":[],"label_agreement":null},{"id":"W4394745991","doi":"10.1093/bioinformatics/btae182","title":"Pacybara: accurate long-read sequencing for barcoded mutagenized allelic libraries","year":2024,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Rare Diseases","field":"Biochemistry, Genetics and Molecular Biology","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Lunenfeld-Tanenbaum Research Institute; University of Toronto","funders":"National Institute of General Medical Sciences; National Human Genome Research Institute; National Heart, Lung, and Blood Institute; Canadian Institutes of Health Research; National Institutes of Health","keywords":"Barcode; Identification (biology); Biology; Genetics; Computational biology; Identifier; DNA sequencing; Computer science; Gene; Programming language","score_opus":0.021092162970628367,"score_gpt":0.2625775321882256,"score_spread":0.24148536921759722,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4394745991","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.009344763,0.0019953381,0.66469306,0.00079407304,0.00051235483,0.0011769297,0.018538248,0.29592073,0.007024468],"genre_scores_gemma":[0.032335173,0.0013626438,0.84214246,0.0012218902,0.00018590484,0.0022740315,0.050544735,0.060052242,0.009880886],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9935608,0.0012539054,0.0005351379,0.0019460479,0.0022400175,0.00046421192],"domain_scores_gemma":[0.9916653,0.0031686032,0.0014538888,0.0016103123,0.001528436,0.00057337154],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.006496811,0.003165222,0.002458179,0.0026820977,0.0018719854,0.0043183337,0.007091471,0.00351533,0.023363953],"category_scores_gemma":[0.022412807,0.0031980034,0.0017581752,0.0022802514,0.0014907271,0.003671309,0.003194432,0.0057012914,0.04093759],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.003836024,0.00047139847,0.0057534,0.005357483,0.0007926131,0.0017904312,0.0016754874,0.011306414,0.3586726,0.022423541,0.28289452,0.30502608],"study_design_scores_gemma":[0.0004188009,0.00066056155,0.003955364,0.0007929997,0.00025409905,0.0018777554,0.00017005409,0.057267625,0.54556036,0.011007177,0.37737584,0.0006593858],"about_ca_topic_score_codex":0.0028756,"about_ca_topic_score_gemma":0.0041928524,"teacher_disagreement_score":0.023363953,"about_ca_system_score_codex":0.0012410466,"about_ca_system_score_gemma":0.002772025,"threshold_uncertainty_score":0.07816017},"labels":[],"label_agreement":null},{"id":"W4394759342","doi":"10.1093/bioinformatics/btae161","title":"phylaGAN: data augmentation through conditional GANs and autoencoders for improving disease prediction accuracy using microbiome data","year":2024,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gut microbiota and health","field":"Biochemistry, Genetics and Molecular Biology","cited_by":22,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Princess Margaret Cancer Centre; Public Health Ontario; University of Toronto; University Health Network","funders":"","keywords":"Autoencoder; Machine learning; Artificial intelligence; Computer science; Microbiome; Deep learning; Generative grammar; Subspace topology; Pattern recognition (psychology); Bioinformatics","score_opus":0.0742748241898758,"score_gpt":0.35470221034668203,"score_spread":0.28042738615680624,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4394759342","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.15794137,0.0024839747,0.8164339,0.0020950795,0.0004477288,0.00020250103,0.003971012,0.012931066,0.0034932778],"genre_scores_gemma":[0.78655016,0.0007260118,0.19754292,0.0012628501,0.00018596406,0.00029998968,0.009434559,0.00042300404,0.0035746016],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9994252,0.0002676605,0.000028009192,0.0001405485,0.00008490653,0.000053655538],"domain_scores_gemma":[0.9981141,0.0011993535,0.00012716712,0.00026305226,0.00021604252,0.00008024729],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0021479803,0.0011721328,0.0007438946,0.0006220924,0.00025558867,0.0006065282,0.0013298715,0.00083831884,0.0017296573],"category_scores_gemma":[0.0054295305,0.00039834387,0.0009186092,0.00041096078,0.0004947881,0.0008667706,0.0013314221,0.0019720662,0.0008321274],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006071656,0.00033234907,0.017450532,0.0002065015,0.0002977134,0.00026703253,0.00011435559,0.7650415,0.0071075256,0.002938644,0.016443213,0.18919347],"study_design_scores_gemma":[0.000014112564,0.000045689743,0.0005579022,0.000017191924,0.000013045429,0.000030765495,0.0000068859213,0.9952473,0.0015571484,0.0019180003,0.00058296946,0.000008866962],"about_ca_topic_score_codex":0.004301873,"about_ca_topic_score_gemma":0.005926781,"teacher_disagreement_score":0.004301873,"about_ca_system_score_codex":0.00053953903,"about_ca_system_score_gemma":0.0007095788,"threshold_uncertainty_score":0.011359751},"labels":[],"label_agreement":null},{"id":"W4394769874","doi":"10.1093/bioinformatics/btae196","title":"Effect of tokenization on transformers for biological sequences","year":2024,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Machine Learning in Bioinformatics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":41,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Tel Aviv University; Azrieli Foundation; Israel Science Foundation","keywords":"Lexical analysis; Computer science; Artificial intelligence; Natural language processing; Biological network; Inference; Security token; Machine learning; Computational biology; Biology","score_opus":0.009661397653894933,"score_gpt":0.2905729843578977,"score_spread":0.28091158670400274,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4394769874","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8267384,0.002184932,0.14381316,0.0016953329,0.00113564,0.00028504265,0.0013828943,0.017367458,0.005397153],"genre_scores_gemma":[0.9041687,0.000547124,0.08826085,0.0006207575,0.00012272075,0.00013878502,0.0029067225,0.0011791023,0.0020551677],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99560237,0.0014663456,0.00057623646,0.0012131237,0.00060424715,0.00053770107],"domain_scores_gemma":[0.95991,0.029617425,0.0020544494,0.0050767534,0.0023846377,0.0009567727],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0072134407,0.0020253025,0.0011289486,0.0012509129,0.00082785386,0.0022296314,0.002572385,0.0013817481,0.0041327844],"category_scores_gemma":[0.040211,0.00061819615,0.0010699553,0.0011808313,0.0020077818,0.0069100726,0.0020413257,0.002950038,0.0015673924],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.011671853,0.0015814147,0.024376765,0.0015436871,0.00057826575,0.0010332826,0.0006535806,0.4199114,0.058404826,0.015631061,0.014233545,0.45038036],"study_design_scores_gemma":[0.00041752882,0.0014204131,0.00218717,0.00009690608,0.00028590785,0.0003331917,0.0004349427,0.88036484,0.09612203,0.013904162,0.004353477,0.00007955692],"about_ca_topic_score_codex":0.005715546,"about_ca_topic_score_gemma":0.006271221,"teacher_disagreement_score":0.0072134407,"about_ca_system_score_codex":0.0018148895,"about_ca_system_score_gemma":0.0032468739,"threshold_uncertainty_score":0.03814882},"labels":[],"label_agreement":null},{"id":"W4396561183","doi":"10.1093/bioinformatics/btae295","title":"SharePro: an accurate and efficient genetic colocalization method accounting for multiple causal signals","year":2024,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":21,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto; Université de Montréal; McGill University; Montreal Heart Institute","funders":"Canada First Research Excellence Fund","keywords":"Colocalization; Genome-wide association study; Linkage disequilibrium; Computational biology; Computer science; Locus (genetics); Genetic association; Statistical power; False positive paradox; Biology; Machine learning; Statistics; Genetics; Neuroscience; Allele; Mathematics; Genotype; Single-nucleotide polymorphism; Gene; Haplotype","score_opus":0.025460759315474625,"score_gpt":0.32263423724643114,"score_spread":0.2971734779309565,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4396561183","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00747467,0.00020970868,0.97793055,0.00026450466,0.000053140713,0.00007197972,0.001129154,0.011999944,0.0008663194],"genre_scores_gemma":[0.17040947,0.00024860605,0.81306463,0.00048555902,0.00012397552,0.000441893,0.004260409,0.0068901074,0.0040753335],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9986333,0.00050406856,0.00006814296,0.00042387948,0.00027907855,0.00009152313],"domain_scores_gemma":[0.996473,0.002170398,0.00025642317,0.00059735536,0.000346795,0.0001560439],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00370092,0.0012397369,0.0016085437,0.0017069759,0.0010873771,0.0017868723,0.0028223833,0.0014303848,0.013357379],"category_scores_gemma":[0.012669672,0.0010303952,0.0020762782,0.0013230338,0.0011365798,0.0015541309,0.0028455483,0.0017429505,0.0028002106],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014191101,0.00027592565,0.028995173,0.0015960316,0.002008999,0.0018287871,0.00083588396,0.260168,0.03204617,0.08666222,0.08469976,0.49946395],"study_design_scores_gemma":[0.0003081559,0.000077448625,0.0028608453,0.000057764213,0.00017743626,0.00074403844,0.00007996702,0.89891016,0.007537022,0.07059124,0.018550867,0.00010505755],"about_ca_topic_score_codex":0.0069112866,"about_ca_topic_score_gemma":0.010489785,"teacher_disagreement_score":0.013357379,"about_ca_system_score_codex":0.0007407527,"about_ca_system_score_gemma":0.0027621514,"threshold_uncertainty_score":0.044684887},"labels":[],"label_agreement":null},{"id":"W4396745594","doi":"10.1093/bioinformatics/btae444","title":"A curated rotamer library for common post-translational modifications of proteins","year":2024,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Protein Structure and Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Hospital for Sick Children; University of Toronto","funders":"National Institutes of Health; National Institute of General Medical Sciences; Natural Sciences and Engineering Research Council of Canada; Canada Research Chairs","keywords":"Protein Data Bank (RCSB PDB); Dihedral angle; Conformational isomerism; Computational biology; Intrinsically disordered proteins; Computer science; Protein structure; Conformational ensembles; Database; Bioinformatics; Chemistry; Biology; Biochemistry; Molecule","score_opus":0.009658113960087922,"score_gpt":0.24955730278967478,"score_spread":0.23989918882958686,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4396745594","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.07304556,0.0019716215,0.47980163,0.000439525,0.0005105408,0.0008948238,0.1724798,0.25830913,0.012547364],"genre_scores_gemma":[0.09276269,0.0011149124,0.4849402,0.000563072,0.00014477526,0.0018749818,0.37533295,0.03799385,0.0052726646],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9982969,0.00038629727,0.00017089752,0.0005083415,0.0005010609,0.00013659563],"domain_scores_gemma":[0.9972562,0.00075321394,0.00028131515,0.00087728957,0.0006066431,0.00022542328],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0026373335,0.0023515632,0.0016522958,0.0022263518,0.0013841308,0.0013468587,0.0031641584,0.0012574015,0.015370925],"category_scores_gemma":[0.008156808,0.0010520398,0.0017950458,0.002933017,0.0006199893,0.0016129015,0.0021242555,0.0024552646,0.015518785],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0031237637,0.00097385334,0.0139513565,0.006623116,0.00074526644,0.0024034288,0.0009009421,0.09154948,0.23493463,0.020511039,0.38319358,0.24108954],"study_design_scores_gemma":[0.0007786236,0.0008507565,0.006621564,0.0005835181,0.0005171055,0.0025219826,0.00014205932,0.25573224,0.3188924,0.025732026,0.38722458,0.00040318907],"about_ca_topic_score_codex":0.0011095416,"about_ca_topic_score_gemma":0.0024782668,"teacher_disagreement_score":0.015370925,"about_ca_system_score_codex":0.00083838234,"about_ca_system_score_gemma":0.0024140957,"threshold_uncertainty_score":0.051420927},"labels":[],"label_agreement":null},{"id":"W4399117787","doi":"10.1093/bioinformatics/btae342","title":"Representations of lipid nanoparticles using large language models for transfection efficiency prediction","year":2024,"lang":"en","type":"article","venue":"Bioinformatics","topic":"RNA Interference and Gene Delivery","field":"Biochemistry, Genetics and Molecular Biology","cited_by":27,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Sanofi (Canada)","funders":"Sanofi","keywords":"Payload (computing); Transfection; Nanoparticle; Gene delivery; Computer science; Messenger RNA; Chemistry; Nanotechnology; Computational biology; Cell biology; Biochemistry; Biology; Materials science; Gene; Computer network","score_opus":0.01932988650558984,"score_gpt":0.2876306062438401,"score_spread":0.2683007197382503,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4399117787","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.08094772,0.0008010317,0.899934,0.0009446326,0.00015624678,0.00013902682,0.0035443692,0.011031284,0.0025017408],"genre_scores_gemma":[0.6570758,0.0007220444,0.31882447,0.00068047136,0.00015818219,0.00077566074,0.015179127,0.0012651033,0.005319096],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99942493,0.00021119599,0.00003843363,0.00015941485,0.0000984789,0.00006747552],"domain_scores_gemma":[0.9979665,0.0014067007,0.00012383843,0.00011750252,0.00031428604,0.00007112392],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012213972,0.0013379107,0.00078756985,0.0011374208,0.0003847114,0.001259408,0.0010960256,0.001147864,0.0032882926],"category_scores_gemma":[0.005208115,0.00041062673,0.0017553688,0.00059900637,0.00042559704,0.0011713437,0.00080249686,0.0015170664,0.0027127904],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005988379,0.00040302775,0.005150347,0.00040594846,0.00018580005,0.000314853,0.00016633516,0.77434236,0.020514358,0.011928201,0.01296791,0.17302196],"study_design_scores_gemma":[0.000008332568,0.000022136528,0.000095546624,0.0000062716567,0.000009268175,0.00001419919,0.0000075025628,0.99341416,0.0017117012,0.004165567,0.0005387319,0.000006499487],"about_ca_topic_score_codex":0.004670535,"about_ca_topic_score_gemma":0.005073801,"teacher_disagreement_score":0.004670535,"about_ca_system_score_codex":0.0011934179,"about_ca_system_score_gemma":0.0010772147,"threshold_uncertainty_score":0.011000454},"labels":[],"label_agreement":null},{"id":"W4400128023","doi":"10.1093/bioinformatics/btae229","title":"Median and small parsimony problems on RNA trees","year":2024,"lang":"en","type":"article","venue":"Bioinformatics","topic":"RNA and protein synthesis mechanisms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Sherbrooke","funders":"Natural Sciences and Engineering Research Council of Canada; Canada Research Chairs; Agence Nationale de la Recherche; Institut Polytechnique de Paris","keywords":"RNA; Biology; Computational biology; Nucleic acid secondary structure; Nucleic acid structure; Non-coding RNA; Evolutionary biology; Homologous chromosome; Genetics; RNA editing; Protein secondary structure; Structural motif; Gene","score_opus":0.016641647047241314,"score_gpt":0.22389361941666736,"score_spread":0.20725197236942605,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4400128023","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.067697056,0.0026708064,0.91589797,0.003679349,0.0001298781,0.00023820576,0.002081149,0.0009664057,0.006639185],"genre_scores_gemma":[0.36466682,0.0017920353,0.61920375,0.00061237824,0.00046390318,0.00066837395,0.004701919,0.0004992402,0.0073915855],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9975228,0.0009918835,0.00014434304,0.0006897743,0.00039654458,0.00025457767],"domain_scores_gemma":[0.9706588,0.026354788,0.0010279664,0.0007801618,0.0006474193,0.0005308506],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002571334,0.0016272864,0.002708361,0.001823183,0.0017054416,0.0030574242,0.002994147,0.0032733479,0.008289599],"category_scores_gemma":[0.020952446,0.00087760336,0.0016787482,0.003752885,0.0020858124,0.006636318,0.0023508128,0.003624963,0.001080972],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006704529,0.00027630432,0.0024537367,0.001140629,0.0002040064,0.0004957604,0.00073870615,0.6360845,0.0022383092,0.19962776,0.019641787,0.13642809],"study_design_scores_gemma":[0.00008351276,0.00005354372,0.00032104275,0.00004027214,0.00002232539,0.00013694832,0.000108532135,0.6389727,0.0005912566,0.3573901,0.002261811,0.000017946262],"about_ca_topic_score_codex":0.0024471832,"about_ca_topic_score_gemma":0.002674435,"teacher_disagreement_score":0.008289599,"about_ca_system_score_codex":0.0026325001,"about_ca_system_score_gemma":0.0015842196,"threshold_uncertainty_score":0.027731478},"labels":[],"label_agreement":null},{"id":"W4400140691","doi":"10.1093/bioinformatics/btae252","title":"Floria: fast and accurate strain haplotyping in metagenomes","year":2024,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":16,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Metagenomics; Strain (injury); Computer science; Pipeline (software); Workflow; Computational biology; Set (abstract data type); Haplotype; Genome; Data mining; Biology; Genetics; Database; Gene; Allele; Operating system; Programming language","score_opus":0.015796229212360697,"score_gpt":0.2470699199430443,"score_spread":0.2312736907306836,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4400140691","genre_codex":"software","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.04665708,0.0014124068,0.26801175,0.0007833575,0.00036057422,0.00047241847,0.037238,0.64054155,0.004522884],"genre_scores_gemma":[0.16393557,0.0005727552,0.7024438,0.0011137286,0.00014269953,0.0015936821,0.07428153,0.05244864,0.003467627],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9981217,0.00026581288,0.00012396628,0.00074301177,0.0005333394,0.00021215685],"domain_scores_gemma":[0.9976394,0.00076854805,0.00026760204,0.0007802092,0.00029067046,0.00025357903],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0036473402,0.002267199,0.0016175391,0.0017885342,0.0010489493,0.0022173738,0.00373768,0.0015264824,0.011283732],"category_scores_gemma":[0.008413673,0.0021305983,0.002544138,0.0008995647,0.0008911148,0.0025426345,0.0041700224,0.0032049092,0.012485095],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.005487414,0.00045753596,0.032896325,0.004125105,0.0019053858,0.0014278494,0.0019969386,0.02666217,0.29840916,0.010227996,0.30254948,0.31385458],"study_design_scores_gemma":[0.0020434184,0.0010646255,0.030948475,0.0009450301,0.00042909014,0.0021941524,0.0005432086,0.3710434,0.31612545,0.022163155,0.25130266,0.0011973822],"about_ca_topic_score_codex":0.0026449412,"about_ca_topic_score_gemma":0.004229635,"teacher_disagreement_score":0.011283732,"about_ca_system_score_codex":0.000981077,"about_ca_system_score_gemma":0.0015215948,"threshold_uncertainty_score":0.0377478},"labels":[],"label_agreement":null},{"id":"W4400140816","doi":"10.1093/bioinformatics/btae263","title":"Adaptive digital tissue deconvolution","year":2024,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Single-cell and spatial transcriptomics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Bundesministerium für Bildung und Forschung; Deutsche Forschungsgemeinschaft; European Commission; Trond Mohn stiftelse; Ministère de l'Économie, de la Science et de l'Innovation - Québec; California Department of Fish and Game","keywords":"Deconvolution; Computer science; Blind deconvolution; Algorithm; Artificial intelligence","score_opus":0.013059272636809205,"score_gpt":0.23184750945628338,"score_spread":0.21878823681947418,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4400140816","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0023866424,0.00015711706,0.9939528,0.00013877159,0.000037261423,0.000025596875,0.00024846624,0.0014658648,0.0015874556],"genre_scores_gemma":[0.08492012,0.00064840977,0.90685475,0.0003684698,0.00007786844,0.00012085161,0.0013009501,0.00054791715,0.005160574],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9996357,0.00006192912,0.000022174148,0.00010087025,0.00014940105,0.000029892095],"domain_scores_gemma":[0.99945086,0.00020989511,0.000054375374,0.0001178508,0.00012626099,0.000040813142],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00079593353,0.00078655884,0.00048174016,0.0006936124,0.0002798994,0.0011013431,0.0014040011,0.0010019138,0.0049776714],"category_scores_gemma":[0.0020055587,0.0003555059,0.0008883139,0.00076416787,0.0009443491,0.0010290811,0.0015396761,0.0011476169,0.0026970971],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00039949667,0.00008867528,0.003396849,0.0008477549,0.00013497894,0.00053116924,0.00025352708,0.31482753,0.12683536,0.04872584,0.015573852,0.4883851],"study_design_scores_gemma":[0.000022142996,0.00003194155,0.0005548942,0.000038723803,0.000023142453,0.00054061343,0.000025700194,0.91750956,0.04438847,0.020605441,0.016216524,0.000042923555],"about_ca_topic_score_codex":0.0023158907,"about_ca_topic_score_gemma":0.0017630829,"teacher_disagreement_score":0.0049776714,"about_ca_system_score_codex":0.0007291632,"about_ca_system_score_gemma":0.0009825849,"threshold_uncertainty_score":0.016651988},"labels":[],"label_agreement":null},{"id":"W4400140828","doi":"10.1093/bioinformatics/btae237","title":"Predicting protein functions using positive-unlabeled ranking with ontology-based priors","year":2024,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Biomedical Text Mining and Ontologies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"King Abdullah University of Science and Technology","keywords":"Computer science; Prior probability; Benchmark (surveying); Classifier (UML); Ranking (information retrieval); Artificial intelligence; Machine learning; Data mining; Source code; Gene ontology; Function (biology); Ontology; Pattern recognition (psychology); Bayesian probability","score_opus":0.015018811350751586,"score_gpt":0.2557952577102794,"score_spread":0.2407764463595278,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4400140828","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.15360378,0.0021703406,0.80591136,0.0013277704,0.00018800556,0.00031375364,0.011715661,0.016670913,0.008098375],"genre_scores_gemma":[0.58115494,0.00053224416,0.37714922,0.0006529605,0.0002530995,0.00032585798,0.034498088,0.000983884,0.004449745],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9969439,0.00084357854,0.0001438835,0.0009123813,0.0009419553,0.00021426743],"domain_scores_gemma":[0.99402195,0.002835857,0.00049499265,0.0014556252,0.0009391596,0.0002523677],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0033715232,0.0017754812,0.001682004,0.004324429,0.0010423572,0.0020429757,0.0029522325,0.002174607,0.002093424],"category_scores_gemma":[0.013091617,0.0004940143,0.0013954185,0.0024608255,0.0009447964,0.0035875363,0.0019864966,0.0018802519,0.0023965703],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013934359,0.0017441652,0.035308644,0.0008947916,0.0003836644,0.0004503644,0.00029970478,0.32199413,0.027199559,0.026412996,0.06615528,0.5177633],"study_design_scores_gemma":[0.000039123173,0.0000760882,0.0023496267,0.000031500575,0.000034878,0.00013963114,0.000051282408,0.95115995,0.0055457223,0.036375854,0.0041674674,0.000028820754],"about_ca_topic_score_codex":0.005282251,"about_ca_topic_score_gemma":0.013880524,"teacher_disagreement_score":0.005282251,"about_ca_system_score_codex":0.0015582555,"about_ca_system_score_gemma":0.0017158174,"threshold_uncertainty_score":0.01783055},"labels":[],"label_agreement":null},{"id":"W4400497719","doi":"10.1093/bioinformatics/btae405","title":"Geometric epitope and paratope prediction","year":2024,"lang":"en","type":"article","venue":"Bioinformatics","topic":"vaccines and immunoinformatics approaches","field":"Biochemistry, Genetics and Molecular Biology","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"H2020 European Research Council; Sapienza Università di Roma; European Commission","keywords":"Paratope; Epitope; Computer science; Computational biology; Artificial intelligence; Biology; Antibody; Genetics","score_opus":0.012131777674846557,"score_gpt":0.22608040107793448,"score_spread":0.21394862340308793,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4400497719","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.21203622,0.002007206,0.7658036,0.0011621034,0.0001329967,0.00011241176,0.002811471,0.005637751,0.010296221],"genre_scores_gemma":[0.80243915,0.0014101353,0.18567571,0.00039393452,0.00009729237,0.00009514388,0.0060368166,0.0004809299,0.0033708538],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99956745,0.000096628275,0.00001619075,0.0001134998,0.00014881567,0.000057411595],"domain_scores_gemma":[0.99903,0.0003518819,0.00017175249,0.000163564,0.00020714637,0.00007577261],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00066258595,0.0009221929,0.0007484251,0.0006739615,0.0002517382,0.0007850732,0.0010557717,0.00080145366,0.0043508657],"category_scores_gemma":[0.0023793662,0.00025660172,0.0005437544,0.0006505229,0.0006896888,0.0015295445,0.0010572114,0.0011298766,0.0025107337],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0004567949,0.00023016233,0.010199512,0.0006023215,0.000090717724,0.00023604467,0.000061245846,0.7280135,0.036080614,0.02300726,0.011126056,0.18989573],"study_design_scores_gemma":[0.000016756609,0.0000989445,0.0015296753,0.00002158443,0.00001562488,0.0001485799,0.000025623593,0.95567626,0.017945461,0.021342881,0.0031591074,0.000019489424],"about_ca_topic_score_codex":0.0024694048,"about_ca_topic_score_gemma":0.0018686764,"teacher_disagreement_score":0.0043508657,"about_ca_system_score_codex":0.0005653638,"about_ca_system_score_gemma":0.00076034374,"threshold_uncertainty_score":0.014555097},"labels":[],"label_agreement":null},{"id":"W4400597743","doi":"10.1093/bioinformatics/btae448","title":"isolateR: an R package for generating microbial libraries from Sanger sequencing data","year":2024,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Saskatchewan; University of Guelph","funders":"Breakthrough T1D Canada; Canada First Research Excellence Fund; National Institute of General Medical Sciences; Canada Research Chairs","keywords":"Cataloging; Sanger sequencing; Benchmarking; Computer science; Biology; Pipeline (software); Computational biology; Information retrieval; DNA sequencing; World Wide Web; Genetics; Programming language; Gene","score_opus":0.04105578436180538,"score_gpt":0.26225966537620315,"score_spread":0.22120388101439778,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4400597743","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0027648087,0.0012780254,0.3434949,0.0006440642,0.0003818446,0.00059892883,0.10296696,0.5417494,0.0061210026],"genre_scores_gemma":[0.025240079,0.002004332,0.5752556,0.0014359631,0.00021197823,0.0043744766,0.16044082,0.22378962,0.007247198],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9958723,0.0016993032,0.00035870363,0.00091607694,0.0008232585,0.00033040126],"domain_scores_gemma":[0.99202955,0.004789114,0.00096553296,0.0010795281,0.0007356415,0.00040054804],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007494866,0.0055354564,0.004401592,0.0034182763,0.001121037,0.0044773,0.004940022,0.0017132448,0.067802705],"category_scores_gemma":[0.025205752,0.0030060902,0.003502814,0.0028041387,0.0014554787,0.0029133665,0.004481889,0.005333986,0.090989865],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0018592983,0.00017283004,0.004775735,0.0096513275,0.0021098729,0.0009484572,0.00086581864,0.012453493,0.019720098,0.019130826,0.82122356,0.10708875],"study_design_scores_gemma":[0.0016623955,0.0004055817,0.0050355736,0.0016068129,0.0009441566,0.0013404186,0.00023182677,0.06663733,0.027134126,0.05020273,0.84427494,0.0005240709],"about_ca_topic_score_codex":0.0020920357,"about_ca_topic_score_gemma":0.0024483954,"teacher_disagreement_score":0.067802705,"about_ca_system_score_codex":0.0009810681,"about_ca_system_score_gemma":0.0038388423,"threshold_uncertainty_score":0.22682261},"labels":[],"label_agreement":null},{"id":"W4400920210","doi":"10.1093/bioinformatics/btae446","title":"Adaptive space search-based molecular evolution optimization algorithm","year":2024,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Computational Drug Discovery Methods","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Key Research and Development Program of China; Natural Science Foundation of Anhui Province; National Natural Science Foundation of China; Institute for Catastrophic Loss Reduction","keywords":"Computer science; Space (punctuation); Optimization problem; Process (computing); Encoder; Chemical space; Algorithm; Mathematical optimization; Theoretical computer science; Mathematics; Drug discovery; Bioinformatics","score_opus":0.017765906789825977,"score_gpt":0.2777247315030628,"score_spread":0.25995882471323684,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4400920210","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.022504939,0.001042007,0.9670314,0.00039922085,0.00011414816,0.00019982755,0.00014331321,0.0011540304,0.0074110962],"genre_scores_gemma":[0.36579368,0.0007558547,0.62471,0.0004447014,0.00010402189,0.0010734812,0.0006120377,0.00027319795,0.0062330845],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.999516,0.00014757355,0.000027118129,0.00009533354,0.00015690782,0.000056994213],"domain_scores_gemma":[0.9993685,0.00036755975,0.000058867834,0.00003510636,0.00013315376,0.000036853202],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007706166,0.0010881745,0.0015121232,0.0008949323,0.00049764616,0.00064833957,0.0015895005,0.0013836345,0.004177917],"category_scores_gemma":[0.0020341703,0.00047358722,0.0009390895,0.0010041709,0.00069376477,0.0006299846,0.0011102664,0.0013554359,0.000809798],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000576057,0.000052572133,0.0008111904,0.00011383567,0.000055684417,0.00007184025,0.00005461134,0.9262811,0.001512754,0.008585903,0.0020601205,0.060342707],"study_design_scores_gemma":[0.00002588058,0.000024411924,0.00006861462,0.0000074600207,0.000010313854,0.000017334452,0.000004748521,0.99713147,0.00023163576,0.00149918,0.0009746209,0.00000428879],"about_ca_topic_score_codex":0.00508566,"about_ca_topic_score_gemma":0.004533794,"teacher_disagreement_score":0.00508566,"about_ca_system_score_codex":0.0007191813,"about_ca_system_score_gemma":0.0014684267,"threshold_uncertainty_score":0.013976574},"labels":[],"label_agreement":null},{"id":"W4401322700","doi":"10.1093/bioinformatics/btae483","title":"Cellular proliferation biases clonal lineage tracing and trajectory inference","year":2024,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Single-cell and spatial transcriptomics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"Natural Sciences and Engineering Research Council of Canada; Yale University","keywords":"Python (programming language); Computer science; Inference; Tracing; Probabilistic logic; Source code; Skew; Sampling bias; Theoretical computer science; Computational biology; Algorithm; Biology; Artificial intelligence; Statistics; Programming language; Mathematics; Sample size determination","score_opus":0.02444510163836968,"score_gpt":0.2530713318885632,"score_spread":0.2286262302501935,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4401322700","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.11294629,0.000980076,0.87240356,0.0019647533,0.0001992306,0.00014244072,0.0036345362,0.0036867596,0.004042313],"genre_scores_gemma":[0.7804726,0.0009584524,0.20798369,0.000984393,0.00022314735,0.00042267374,0.004810085,0.0011504034,0.0029945842],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9975925,0.0009137086,0.00014473662,0.00070715876,0.0004997599,0.00014218062],"domain_scores_gemma":[0.9644658,0.02844052,0.0026178455,0.0023542563,0.0016304627,0.0004911834],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008295477,0.0007230661,0.00090407324,0.0008094955,0.0013185869,0.0024322765,0.0021494913,0.0015497213,0.00405853],"category_scores_gemma":[0.06013531,0.00065333187,0.0009484184,0.001551327,0.0020485637,0.002208912,0.0019422227,0.0019996804,0.0010565927],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00050628977,0.000109178305,0.109519705,0.0010257522,0.00035079184,0.0006349679,0.00083185197,0.62726045,0.023200192,0.13287914,0.01635237,0.08732933],"study_design_scores_gemma":[0.00004984112,0.00004498156,0.0067237406,0.00012683452,0.00008867355,0.00040214203,0.000082138155,0.8569738,0.016428102,0.110345006,0.008670878,0.00006393114],"about_ca_topic_score_codex":0.013987005,"about_ca_topic_score_gemma":0.012805338,"teacher_disagreement_score":0.013987005,"about_ca_system_score_codex":0.0023505017,"about_ca_system_score_gemma":0.0025389185,"threshold_uncertainty_score":0.043871224},"labels":[],"label_agreement":null},{"id":"W4401666005","doi":"10.1093/bioinformatics/btae517","title":"Spaln3: improvement in speed and accuracy of genome mapping and spliced alignment of protein query sequences","year":2024,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Institute of Genetics; National Institute of Advanced Industrial Science and Technology","keywords":"Vectorization (mathematics); Computer science; SIMD; Genome; Multiple sequence alignment; Sequence (biology); Smith–Waterman algorithm; Dynamic programming; Sequence alignment; Alignment-free sequence analysis; Speedup; Computational biology; Reference genome; Parallel computing; Gene; Algorithm; Genetics; Biology; Peptide sequence","score_opus":0.014101316815504597,"score_gpt":0.23624416687569055,"score_spread":0.22214285006018597,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4401666005","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.15330294,0.0014039072,0.46833885,0.0014148912,0.0006646176,0.000267376,0.017458467,0.34790248,0.009246445],"genre_scores_gemma":[0.19361176,0.00061606144,0.71273845,0.0006884036,0.00013137322,0.00052151235,0.05209671,0.031172588,0.00842316],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99784565,0.00040820308,0.00019485324,0.00060008746,0.0007224964,0.00022866213],"domain_scores_gemma":[0.99782,0.0007395421,0.00015852686,0.00052798865,0.0006088012,0.00014506927],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0036826269,0.0016278614,0.0010715526,0.0012483187,0.0008161414,0.0015016712,0.0027513169,0.0009894191,0.013622514],"category_scores_gemma":[0.0063794646,0.00095039123,0.0013055756,0.0021112834,0.0005735201,0.0023649784,0.0021594493,0.0023953815,0.011947457],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0047196536,0.00049227686,0.01100641,0.0013527226,0.0003944633,0.0005373336,0.00070454815,0.015969733,0.28829834,0.009340821,0.23590524,0.4312785],"study_design_scores_gemma":[0.0009320495,0.0005639064,0.009549367,0.00016773933,0.00018617584,0.0012280688,0.00019702164,0.44758272,0.40653768,0.009215845,0.12353842,0.00030097374],"about_ca_topic_score_codex":0.0053079613,"about_ca_topic_score_gemma":0.0053572557,"teacher_disagreement_score":0.013622514,"about_ca_system_score_codex":0.0012115482,"about_ca_system_score_gemma":0.0019674357,"threshold_uncertainty_score":0.045571864},"labels":[],"label_agreement":null},{"id":"W4401835088","doi":"10.1093/bioinformatics/btae525","title":"FateNet: an integration of dynamical systems and deep learning for cell fate prediction","year":2024,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Single-cell and spatial transcriptomics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; University of Waterloo","funders":"Fonds de recherche du Québec – Nature et technologies","keywords":"Computer science; Deep learning; Artificial intelligence; Machine learning","score_opus":0.012743921212909527,"score_gpt":0.23074258807812947,"score_spread":0.21799866686521996,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4401835088","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.03063241,0.0007616532,0.91265565,0.0010070768,0.00032258144,0.00017047559,0.0063530267,0.04266535,0.0054317256],"genre_scores_gemma":[0.34309462,0.0009092688,0.6263619,0.00079363596,0.00016785474,0.0009815969,0.016672438,0.0040036417,0.007015073],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99986243,0.00003366417,0.000008210782,0.000042052176,0.000040112693,0.000013473897],"domain_scores_gemma":[0.9995648,0.00029379368,0.000022039467,0.000037538266,0.00004314822,0.00003866638],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00068489223,0.0009439776,0.00057894364,0.00074467866,0.00038798695,0.00089897437,0.0011406385,0.00096144545,0.0062483605],"category_scores_gemma":[0.0024038248,0.0005064708,0.0007169969,0.00044146602,0.00048741107,0.00095944054,0.0011854068,0.0013817461,0.0016187484],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00033790167,0.00020264462,0.0039505367,0.00049391156,0.00022002259,0.00031711897,0.00010707247,0.7670421,0.007916603,0.029286826,0.041557625,0.14856766],"study_design_scores_gemma":[0.000013569846,0.000016411796,0.000121607256,0.000008512233,0.0000057357333,0.000015559774,0.0000048693064,0.98543984,0.0009555896,0.011265173,0.0021455623,0.000007514448],"about_ca_topic_score_codex":0.004908288,"about_ca_topic_score_gemma":0.009506968,"teacher_disagreement_score":0.0062483605,"about_ca_system_score_codex":0.00078039983,"about_ca_system_score_gemma":0.0011689673,"threshold_uncertainty_score":0.020902812},"labels":[],"label_agreement":null},{"id":"W4402253411","doi":"10.1093/bioinformatics/btae398","title":"skandiver: a divergence-based analysis tool for identifying intercellular mobile genetic elements","year":2024,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bacteriophages and microbial interactions","field":"Environmental Science","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"The Scarborough Hospital; University of Toronto","funders":"Natural Sciences and Engineering Research Council of Canada; Carnegie Mellon University","keywords":"Mobile genetic elements; Genome; Scalability; Transposable element; Plasmid; Computational biology; Annotation; Biology; Computer science; Genetics; Gene; Database","score_opus":0.017636749436864878,"score_gpt":0.2702004668755518,"score_spread":0.25256371743868694,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4402253411","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.08003789,0.001405779,0.5737954,0.00036097522,0.00019222402,0.00039051916,0.02518442,0.31092724,0.007705543],"genre_scores_gemma":[0.21328615,0.000670588,0.7210939,0.00039098307,0.000055740144,0.0009012542,0.041039683,0.018281402,0.004280228],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99891245,0.00012841904,0.00014078352,0.0003535974,0.00036654843,0.00009827102],"domain_scores_gemma":[0.9977507,0.0012352852,0.00035104508,0.00028865665,0.00025558082,0.000118779055],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0017494763,0.0022868663,0.0014128449,0.004004983,0.0011361277,0.0022121063,0.0023324103,0.0011525127,0.008875507],"category_scores_gemma":[0.0064182244,0.0008687616,0.0020689385,0.0018675668,0.0007464429,0.002951636,0.0031491313,0.0013766662,0.004355917],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.004369661,0.0004603865,0.05410089,0.005803121,0.0013416539,0.0022824185,0.0033077518,0.0306516,0.17286165,0.02347249,0.13268898,0.56865937],"study_design_scores_gemma":[0.00070347084,0.0010381475,0.023583014,0.0008938022,0.0006156495,0.0038991899,0.0011686753,0.5731264,0.17018726,0.043657217,0.18047585,0.00065128884],"about_ca_topic_score_codex":0.002041965,"about_ca_topic_score_gemma":0.003430215,"teacher_disagreement_score":0.008875507,"about_ca_system_score_codex":0.0008921271,"about_ca_system_score_gemma":0.0015471551,"threshold_uncertainty_score":0.029691577},"labels":[],"label_agreement":null},{"id":"W4402253482","doi":"10.1093/bioinformatics/btae391","title":"GGTyper: genotyping complex structural variants using short-read sequencing data","year":2024,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Rare Diseases","field":"Biochemistry, Genetics and Molecular Biology","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Amgen (Canada)","funders":"Berlin Institute of Health","keywords":"Genotyping; Computer science; Computational biology; Genetics; Biology; Genotype; Gene","score_opus":0.09498124191876284,"score_gpt":0.3235446239675812,"score_spread":0.22856338204881838,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4402253482","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.034213517,0.00058104575,0.8297546,0.0004743757,0.00019172745,0.00040396967,0.02099187,0.11077042,0.0026185538],"genre_scores_gemma":[0.14842626,0.00025994907,0.8101222,0.0006908171,0.00014840493,0.0006452715,0.028335858,0.008499031,0.0028721923],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.996046,0.0013080164,0.00023549133,0.0013818041,0.0008786382,0.0001500619],"domain_scores_gemma":[0.9895734,0.007201568,0.0007319966,0.0015426844,0.00059942255,0.00035093806],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004751904,0.0017202909,0.0013309219,0.0014868765,0.00054573314,0.001573814,0.0034241455,0.0020224017,0.009816049],"category_scores_gemma":[0.020196648,0.00087463396,0.001694275,0.0015203535,0.00070589414,0.0015202733,0.0018489566,0.0016094039,0.00537619],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0041930033,0.00042324793,0.06573223,0.0033551813,0.0021706708,0.0017373242,0.00086112367,0.2965826,0.06979901,0.019856928,0.10523706,0.4300516],"study_design_scores_gemma":[0.00065696455,0.0003852729,0.011079212,0.00015211754,0.00017179066,0.0010728022,0.0001019144,0.89906573,0.03373056,0.029348372,0.02400692,0.0002283716],"about_ca_topic_score_codex":0.005237321,"about_ca_topic_score_gemma":0.008285069,"teacher_disagreement_score":0.009816049,"about_ca_system_score_codex":0.0007569686,"about_ca_system_score_gemma":0.0014344335,"threshold_uncertainty_score":0.032837987},"labels":[],"label_agreement":null},{"id":"W4402348973","doi":"10.1093/bioinformatics/btae546","title":"Bin2cell reconstructs cells from high resolution Visium HD data","year":2024,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Single-cell and spatial transcriptomics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":75,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Engineering and Physical Sciences Research Council; Medical Research Council; UK Research and Innovation; Wellcome Trust; Canadian Institute for Advanced Research","keywords":"Python (programming language); Computer science; Software; Documentation; Segmentation; Open source; Image resolution; Artificial intelligence; Data mining; Information retrieval; Computer graphics (images); Computer vision; Operating system","score_opus":0.023994808084351116,"score_gpt":0.23487815531196674,"score_spread":0.21088334722761562,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4402348973","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.07205095,0.00077151356,0.6913163,0.00083994796,0.00034652944,0.00015968022,0.045880873,0.17452465,0.014109546],"genre_scores_gemma":[0.22981994,0.00094890856,0.59874815,0.0012802939,0.00006983724,0.0007498361,0.09572289,0.061396845,0.011263294],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9997335,0.0000246154,0.000013421376,0.00007249828,0.000114655515,0.00004129717],"domain_scores_gemma":[0.99960536,0.00013815892,0.000026076057,0.00010063361,0.00009003721,0.000039707735],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007440616,0.0008313496,0.0006163844,0.0008885027,0.00061447953,0.0018036308,0.0012705439,0.00081490114,0.019327525],"category_scores_gemma":[0.0018232508,0.00060224766,0.0006112415,0.0009235145,0.00036413848,0.00087476097,0.0013034926,0.0010946117,0.00827021],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0012471505,0.0001807,0.012570493,0.0014345335,0.00028503922,0.0007042238,0.0007863271,0.045021694,0.48273835,0.018577907,0.2514961,0.18495743],"study_design_scores_gemma":[0.000101202946,0.00010330692,0.008406495,0.0001387741,0.00006612715,0.0005818709,0.00038644695,0.31906852,0.45285812,0.015143676,0.20298043,0.00016509187],"about_ca_topic_score_codex":0.0033470893,"about_ca_topic_score_gemma":0.0082893595,"teacher_disagreement_score":0.019327525,"about_ca_system_score_codex":0.0007241547,"about_ca_system_score_gemma":0.0012148687,"threshold_uncertainty_score":0.06465697},"labels":[],"label_agreement":null},{"id":"W4402947492","doi":"10.1093/bioinformatics/btae561","title":"RetroCaptioner: beyond attention in end-to-end retrosynthesis transformer via contrastively captioned learnable graph representation","year":2024,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Machine Learning in Materials Science","field":"Materials Science","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Young Scientists Fund; National Natural Science Foundation of China; Central South University; Institute for Catastrophic Loss Reduction","keywords":"Computer science; Encoder; Transformer; Artificial intelligence; Graph; Natural language processing; Theoretical computer science; Engineering","score_opus":0.012636624051815783,"score_gpt":0.2618396766940532,"score_spread":0.24920305264223744,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4402947492","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.021884937,0.0011596153,0.88639677,0.0006359836,0.00030368945,0.00026820443,0.0035796792,0.07988552,0.005885703],"genre_scores_gemma":[0.24018317,0.00088488334,0.72080237,0.001118815,0.00016573622,0.0004895445,0.018959358,0.004401577,0.012994514],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99941504,0.000103756916,0.00002425865,0.00024626398,0.0001539497,0.00005666557],"domain_scores_gemma":[0.99907625,0.00035883806,0.000069162605,0.00028727832,0.00015161707,0.000056770743],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008919376,0.0022787123,0.0010523936,0.0008961188,0.0004332939,0.0013802876,0.0033447328,0.0019157312,0.010935445],"category_scores_gemma":[0.0034761422,0.00061553903,0.0014320709,0.00077798846,0.00080345746,0.0029609671,0.0020820303,0.0031187774,0.0065096333],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007025927,0.00043878073,0.0015939936,0.0010161656,0.00018101209,0.0005849199,0.00019910473,0.21434024,0.05210254,0.019261556,0.05823243,0.6513466],"study_design_scores_gemma":[0.000053335272,0.00014518022,0.00016786411,0.000032674605,0.00003701367,0.00012686767,0.000030033469,0.9290181,0.04198037,0.015354006,0.0130191725,0.000035419544],"about_ca_topic_score_codex":0.0030742823,"about_ca_topic_score_gemma":0.0057634236,"teacher_disagreement_score":0.010935445,"about_ca_system_score_codex":0.0010425926,"about_ca_system_score_gemma":0.0015138326,"threshold_uncertainty_score":0.036582768},"labels":[],"label_agreement":null},{"id":"W4403113340","doi":"10.1093/bioinformatics/btae589","title":"Learning to explain is a good biomedical few-shot learner","year":2024,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Explainable Artificial Intelligence (XAI)","field":"Computer Science","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Natural Science Foundation of China; Fundamental Research Funds for the Central Universities; Institute for Catastrophic Loss Reduction","keywords":"Computer science; Shot (pellet); One shot; Artificial intelligence; Chemistry; Engineering; Mechanical engineering","score_opus":0.035562380217422467,"score_gpt":0.29649448069667717,"score_spread":0.2609321004792547,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4403113340","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.059863266,0.0020945903,0.92693734,0.002486222,0.00016310292,0.00015903801,0.0007380035,0.0041729882,0.0033852917],"genre_scores_gemma":[0.6822525,0.0008337746,0.30100173,0.0020792237,0.00028686557,0.00020690252,0.0045174775,0.00029735116,0.008524195],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9989497,0.00034985406,0.000049571943,0.00044798938,0.00014804516,0.00005491443],"domain_scores_gemma":[0.99661225,0.002513453,0.00012295852,0.00033928253,0.00026219696,0.00014979621],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0021161514,0.0010843612,0.00088355545,0.00060880056,0.0006777631,0.0010290748,0.0016771348,0.0021432787,0.0040258477],"category_scores_gemma":[0.008507232,0.00035843937,0.0010275905,0.00041987936,0.0010465902,0.002289026,0.001665204,0.002419539,0.0012222851],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007751926,0.00073449494,0.011148645,0.0011361243,0.00037368678,0.0010721317,0.0009375421,0.25420913,0.017210102,0.025480106,0.025724743,0.66119814],"study_design_scores_gemma":[0.000063316635,0.00023888481,0.0010308756,0.000083046114,0.000061813,0.00039664604,0.00013571013,0.9409588,0.010317437,0.041568577,0.0051065153,0.00003840971],"about_ca_topic_score_codex":0.001870847,"about_ca_topic_score_gemma":0.0029395504,"teacher_disagreement_score":0.0040258477,"about_ca_system_score_codex":0.0006795997,"about_ca_system_score_gemma":0.0011045355,"threshold_uncertainty_score":0.013467789},"labels":[],"label_agreement":null},{"id":"W4403299283","doi":"10.1093/bioinformatics/btae601","title":"MNBC: a multithreaded Minimizer-based Naïve Bayes Classifier for improved metagenomic sequence classification","year":2024,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Public Health Agency of Canada; University of Manitoba; Agriculture and Agri-Food Canada; Canadian Food Inspection Agency","funders":"Public Health Agency; Public Health Agency of Canada; Agriculture and Agri-Food Canada; Canadian Food Inspection Agency; University of Lethbridge","keywords":"Metagenomics; Naive Bayes classifier; Classifier (UML); Computer science; Bayes' theorem; Artificial intelligence; Bayes classifier; Pattern recognition (psychology); Sequence (biology); Machine learning; Computational biology; Bayesian probability; Biology; Support vector machine; Genetics","score_opus":0.05319947061359019,"score_gpt":0.29396832035820253,"score_spread":0.24076884974461232,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4403299283","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.026881048,0.0014185758,0.9275458,0.00079072616,0.00033640108,0.0005001798,0.0033169235,0.0372513,0.0019591164],"genre_scores_gemma":[0.11086162,0.0002966096,0.8754284,0.0006595201,0.00015796241,0.0006162139,0.008070659,0.001798402,0.002110743],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9968623,0.00088099757,0.00025433514,0.00069263583,0.0011336381,0.00017614038],"domain_scores_gemma":[0.99577963,0.0018499113,0.00026533764,0.0004798517,0.0014207994,0.00020448597],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0056559113,0.0018347254,0.0016477964,0.0018364157,0.0010361859,0.0018247605,0.0031012269,0.0016590087,0.004222439],"category_scores_gemma":[0.016598867,0.0008594077,0.001299443,0.0013893798,0.0005537188,0.0017549862,0.0015535783,0.0029081549,0.002590547],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00121638,0.0005116121,0.016392397,0.0011033877,0.00058433757,0.0002375782,0.00034307514,0.20018566,0.029984089,0.00741464,0.06972952,0.6722973],"study_design_scores_gemma":[0.00009313922,0.00009002748,0.0007517915,0.000047753314,0.000041609175,0.00005220199,0.000029175317,0.97936183,0.007384581,0.0055722524,0.006538125,0.000037482645],"about_ca_topic_score_codex":0.014940811,"about_ca_topic_score_gemma":0.019640185,"teacher_disagreement_score":0.014940811,"about_ca_system_score_codex":0.001557608,"about_ca_system_score_gemma":0.0041985502,"threshold_uncertainty_score":0.029911697},"labels":[],"label_agreement":null},{"id":"W4403597831","doi":"10.1093/bioinformatics/btae634","title":"PONYTA: prioritization of phenotype-related genes from mouse KO events using PU learning on a biological network","year":2024,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Artificial Intelligence in Medicine (Canada)","funders":"Institute for Information and Communications Technology Promotion; Ministry of Science and ICT, South Korea; Seoul National University; National Research Foundation","keywords":"False positive paradox; Cutoff; Phenotype; Transcriptome; Computational biology; Gene; Benchmark (surveying); Biology; Prioritization; Gene regulatory network; Computer science; Genetics; Gene expression; Machine learning; Physics","score_opus":0.015840838371398647,"score_gpt":0.24211639001666194,"score_spread":0.22627555164526328,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4403597831","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.02031918,0.0002257817,0.9404249,0.00024977507,0.00008764377,0.00023188816,0.000977117,0.03601845,0.0014651862],"genre_scores_gemma":[0.26994237,0.00032709094,0.71675885,0.00039705692,0.00008813508,0.00063587114,0.005230209,0.002564503,0.004055875],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9996152,0.0000782313,0.00002018961,0.00012466348,0.00011478603,0.00004699914],"domain_scores_gemma":[0.9989261,0.00049678795,0.00012381146,0.00014992637,0.00019111045,0.000112207985],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013787833,0.0013519034,0.00080986466,0.0012296049,0.00048698505,0.0011778229,0.0019848372,0.00092785474,0.0072534974],"category_scores_gemma":[0.0040466166,0.00047959518,0.0010694563,0.0004623913,0.0006667948,0.0013932276,0.0017776738,0.0017432908,0.0014397186],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0012824181,0.000618978,0.015243003,0.0011788496,0.00037609137,0.0007126982,0.00025122182,0.34285593,0.060636543,0.014088922,0.03720179,0.52555364],"study_design_scores_gemma":[0.000038696526,0.00009942106,0.00076274213,0.00002025915,0.00002623492,0.00012400236,0.000017593113,0.98135513,0.008716556,0.006373372,0.0024506396,0.000015452806],"about_ca_topic_score_codex":0.0026089386,"about_ca_topic_score_gemma":0.0039348914,"teacher_disagreement_score":0.0072534974,"about_ca_system_score_codex":0.0007948315,"about_ca_system_score_gemma":0.0012302839,"threshold_uncertainty_score":0.024265349},"labels":[],"label_agreement":null},{"id":"W4403597865","doi":"10.1093/bioinformatics/btae622","title":"Biologically-informed killer cell immunoglobulin-like receptor gene annotation tool","year":2024,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Immune Cell Function and Interaction","field":"Immunology and Microbiology","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Victoria","funders":"National Cancer Institute; National Institutes of Health","keywords":"Genotyping; Annotation; Computational biology; Computer science; Gene; Identification (biology); Biology; Genetics; Genotype","score_opus":0.013490382800119905,"score_gpt":0.23653785458795284,"score_spread":0.22304747178783293,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4403597865","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.01464983,0.0008445481,0.44376665,0.00051157107,0.00047494142,0.00032756492,0.07226016,0.45927343,0.007891308],"genre_scores_gemma":[0.071999766,0.0005636489,0.63027376,0.00090360694,0.00010689177,0.0010070846,0.24477223,0.04343097,0.006942008],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9989286,0.00020044591,0.00011738641,0.00038163087,0.00027981086,0.00009207028],"domain_scores_gemma":[0.9982998,0.00080690556,0.00020759575,0.00026623582,0.00028534155,0.00013419811],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0025634288,0.0018490939,0.0009335953,0.0017759249,0.0010870135,0.0021547363,0.0029011308,0.0014208028,0.024733964],"category_scores_gemma":[0.005926488,0.0007722986,0.0015891948,0.0014639142,0.0004188492,0.0014912885,0.0024797583,0.0020537064,0.018455587],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.003694394,0.00037687182,0.0119628245,0.006258953,0.0009864098,0.0019525058,0.0016673764,0.020980448,0.10891745,0.015186982,0.5791063,0.24890956],"study_design_scores_gemma":[0.0010100547,0.00061134575,0.009702441,0.00079423055,0.0005213859,0.0027273481,0.000545267,0.20336226,0.13278756,0.034238543,0.61306846,0.00063119177],"about_ca_topic_score_codex":0.0018713651,"about_ca_topic_score_gemma":0.0026435368,"teacher_disagreement_score":0.024733964,"about_ca_system_score_codex":0.00067342544,"about_ca_system_score_gemma":0.0015258612,"threshold_uncertainty_score":0.08274335},"labels":[],"label_agreement":null},{"id":"W4403680155","doi":"10.1093/bioinformatics/btae636","title":"EuDockScore: Euclidean graph neural networks for scoring protein–protein interfaces","year":2024,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Monoclonal and Polyclonal Antibodies Research","field":"Medicine","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canada Research Chairs; University of Toronto; University of New Brunswick","funders":"Canadian Institutes of Health Research; Alliance de recherche numérique du Canada","keywords":"Computer science; Macromolecular docking; Docking (animal); Artificial intelligence; Machine learning; Artificial neural network; Graph; Protein–protein interaction; Computational biology; Protein structure; Theoretical computer science; Biology","score_opus":0.03702823887180946,"score_gpt":0.31701686099494925,"score_spread":0.2799886221231398,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4403680155","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.11142322,0.004110133,0.7762486,0.0022691493,0.0007346517,0.00042445838,0.017333582,0.06897693,0.01847936],"genre_scores_gemma":[0.6031444,0.001082379,0.34441668,0.0013619502,0.00014859672,0.0006517606,0.028818127,0.0034218114,0.01695432],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9992968,0.000183691,0.00003831592,0.00016454479,0.00023423962,0.00008236923],"domain_scores_gemma":[0.998909,0.00048834726,0.00008854146,0.0001714574,0.00024898082,0.000093623545],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012291933,0.0016653832,0.0008093933,0.0009174856,0.0004257618,0.0010562163,0.0026941178,0.00176268,0.008352508],"category_scores_gemma":[0.0063821613,0.00048352493,0.00068671705,0.0009832507,0.000567034,0.0016060623,0.001133643,0.0020861987,0.0027492507],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00067395705,0.00024630033,0.006593407,0.00064876245,0.00030144528,0.00019830898,0.000056165143,0.66310006,0.004588249,0.014455433,0.111194685,0.19794324],"study_design_scores_gemma":[0.00004297314,0.00004357621,0.0003913044,0.00001923385,0.000011362027,0.000026920414,0.000007905873,0.9880625,0.0020260278,0.006533285,0.0028207968,0.000014042064],"about_ca_topic_score_codex":0.014259288,"about_ca_topic_score_gemma":0.020529155,"teacher_disagreement_score":0.014259288,"about_ca_system_score_codex":0.0015342779,"about_ca_system_score_gemma":0.001654402,"threshold_uncertainty_score":0.028352559},"labels":[],"label_agreement":null},{"id":"W4404091271","doi":"10.1093/bioinformatics/btae660","title":"ADTGP: correcting single-cell antibody sequencing data using Gaussian process regression","year":2024,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Single-cell and spatial transcriptomics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Princess Margaret Cancer Centre; University of Toronto","funders":"","keywords":"Interpretability; Computer science; R package; Isotype; Raw data; Regression; Process (computing); Count data; Data mining; Algorithm; Computational biology; Statistics; Mathematics; Artificial intelligence; Antibody; Biology; Immunology; Computational science","score_opus":0.06228579374345745,"score_gpt":0.3170366147317804,"score_spread":0.25475082098832297,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4404091271","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0045653316,0.0007511341,0.85425305,0.00036007343,0.0005995183,0.00016899777,0.02469941,0.11288308,0.0017194296],"genre_scores_gemma":[0.038592793,0.0007360203,0.8324974,0.00077625073,0.00020972834,0.0013657381,0.036444943,0.081566654,0.007810515],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99603266,0.0012555699,0.00021252278,0.0011435376,0.0011581974,0.00019756984],"domain_scores_gemma":[0.99289507,0.0035123986,0.00054178154,0.0014120571,0.0014652101,0.00017347088],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0075151715,0.0031243362,0.0022440602,0.0019880002,0.0011723019,0.0029498208,0.0034647535,0.0015767682,0.041172635],"category_scores_gemma":[0.023591314,0.0016246492,0.002710906,0.0019116758,0.00091623905,0.0019177949,0.002209956,0.0041611474,0.03076059],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000724233,0.00022278614,0.014101853,0.0034320387,0.0019820577,0.0005259456,0.0006616469,0.0730851,0.043299977,0.024410997,0.52024823,0.3173052],"study_design_scores_gemma":[0.00031444195,0.00024739606,0.007567327,0.00036948308,0.00048039123,0.0007667573,0.00014322322,0.48738787,0.07838438,0.05315548,0.3707387,0.00044450312],"about_ca_topic_score_codex":0.005708324,"about_ca_topic_score_gemma":0.009076366,"teacher_disagreement_score":0.041172635,"about_ca_system_score_codex":0.0011257661,"about_ca_system_score_gemma":0.0032655878,"threshold_uncertainty_score":0.13773614},"labels":[],"label_agreement":null},{"id":"W4404185145","doi":"10.1093/bioinformatics/btae666","title":"MR Corge: sensitivity analysis of Mendelian randomization based on the core gene hypothesis for polygenic exposures","year":2024,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; Montreal Heart Institute","funders":"Japan Society for the Promotion of Science; Canadian Institutes of Health Research; Compute Canada; Division of Graduate Education; University of Wisconsin-Madison","keywords":"Mendelian randomization; Sensitivity (control systems); Computational biology; Mendelian inheritance; Biology; Core (optical fiber); Genetics; Gene; Computer science; Statistics; Mathematics; Genetic variants; Genotype","score_opus":0.027509506983968125,"score_gpt":0.2652732472096275,"score_spread":0.2377637402256594,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4404185145","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.017300464,0.0016141351,0.9586342,0.0009066489,0.00037125018,0.0016639413,0.006933208,0.010769336,0.0018068114],"genre_scores_gemma":[0.26559773,0.0007704251,0.7121974,0.0012821214,0.0003005656,0.007246895,0.005335933,0.0054165255,0.0018524029],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.8828539,0.10073463,0.0031280203,0.008143778,0.0040788557,0.0010607985],"domain_scores_gemma":[0.5651153,0.39256307,0.011702962,0.025810484,0.0038914303,0.00091680634],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.09731776,0.00258698,0.004590082,0.0047185523,0.00096975657,0.0033777794,0.003993871,0.002200917,0.03442488],"category_scores_gemma":[0.37620065,0.0015101015,0.008962313,0.0044063884,0.0028040253,0.0022248873,0.0048264246,0.0048562097,0.0023490533],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.009753655,0.00051336887,0.068451166,0.014055097,0.053672448,0.0026833168,0.0020172761,0.21020982,0.005808093,0.13275513,0.10896897,0.39111164],"study_design_scores_gemma":[0.0036680966,0.0016567421,0.025425253,0.0017789179,0.009686224,0.0021925254,0.00035325412,0.607926,0.007002695,0.26660597,0.073068224,0.0006361729],"about_ca_topic_score_codex":0.0030490058,"about_ca_topic_score_gemma":0.002383078,"teacher_disagreement_score":0.09731776,"about_ca_system_score_codex":0.0016182269,"about_ca_system_score_gemma":0.003377399,"threshold_uncertainty_score":0.5146717},"labels":[],"label_agreement":null},{"id":"W4404607640","doi":"10.1093/bioinformatics/btae701","title":"Sparse Neighbor Joining: rapid phylogenetic inference using a sparse distance matrix","year":2024,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"Vetenskapsrådet","keywords":"Distance matrix; Python (programming language); Computer science; Sparse matrix; Distance matrices in phylogeny; Phylogenetic tree; Inference; Computation; Matrix (chemical analysis); Set (abstract data type); Tree (set theory); Algorithm; Theoretical computer science; Data mining; Combinatorics; Artificial intelligence; Mathematics; Biology","score_opus":0.030483514826302148,"score_gpt":0.2807530257646901,"score_spread":0.25026951093838795,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4404607640","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0058582327,0.00021155215,0.9832941,0.00026449922,0.000097051,0.0001243039,0.0010163795,0.008026407,0.0011074602],"genre_scores_gemma":[0.0494625,0.00025802953,0.9438091,0.00014236024,0.00011416299,0.00023090871,0.0037366727,0.0010640288,0.0011822741],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.996655,0.0009824217,0.0001865455,0.00083062326,0.0011958688,0.0001496136],"domain_scores_gemma":[0.9934597,0.0029792078,0.0007337312,0.0014937329,0.0009486915,0.00038484548],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0035162806,0.0011908716,0.0017145453,0.0025191586,0.0018278699,0.0025376666,0.0039968276,0.0016735984,0.008031691],"category_scores_gemma":[0.021185257,0.00133535,0.0012718423,0.0037464385,0.001343845,0.003925392,0.003737602,0.0029365905,0.0070383493],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009315716,0.0004142451,0.007445529,0.0019142345,0.0004807229,0.00055832946,0.0016330751,0.16331144,0.05079577,0.056282565,0.065036245,0.6511962],"study_design_scores_gemma":[0.00017176881,0.0001314613,0.0012665349,0.00010830038,0.00006366481,0.0006494114,0.00020058459,0.8699013,0.012790444,0.08576831,0.028848773,0.00009944004],"about_ca_topic_score_codex":0.0029045937,"about_ca_topic_score_gemma":0.0042133224,"teacher_disagreement_score":0.008031691,"about_ca_system_score_codex":0.0006588698,"about_ca_system_score_gemma":0.0022022265,"threshold_uncertainty_score":0.026868701},"labels":[],"label_agreement":null},{"id":"W4404691004","doi":"10.1093/bioinformatics/btae707","title":"<i>lefser</i>: implementation of metagenomic biomarker discovery tool, <i>LEfSe</i>, in R","year":2024,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":45,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Institute of Population and Public Health","funders":"National Cancer Institute; National Institutes of Health","keywords":"Bioconductor; Python (programming language); Metagenomics; Computer science; Software; Data mining; Source code; JavaScript; Benchmarking; Visualization; Population; Data science; Biology; World Wide Web; Operating system","score_opus":0.01587551541960474,"score_gpt":0.29522894954475254,"score_spread":0.2793534341251478,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4404691004","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.002299847,0.00040216773,0.2528439,0.001130114,0.0004811842,0.00038676924,0.105418704,0.63141906,0.005618281],"genre_scores_gemma":[0.036145516,0.00050863624,0.47700602,0.0026066052,0.00033103194,0.0033567096,0.17253594,0.29635295,0.0111565925],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99738973,0.0005228624,0.00026800515,0.000853169,0.0007079424,0.00025837964],"domain_scores_gemma":[0.9941309,0.0024700707,0.0007776475,0.001132632,0.0010297528,0.00045896476],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00604694,0.0034157967,0.0021847226,0.0030974506,0.0012360681,0.003903711,0.0052278917,0.0012972035,0.09271686],"category_scores_gemma":[0.019824823,0.001703132,0.0028601137,0.0022100064,0.0011746661,0.0030354937,0.0033402059,0.0039242976,0.103551656],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00044830487,0.000060966984,0.0039668437,0.0016035103,0.00044763368,0.00032400744,0.00029138805,0.0038618902,0.009196633,0.0057111736,0.911528,0.06255966],"study_design_scores_gemma":[0.0006542514,0.00024618546,0.0077561648,0.00056296715,0.00030080858,0.0013546867,0.00014625133,0.07009154,0.052788936,0.05541155,0.81010437,0.0005822624],"about_ca_topic_score_codex":0.003418036,"about_ca_topic_score_gemma":0.004702939,"teacher_disagreement_score":0.09271686,"about_ca_system_score_codex":0.001135782,"about_ca_system_score_gemma":0.003392357,"threshold_uncertainty_score":0.3101688},"labels":[],"label_agreement":null},{"id":"W4405218992","doi":"10.1093/bioinformatics/btae715","title":"VSS-Hi-C: variance-stabilized signals for chromatin contacts","year":2024,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"Natural Sciences and Engineering Research Council of Canada; Compute Canada","keywords":"Variance (accounting); Computer science; Chromatin; Biology; Genetics; DNA; Business","score_opus":0.011047146874385268,"score_gpt":0.24874267685226964,"score_spread":0.23769552997788437,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4405218992","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.03100534,0.0007844348,0.86816233,0.0005402438,0.00039166232,0.0002063993,0.01562046,0.08071783,0.0025714051],"genre_scores_gemma":[0.18829964,0.00053101825,0.7470146,0.0006578243,0.00027843527,0.0012722525,0.03153032,0.02669642,0.0037194728],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.997721,0.00058007374,0.000095334566,0.00067340536,0.00073692494,0.00019316032],"domain_scores_gemma":[0.994351,0.0030781324,0.00044699106,0.00096023135,0.0009296028,0.00023398231],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0037782232,0.0015055377,0.0011874128,0.0014961874,0.00094146666,0.0018342277,0.002749716,0.0013851007,0.015749132],"category_scores_gemma":[0.020497499,0.000824846,0.0013222862,0.0018373276,0.0010712434,0.0013005753,0.0022560111,0.0028782878,0.007484533],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0030759168,0.00037891022,0.025864167,0.004135248,0.0012084274,0.0006451024,0.0015975067,0.07334678,0.19683167,0.04018619,0.21085659,0.44187352],"study_design_scores_gemma":[0.0003785864,0.00036846887,0.017874857,0.00031977397,0.00025243402,0.00060529675,0.00023891723,0.65975636,0.20623137,0.034839448,0.07864811,0.00048628956],"about_ca_topic_score_codex":0.0031574697,"about_ca_topic_score_gemma":0.004462254,"teacher_disagreement_score":0.015749132,"about_ca_system_score_codex":0.00088076864,"about_ca_system_score_gemma":0.001513736,"threshold_uncertainty_score":0.052686095},"labels":[],"label_agreement":null},{"id":"W4405221092","doi":"10.1093/bioinformatics/btae719","title":"PheWAS analysis on large-scale biobank data with PheTK","year":2024,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"National Human Genome Research Institute; National Institutes of Health","keywords":"Biobank; Computer science; Scale (ratio); Data science; Bioinformatics; Biology; Geography; Cartography","score_opus":0.01748673113263469,"score_gpt":0.2820495558834093,"score_spread":0.2645628247507746,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4405221092","genre_codex":"dataset","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.041892365,0.0011242152,0.24224736,0.0027923814,0.0010752738,0.0013190559,0.4217338,0.2700148,0.017800715],"genre_scores_gemma":[0.13135368,0.00075398805,0.22412275,0.0023040096,0.00030364774,0.005126603,0.5680502,0.055421423,0.012563598],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9944699,0.0013101031,0.000657972,0.0018737479,0.001220357,0.0004680038],"domain_scores_gemma":[0.9832622,0.008578945,0.0012144647,0.0039975625,0.0023838244,0.00056303554],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008498127,0.0016455937,0.0014119813,0.0036217235,0.0015938794,0.0048675486,0.0022464904,0.000987289,0.04532903],"category_scores_gemma":[0.041614976,0.0011320534,0.0027840277,0.004736164,0.0011016116,0.003335664,0.0053193225,0.0024150813,0.028907293],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.003983064,0.00035783276,0.051058147,0.004702367,0.0028387895,0.001414976,0.0014974971,0.020586433,0.009768601,0.012304584,0.74482024,0.14666744],"study_design_scores_gemma":[0.0016879091,0.0005870672,0.075481534,0.0011047707,0.0012091692,0.0019200151,0.0011560486,0.14597824,0.02403506,0.08993114,0.6562217,0.00068728253],"about_ca_topic_score_codex":0.0058024195,"about_ca_topic_score_gemma":0.008301069,"teacher_disagreement_score":0.04532903,"about_ca_system_score_codex":0.001003391,"about_ca_system_score_gemma":0.005361785,"threshold_uncertainty_score":0.15164071},"labels":[],"label_agreement":null},{"id":"W4405231346","doi":"10.1093/bioinformatics/btae729","title":"A BLAST from the past: revisiting blastp’s <i>E</i>-value","year":2024,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":12,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Value (mathematics); Computer science; History; Psychology; Machine learning","score_opus":0.009638727986995934,"score_gpt":0.22526681105481872,"score_spread":0.21562808306782277,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4405231346","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.015151914,0.017461166,0.88636434,0.049546737,0.008591396,0.00028580194,0.003410603,0.0084176045,0.010770424],"genre_scores_gemma":[0.06641064,0.0085356375,0.90057087,0.00874354,0.0028353934,0.0003979625,0.0033422261,0.004592085,0.004571581],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9827602,0.0075985417,0.001534919,0.0029205405,0.004680422,0.00050537253],"domain_scores_gemma":[0.9410162,0.037428766,0.0035768037,0.0073358775,0.009288421,0.0013539605],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.030793078,0.0023212149,0.0028418961,0.0039779167,0.002925192,0.008175551,0.0064155613,0.005357103,0.0074852766],"category_scores_gemma":[0.14611223,0.0016474609,0.0016220903,0.006954615,0.003809344,0.011147035,0.00516418,0.010143077,0.008417513],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001115335,0.0001538532,0.00938281,0.0025463465,0.00025650553,0.0015090352,0.0013691173,0.025547275,0.009826904,0.15044768,0.22377634,0.5740687],"study_design_scores_gemma":[0.000125545,0.0004365637,0.002920277,0.0022523042,0.00020327751,0.0035665804,0.0009181552,0.17348929,0.010575047,0.36105663,0.4441237,0.00033277614],"about_ca_topic_score_codex":0.004810682,"about_ca_topic_score_gemma":0.004436698,"teacher_disagreement_score":0.030793078,"about_ca_system_score_codex":0.002994222,"about_ca_system_score_gemma":0.003730779,"threshold_uncertainty_score":0.16285133},"labels":[],"label_agreement":null},{"id":"W4405356247","doi":"10.1093/bioinformatics/btae738","title":"DiffPaSS—high-performance differentiable pairing of protein sequences using soft scores","year":2024,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Institute of Infection and Immunity","funders":"H2020 European Research Council; European Commission; Cancer Research UK","keywords":"Python (programming language); Hyperparameter; Computer science; Differentiable function; Pairing; Benchmark (surveying); Artificial intelligence; Similarity (geometry); Machine learning; Theoretical computer science; Graph; Data mining; Mathematics","score_opus":0.013342887466565215,"score_gpt":0.2266625405328697,"score_spread":0.21331965306630446,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4405356247","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.08071121,0.00094058027,0.7988059,0.00052173686,0.0002605414,0.0003694891,0.0048720962,0.10590973,0.0076087574],"genre_scores_gemma":[0.3107504,0.00038783732,0.6562675,0.00046541655,0.00012576341,0.0005336246,0.01870512,0.00605984,0.006704552],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9981331,0.00043971324,0.0001312106,0.0004615258,0.00067038543,0.00016408053],"domain_scores_gemma":[0.9982469,0.00073280337,0.00016515444,0.0004035561,0.0002462996,0.00020519076],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0024834494,0.0024802762,0.0023025572,0.0015710532,0.0009085165,0.002474262,0.003200458,0.0018557063,0.013282372],"category_scores_gemma":[0.008303169,0.0010347826,0.0016366536,0.0018851903,0.00103638,0.00378663,0.0035153115,0.0029051946,0.008424259],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0017446141,0.00074163626,0.010446442,0.0011929956,0.0005295951,0.00053226145,0.000265408,0.3983347,0.026112335,0.03354518,0.09385669,0.43269813],"study_design_scores_gemma":[0.00009575067,0.00012773887,0.00059128605,0.000015159438,0.0000117954,0.00008799909,0.000024989062,0.9761202,0.00546379,0.0128575405,0.004571177,0.000032710082],"about_ca_topic_score_codex":0.004534,"about_ca_topic_score_gemma":0.0065656737,"teacher_disagreement_score":0.013282372,"about_ca_system_score_codex":0.0012979741,"about_ca_system_score_gemma":0.0024865544,"threshold_uncertainty_score":0.04443389},"labels":[],"label_agreement":null},{"id":"W4405596815","doi":"10.1093/bioinformatics/btae748","title":"CParty: hierarchically constrained partition function of RNA pseudoknots","year":2024,"lang":"en","type":"article","venue":"Bioinformatics","topic":"RNA and protein synthesis mechanisms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Victoria; University of Alberta","funders":"Natural Sciences and Engineering Research Council of Canada; Agence Nationale de la Recherche","keywords":"Pseudoknot; Partition function (quantum field theory); Algorithm; Partition (number theory); Partition problem; Computer science; Dynamic programming; Function (biology); Time complexity; Nucleic acid secondary structure; Mathematics; RNA; Combinatorics; Physics","score_opus":0.011570003130580743,"score_gpt":0.23340572108363922,"score_spread":0.22183571795305848,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4405596815","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.01929739,0.000121833225,0.9770596,0.00010895509,0.00003113122,0.00007730812,0.00018520882,0.0018171822,0.0013013508],"genre_scores_gemma":[0.39099637,0.00021677023,0.6017636,0.00025002917,0.00006781804,0.0007662236,0.0013307633,0.0016379309,0.0029705013],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9994821,0.00013753603,0.000020995121,0.000103494676,0.00018888888,0.00006703897],"domain_scores_gemma":[0.9987024,0.00071295764,0.00011360507,0.00017037951,0.00020299225,0.00009763947],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008884392,0.0011012324,0.0008643059,0.0007851252,0.0009268046,0.0010578668,0.0024123208,0.0015521478,0.0035294606],"category_scores_gemma":[0.004214149,0.0004944755,0.00089020253,0.0006808028,0.0012158384,0.0015710674,0.0015725093,0.001542393,0.00121963],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00010694106,0.000045096855,0.00094136636,0.000094184266,0.000025179346,0.000089518486,0.000088432964,0.9163247,0.0032781796,0.037588317,0.0028516923,0.038566407],"study_design_scores_gemma":[0.0000060932066,0.0000074022932,0.000046413483,0.0000036408287,0.000001207602,0.000010991814,0.000003970336,0.98941445,0.000861331,0.009190914,0.00044881363,0.0000047851445],"about_ca_topic_score_codex":0.0061195786,"about_ca_topic_score_gemma":0.003517684,"teacher_disagreement_score":0.0061195786,"about_ca_system_score_codex":0.0013642535,"about_ca_system_score_gemma":0.0020040055,"threshold_uncertainty_score":0.012167931},"labels":[],"label_agreement":null},{"id":"W4405625946","doi":"10.1093/bioinformatics/btag415","title":"Fitness translocation: improving variant effect prediction with biologically-grounded data augmentation","year":2024,"lang":"en","type":"preprint","venue":"Bioinformatics","topic":"RNA and protein synthesis mechanisms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Bottleneck; Fitness function; Computer science; Fitness landscape; Sequence (biology); Function (biology); Machine learning; Selection (genetic algorithm); Computational biology; Artificial intelligence; Biology; Genetics; Medicine; Genetic algorithm; Population","score_opus":0.02251479689590924,"score_gpt":0.2593764694314939,"score_spread":0.23686167253558466,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4405625946","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.24941564,0.0020097983,0.7228919,0.0015132953,0.0003318328,0.00019187442,0.0051180157,0.015404265,0.0031233036],"genre_scores_gemma":[0.6675661,0.00048067042,0.3147086,0.0008253235,0.00012760329,0.00030572887,0.012984667,0.0011398604,0.0018615759],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9991091,0.00029097596,0.000049555743,0.00034248605,0.0001504321,0.000057452184],"domain_scores_gemma":[0.99733293,0.0017189664,0.00014125255,0.00048151755,0.00023218775,0.00009317109],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0021702494,0.0018476687,0.000950278,0.00084517256,0.00047040184,0.0010966861,0.0015425663,0.0013699874,0.0020338781],"category_scores_gemma":[0.0077389902,0.00047164332,0.0012999752,0.00070165936,0.0009518524,0.0018517227,0.0017470389,0.0022119493,0.0010267093],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00092984,0.0005294028,0.01722771,0.00054511917,0.00042438015,0.00045137908,0.00023422387,0.6954806,0.038087677,0.007816045,0.013828749,0.22444488],"study_design_scores_gemma":[0.000043426142,0.000111971276,0.00066037895,0.000021637135,0.000030127665,0.00007183883,0.000026159223,0.9819619,0.006472619,0.008965655,0.0016129309,0.000021275047],"about_ca_topic_score_codex":0.002406211,"about_ca_topic_score_gemma":0.0038079133,"teacher_disagreement_score":0.002406211,"about_ca_system_score_codex":0.00054905954,"about_ca_system_score_gemma":0.000859574,"threshold_uncertainty_score":0.01147753},"labels":[],"label_agreement":null},{"id":"W4406170484","doi":"10.1093/bioinformatics/btaf009","title":"BetaAlign: a deep learning approach for multiple sequence alignment","year":2024,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Tel Aviv University; Azrieli Foundation; Israel Science Foundation","keywords":"Computer science; Artificial intelligence; Inference; Subspace topology; Multiple sequence alignment; Chunking (psychology); Sequence (biology); Sequence alignment; Smith–Waterman algorithm; Machine learning; Deep learning; Set (abstract data type); Phylogenomics; Phylogenetic tree; Biology","score_opus":0.025010616289056334,"score_gpt":0.25447387497717366,"score_spread":0.22946325868811734,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4406170484","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0058912886,0.00050710095,0.9816737,0.00034696868,0.00009050692,0.000074295756,0.0004086971,0.009477875,0.0015295448],"genre_scores_gemma":[0.106410414,0.00063088175,0.88346857,0.0006491877,0.00012554917,0.0003123249,0.002975129,0.0012468612,0.004181093],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9989221,0.00026731592,0.000062544386,0.00033872674,0.0003276748,0.000081652266],"domain_scores_gemma":[0.99881136,0.00046284165,0.00012471073,0.00024663497,0.0002631813,0.00009132938],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016479266,0.001528213,0.0009932934,0.0012067896,0.0007412448,0.0014632869,0.0036526204,0.0016360357,0.004833415],"category_scores_gemma":[0.004479428,0.0007969995,0.0009965152,0.0016364479,0.0012033221,0.00310165,0.0025209421,0.0047368635,0.0025716373],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0004044347,0.00029732843,0.002557852,0.00050330575,0.00022104908,0.00023115761,0.00022999392,0.24213798,0.020706562,0.0513835,0.031066036,0.6502608],"study_design_scores_gemma":[0.000021597838,0.000058245972,0.00016009418,0.000024015764,0.000014504653,0.00006685587,0.00002134526,0.95582587,0.0070369034,0.030302504,0.00645511,0.000012896327],"about_ca_topic_score_codex":0.0030121356,"about_ca_topic_score_gemma":0.004740356,"teacher_disagreement_score":0.004833415,"about_ca_system_score_codex":0.0012300799,"about_ca_system_score_gemma":0.0016935944,"threshold_uncertainty_score":0.016169429},"labels":[],"label_agreement":null},{"id":"W4406213027","doi":"10.1093/bioinformatics/btaf010","title":"FlowPacker: protein side-chain packing with torsional flow matching","year":2025,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Protein Structure and Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"Canadian Institutes of Health Research; Alliance de recherche numérique du Canada","keywords":"Computer science; Side chain; Matching (statistics); Graph; Algorithm; Source code; Folding (DSP implementation); Code (set theory); Chain (unit); Sequence (biology); Flow (mathematics); Set (abstract data type); Theoretical computer science; Programming language; Chemistry; Mathematics; Engineering; Biochemistry","score_opus":0.003672734255921913,"score_gpt":0.21323902876353262,"score_spread":0.20956629450761072,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4406213027","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.039784826,0.0014746399,0.7739261,0.00073129573,0.00041772577,0.00052454247,0.006859014,0.1706858,0.0055961413],"genre_scores_gemma":[0.23382133,0.0010646593,0.7147487,0.00087649527,0.00028994176,0.00085409085,0.01988913,0.01890643,0.009549168],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99959975,0.00006206749,0.000018637285,0.00013018152,0.00013977404,0.000049705588],"domain_scores_gemma":[0.9991978,0.00036226798,0.00008585589,0.00014658998,0.00012361351,0.00008381012],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013252818,0.0018625658,0.0012655841,0.0010690422,0.00078270567,0.0013340166,0.0038084325,0.0017231888,0.015239509],"category_scores_gemma":[0.004947237,0.000953952,0.0012987253,0.0009728924,0.00089815794,0.0026425477,0.0017237862,0.0025526602,0.005330655],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011104518,0.00061932596,0.0068195835,0.0014962049,0.000429589,0.0004489038,0.00037570574,0.36605245,0.028442778,0.03051055,0.17403297,0.38966146],"study_design_scores_gemma":[0.0001046647,0.000054354612,0.00033158806,0.000023352872,0.00001591621,0.00007296997,0.000010736692,0.973678,0.007076942,0.01040766,0.00819712,0.000026700762],"about_ca_topic_score_codex":0.009463669,"about_ca_topic_score_gemma":0.0071777552,"teacher_disagreement_score":0.015239509,"about_ca_system_score_codex":0.0009678698,"about_ca_system_score_gemma":0.0022724418,"threshold_uncertainty_score":0.050981224},"labels":[],"label_agreement":null},{"id":"W4406292104","doi":"10.1093/bioinformatics/btaf014","title":"PhyloMix: enhancing microbiome-trait association prediction through phylogeny-mixing augmentation","year":2025,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gut microbiota and health","field":"Biochemistry, Genetics and Molecular Biology","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Microbiome; Computer science; Trait; Sample (material); Set (abstract data type); Machine learning; Phylogenetic tree; Raw data; Artificial intelligence; Phylogenetics; Data mining; Biology; Bioinformatics","score_opus":0.006999816828427663,"score_gpt":0.25921222452973497,"score_spread":0.2522124077013073,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4406292104","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.11936443,0.0017630018,0.855571,0.0010606549,0.00019053779,0.00017321137,0.0025810353,0.017564733,0.001731434],"genre_scores_gemma":[0.42053163,0.0005306105,0.56506234,0.001120433,0.00018693946,0.0004703166,0.008542538,0.0008465089,0.002708707],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9988637,0.00046326863,0.00004455366,0.0003916051,0.00016508636,0.00007183951],"domain_scores_gemma":[0.99662435,0.0022323208,0.00024990077,0.00048838824,0.00027173903,0.00013337271],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0031193262,0.0017989702,0.0011690698,0.0010901318,0.00067298394,0.0010450532,0.0019381257,0.0013571455,0.00254217],"category_scores_gemma":[0.008586866,0.000694462,0.0017031489,0.0009996106,0.0010197943,0.0022778346,0.0025419341,0.0027989193,0.0011194646],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0016243762,0.0006575346,0.0337336,0.0006025501,0.000562369,0.00038237206,0.0005135005,0.4718584,0.035711747,0.0059795203,0.014785634,0.43358836],"study_design_scores_gemma":[0.000041578678,0.00009272244,0.0010746963,0.000017914961,0.000034551525,0.000060641494,0.000024006178,0.9873455,0.004928168,0.004350599,0.0020080344,0.000021583019],"about_ca_topic_score_codex":0.0030068692,"about_ca_topic_score_gemma":0.0044457177,"teacher_disagreement_score":0.0031193262,"about_ca_system_score_codex":0.00059330766,"about_ca_system_score_gemma":0.0011590283,"threshold_uncertainty_score":0.016496718},"labels":[],"label_agreement":null},{"id":"W4406887020","doi":"10.1093/bioinformatics/btaf042","title":"Multidimensional scaling improves distance-based clustering for microbiome data","year":2025,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gut microbiota and health","field":"Biochemistry, Genetics and Molecular Biology","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"National Institute of General Medical Sciences; School of Medicine and Public Health, University of Wisconsin-Madison","keywords":"Cluster analysis; Computer science; Multidimensional scaling; Medoid; Microbiome; Data mining; Clustering high-dimensional data; Dimensionality reduction; Artificial intelligence; Pattern recognition (psychology); Machine learning; Bioinformatics; Biology","score_opus":0.021011993475996037,"score_gpt":0.29725787483098987,"score_spread":0.27624588135499384,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4406887020","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.057698168,0.0018514976,0.93254036,0.0014297599,0.0003484012,0.000327306,0.0008117738,0.002781393,0.0022114606],"genre_scores_gemma":[0.20684202,0.00058808073,0.7891031,0.0002367906,0.0001783752,0.00033158186,0.001412644,0.0003604144,0.0009469868],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9936771,0.0034337412,0.0004078569,0.0008999723,0.0014345379,0.00014689234],"domain_scores_gemma":[0.9853436,0.0087676635,0.00088198134,0.0024442167,0.0022549734,0.0003075719],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0081276065,0.0016961568,0.0016844111,0.0022585487,0.0016347193,0.0015905298,0.0015567992,0.0014777565,0.0021334814],"category_scores_gemma":[0.034579292,0.0005533672,0.0020640779,0.0029213352,0.0011142747,0.001956679,0.00326477,0.0019950941,0.0019192831],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006878778,0.00031158136,0.013289391,0.0011240391,0.00065113296,0.00029988858,0.0011509073,0.5429109,0.015978727,0.017085228,0.0147102885,0.39180002],"study_design_scores_gemma":[0.00005859564,0.00012768121,0.0021333564,0.000063308005,0.000048662067,0.00013503635,0.00011928354,0.9736319,0.00554367,0.013483124,0.0045985775,0.00005679815],"about_ca_topic_score_codex":0.0060221143,"about_ca_topic_score_gemma":0.0051251994,"teacher_disagreement_score":0.0081276065,"about_ca_system_score_codex":0.0010217407,"about_ca_system_score_gemma":0.0017394818,"threshold_uncertainty_score":0.042983353},"labels":[],"label_agreement":null},{"id":"W4407388488","doi":"10.1093/bioinformatics/btaf060","title":"ImmunoTar—integrative prioritization of cell surface targets for cancer immunotherapy","year":2025,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Cancer Immunotherapy and Biomarkers","field":"Medicine","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canada's Michael Smith Genome Sciences Centre; University of British Columbia; Spinal Cord Injury BC","funders":"National Cancer Institute; Cancer Research UK; National Institutes of Health; Michael Smith Health Research BC; Mark Foundation For Cancer Research","keywords":"Identification (biology); Chimeric antigen receptor; Prioritization; Immunotherapy; Computational biology; Cancer immunotherapy; Cancer; Computer science; Proteomics; Bioinformatics; Biology; Gene","score_opus":0.010638015712514907,"score_gpt":0.29056213018331295,"score_spread":0.27992411447079807,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4407388488","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.29365945,0.0028423595,0.58007425,0.0027139708,0.0004160854,0.00088612386,0.028843133,0.074961334,0.01560333],"genre_scores_gemma":[0.48542583,0.0012191044,0.46177074,0.0011916398,0.0001490605,0.0010424759,0.043896932,0.0028223242,0.00248186],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.999468,0.0001583604,0.000035876128,0.0001571847,0.00013083444,0.000049809547],"domain_scores_gemma":[0.99887997,0.0007053846,0.00008879985,0.0000909407,0.00014793561,0.00008698246],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016316606,0.0012283379,0.0013287334,0.0013983573,0.0005593221,0.0016650074,0.0014632862,0.000659664,0.00527923],"category_scores_gemma":[0.0033505764,0.0003889385,0.0015441725,0.0012361318,0.00040482768,0.0011127723,0.0010202782,0.0009438872,0.0014456592],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.002756957,0.001002493,0.04243423,0.0024494517,0.0011660148,0.0007186333,0.000375639,0.5560206,0.03697712,0.019461673,0.07580838,0.26082876],"study_design_scores_gemma":[0.00021607443,0.00022596623,0.0016446387,0.000039709226,0.00013524546,0.00012539022,0.00006266436,0.9690553,0.009397278,0.009557073,0.009510884,0.000029702092],"about_ca_topic_score_codex":0.0030632012,"about_ca_topic_score_gemma":0.0054073846,"teacher_disagreement_score":0.00527923,"about_ca_system_score_codex":0.0008881041,"about_ca_system_score_gemma":0.00226814,"threshold_uncertainty_score":0.017660856},"labels":[],"label_agreement":null},{"id":"W4407578746","doi":"10.1093/bioinformatics/btaf071","title":"<i>PNL</i>: a software to build polygenic risk scores using a super learner approach based on PairNet, a Convolutional Neural Network","year":2025,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Machine Learning in Healthcare","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université Laval","funders":"Academia Sinica; Institute of Biomedical Sciences, Academia Sinica; Natural Sciences and Engineering Research Council of Canada; National Science and Technology Council","keywords":"Convolutional neural network; Computer science; Polygenic risk score; Software; Artificial intelligence; Machine learning; Artificial neural network; Programming language; Biology; Single-nucleotide polymorphism; Genetics","score_opus":0.019815162476490098,"score_gpt":0.2710087187799389,"score_spread":0.25119355630344875,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4407578746","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0059104282,0.00012785835,0.62151176,0.00041451876,0.00013160965,0.00019059362,0.0071217055,0.36014643,0.0044452096],"genre_scores_gemma":[0.11001914,0.00041786526,0.80150074,0.00097403914,0.00012461698,0.0011740284,0.030402396,0.043139946,0.01224728],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99957615,0.000082605795,0.000040350445,0.00013967387,0.00013224538,0.000028960245],"domain_scores_gemma":[0.99856997,0.00081063196,0.00016670176,0.00019026683,0.0001895838,0.00007280042],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0018551279,0.0014280819,0.0005267958,0.001425167,0.00039919713,0.0010626394,0.0026200626,0.0009947261,0.033776004],"category_scores_gemma":[0.0064243423,0.00091451395,0.0012428843,0.0006919282,0.0005483248,0.0018281466,0.0018395196,0.002220258,0.011405036],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00073680846,0.00028215986,0.015730523,0.0011667695,0.00087657943,0.0012560474,0.0005236272,0.1085716,0.0147280805,0.025445746,0.40291706,0.427765],"study_design_scores_gemma":[0.00016897818,0.00009081441,0.0030161694,0.000094851835,0.000074474774,0.0003754059,0.000030356883,0.90041304,0.015328448,0.020532671,0.059756216,0.00011849053],"about_ca_topic_score_codex":0.005245471,"about_ca_topic_score_gemma":0.006852927,"teacher_disagreement_score":0.033776004,"about_ca_system_score_codex":0.0007463618,"about_ca_system_score_gemma":0.0012165995,"threshold_uncertainty_score":0.11299205},"labels":[],"label_agreement":null},{"id":"W4407613286","doi":"10.1093/bioinformatics/btag199","title":"Refining sequence-to-expression modelling with chromatin accessibility","year":2025,"lang":"en","type":"preprint","venue":"Bioinformatics","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"Natural Sciences and Engineering Research Council of Canada; Alliance de recherche numérique du Canada","keywords":"Refining (metallurgy); Sequence (biology); Chromatin; Expression (computer science); Computational biology; Computer science; Process engineering; Chemistry; Engineering; Biology; Programming language; DNA; Biochemistry","score_opus":0.025698479143283957,"score_gpt":0.2758157139084285,"score_spread":0.2501172347651446,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4407613286","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.18370621,0.00041892796,0.81071377,0.00033628647,0.00006647534,0.000049706647,0.0009101057,0.0021506494,0.0016478848],"genre_scores_gemma":[0.8955003,0.00042198072,0.09733647,0.00020872851,0.000073503514,0.00018320193,0.0019078795,0.00040430602,0.0039635957],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99966586,0.00010031906,0.00001252704,0.00014764298,0.00004551861,0.000028173157],"domain_scores_gemma":[0.9988153,0.000890951,0.00006109863,0.00008759147,0.00009814037,0.000047024765],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00078631006,0.00080106885,0.0006529858,0.00040355982,0.00018989913,0.00082169706,0.000832175,0.0007917228,0.0018525196],"category_scores_gemma":[0.0033041618,0.00043772085,0.001082258,0.00044709462,0.0006529852,0.0008140578,0.0007129493,0.0015291615,0.00060139183],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000994435,0.000039133734,0.0023060914,0.00006839001,0.000038243143,0.00006263622,0.000066812325,0.9688365,0.012920799,0.0024717615,0.00029553205,0.012794593],"study_design_scores_gemma":[0.000004461497,0.0000134959855,0.00034422363,0.0000035443718,0.000006266396,0.0000116662995,0.000003641021,0.993761,0.001845003,0.0036393462,0.00036297433,0.0000044109033],"about_ca_topic_score_codex":0.0042893076,"about_ca_topic_score_gemma":0.005315202,"teacher_disagreement_score":0.0042893076,"about_ca_system_score_codex":0.00059559324,"about_ca_system_score_gemma":0.00070640596,"threshold_uncertainty_score":0.00852865},"labels":[],"label_agreement":null},{"id":"W4408212560","doi":"10.1093/bioinformatics/btaf094","title":"ENACT: End-to-End Analysis of Visium High Definition (HD) Data","year":2025,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Single-cell and spatial transcriptomics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Sanofi (Canada)","funders":"","keywords":"End-to-end principle; Computer science; End user; Operating system; Computer network","score_opus":0.02679882034336073,"score_gpt":0.261496189409761,"score_spread":0.23469736906640026,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4408212560","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.01454686,0.0007092388,0.6054978,0.0006749795,0.00034809974,0.00039842597,0.09071966,0.28007677,0.007028228],"genre_scores_gemma":[0.05127712,0.00057106663,0.676193,0.0009852421,0.000087389104,0.0013003225,0.23301072,0.027421022,0.009154082],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99859816,0.00021925359,0.00009160941,0.0005380309,0.00046124394,0.000091597],"domain_scores_gemma":[0.99803895,0.00069771847,0.0001507511,0.000504492,0.00047638157,0.00013176147],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002720817,0.00180616,0.0010379268,0.0013847663,0.001160936,0.00297896,0.0019608522,0.0010475378,0.016088655],"category_scores_gemma":[0.006030429,0.0010312987,0.0014686743,0.0017010059,0.00088724465,0.0016470288,0.0023077049,0.0022409577,0.021380879],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0027351528,0.00031561407,0.010208057,0.0023876843,0.0007735857,0.0008754517,0.0010474678,0.027151633,0.1979736,0.017390093,0.55019414,0.18894759],"study_design_scores_gemma":[0.0003657114,0.00029498723,0.014828096,0.0002124151,0.00016408549,0.0010201397,0.00049224123,0.29449305,0.22678788,0.039316483,0.4217279,0.00029716917],"about_ca_topic_score_codex":0.0038617891,"about_ca_topic_score_gemma":0.0074888915,"teacher_disagreement_score":0.016088655,"about_ca_system_score_codex":0.0009259643,"about_ca_system_score_gemma":0.0024080914,"threshold_uncertainty_score":0.05382192},"labels":[],"label_agreement":null},{"id":"W4408884143","doi":"10.1093/bioinformatics/btaf129","title":"AI-augmented physics-based docking for antibody-antigen complex prediction","year":2025,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Monoclonal and Polyclonal Antibodies Research","field":"Medicine","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; Ste. Anne's Hospital; National Research Council Canada","funders":"","keywords":"Docking (animal); Computer science; Computational biology; Macromolecular docking; Epitope; Artificial intelligence; Antigen; Biology; Protein structure","score_opus":0.0501294460679583,"score_gpt":0.38286061444346187,"score_spread":0.33273116837550354,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4408884143","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.28410006,0.0014712731,0.6874452,0.0012828641,0.00022578752,0.00014827488,0.0008006386,0.0066283797,0.017897502],"genre_scores_gemma":[0.8622877,0.0005220418,0.13269675,0.00033495156,0.00005669951,0.00013332565,0.0011715479,0.0003447417,0.0024522727],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9996412,0.000109484194,0.000013244727,0.00004987327,0.00014415367,0.00004208475],"domain_scores_gemma":[0.999302,0.00034005483,0.000059185673,0.00009037846,0.00012984125,0.00007847238],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007414306,0.000997351,0.0009938319,0.00063210103,0.00057395163,0.00094365294,0.0012861964,0.0009908358,0.0035864557],"category_scores_gemma":[0.0018969815,0.00036847196,0.0008721664,0.0006781006,0.0005665689,0.0009014897,0.0013259901,0.001505859,0.0007961713],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000075206415,0.000047067682,0.0007987169,0.000052617776,0.00004905171,0.00004917241,0.000023163184,0.98469985,0.001968318,0.0021411935,0.00090249896,0.009193146],"study_design_scores_gemma":[0.000007671811,0.000012003003,0.00010401509,0.0000026811813,0.0000036227839,0.000008887862,0.0000034431844,0.997538,0.0006338638,0.0014422559,0.00023819282,0.000005383476],"about_ca_topic_score_codex":0.008545191,"about_ca_topic_score_gemma":0.0058290325,"teacher_disagreement_score":0.008545191,"about_ca_system_score_codex":0.00084891956,"about_ca_system_score_gemma":0.0013259628,"threshold_uncertainty_score":0.0169909},"labels":[],"label_agreement":null},{"id":"W4408956502","doi":"10.1093/bioinformatics/btaf137","title":"scMUSCL: multi-source transfer learning for clustering scRNA-seq data","year":2025,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Single-cell and spatial transcriptomics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia; Simon Fraser University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Transfer of learning; Cluster analysis; Computer science; Transfer (computing); Data source; Artificial intelligence; Data mining; Parallel computing","score_opus":0.04198762505065439,"score_gpt":0.27984129734352414,"score_spread":0.23785367229286974,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4408956502","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.017430559,0.0004740398,0.9053926,0.0005325236,0.00022152846,0.00024669536,0.0042176032,0.06961386,0.0018706266],"genre_scores_gemma":[0.19620784,0.00048667667,0.7599257,0.0012410716,0.0001468486,0.0012416358,0.027952626,0.0059516896,0.0068459474],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9989575,0.00019142729,0.000052312647,0.0003914367,0.0003154628,0.00009184721],"domain_scores_gemma":[0.9982999,0.0005419473,0.00013181113,0.00048179622,0.00041718062,0.00012730209],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0021676805,0.001996978,0.0009766686,0.0017917029,0.0010792891,0.001219367,0.0041909055,0.0018836367,0.006513853],"category_scores_gemma":[0.006642856,0.00069555885,0.0014590223,0.0018736031,0.0011433003,0.0018196632,0.0033514954,0.0025885974,0.004670861],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007704751,0.00040635213,0.005978201,0.00097469107,0.00048969843,0.00057224964,0.0004363484,0.33231938,0.05270707,0.008248828,0.11123101,0.48586568],"study_design_scores_gemma":[0.00005197502,0.000083402076,0.0009182045,0.000030243527,0.000018952845,0.00011273693,0.0000612628,0.9630184,0.017114095,0.011492025,0.007046637,0.00005200487],"about_ca_topic_score_codex":0.0076475963,"about_ca_topic_score_gemma":0.012415,"teacher_disagreement_score":0.0076475963,"about_ca_system_score_codex":0.0015454509,"about_ca_system_score_gemma":0.0025135274,"threshold_uncertainty_score":0.02179104},"labels":[],"label_agreement":null},{"id":"W4409567869","doi":"10.1093/bioinformatics/btaf173","title":"argNorm: normalization of antibiotic resistance gene annotations to the Antibiotic Resistance Ontology (ARO)","year":2025,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Pharmaceutical and Antibiotic Environmental Impacts","field":"Environmental Science","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Dalhousie University","funders":"National Health and Medical Research Council; Institut National de la Santé et de la Recherche Médicale; Medical Research Council; Università degli Studi di Trento; European Molecular Biology Laboratory; University of Pretoria; International Development Research Centre; Joint Programming Initiative on Antimicrobial Resistance","keywords":"Antibiotics; Antibiotic resistance; Normalization (sociology); Gene ontology; Gene; Computer science; Computational biology; Biology; Microbiology; Genetics; Gene expression; Sociology","score_opus":0.015642746211059447,"score_gpt":0.2729083488235146,"score_spread":0.2572656026124552,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4409567869","genre_codex":"software","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.043678828,0.0008413628,0.2760787,0.00093612925,0.00094382145,0.00079297327,0.30299878,0.35256183,0.021167582],"genre_scores_gemma":[0.06929493,0.00048018372,0.40160853,0.0010058957,0.00012743725,0.0019689186,0.45164648,0.065597035,0.008270619],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99473876,0.0008616392,0.00071645353,0.0016770101,0.0016914998,0.00031463962],"domain_scores_gemma":[0.9951383,0.0015615213,0.00077049347,0.00091462245,0.0014014846,0.00021356711],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0065624462,0.002545312,0.0013376889,0.0043194434,0.0019346782,0.0037922086,0.002214601,0.00093789934,0.016049888],"category_scores_gemma":[0.014866463,0.0009819551,0.0020333729,0.0043606325,0.00095046323,0.0029753242,0.0037270945,0.0029689595,0.015313938],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0025574877,0.00064135634,0.027799668,0.0068738298,0.0008105436,0.0007444271,0.002958215,0.009250084,0.097422145,0.017463394,0.61584663,0.21763225],"study_design_scores_gemma":[0.00030753904,0.00031962476,0.032667752,0.0007834751,0.00028651484,0.0010348704,0.0010967447,0.052564807,0.15434031,0.033243023,0.7228748,0.00048059327],"about_ca_topic_score_codex":0.0035385115,"about_ca_topic_score_gemma":0.004473497,"teacher_disagreement_score":0.016049888,"about_ca_system_score_codex":0.0014486939,"about_ca_system_score_gemma":0.0027220116,"threshold_uncertainty_score":0.05369222},"labels":[],"label_agreement":null},{"id":"W4409886452","doi":"10.1093/bioinformatics/btaf269","title":"PICRUSt2-SC: an update to the reference database used for functional prediction within PICRUSt2","year":2025,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":39,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Dalhousie University","funders":"Natural Sciences and Engineering Research Council of Canada; Alliance de recherche numérique du Canada","keywords":"Database; Computer science; Source code; Code (set theory); Process (computing); Genome; Information retrieval; Set (abstract data type); Programming language; Biology","score_opus":0.027568846823798364,"score_gpt":0.2633363834784288,"score_spread":0.23576753665463043,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4409886452","genre_codex":"dataset","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.06412566,0.008881992,0.15484859,0.001983727,0.003198675,0.0006277594,0.43426254,0.3033862,0.028684765],"genre_scores_gemma":[0.021462135,0.0013242174,0.07651686,0.000595607,0.00021598842,0.00048170608,0.8721174,0.023449382,0.0038367505],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99724764,0.00029565507,0.00040865468,0.0007039478,0.0010423327,0.0003016249],"domain_scores_gemma":[0.9955088,0.00096796936,0.00060353155,0.0009917093,0.0012742497,0.0006537412],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004341065,0.0038918473,0.0026195217,0.0046746023,0.0018660892,0.004966551,0.0059143812,0.0023106632,0.019887762],"category_scores_gemma":[0.010388275,0.0021492827,0.0030337898,0.005829804,0.00076604786,0.0041712625,0.0035998127,0.0043071476,0.02965902],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0056680446,0.00041798342,0.013097595,0.0068112477,0.0008062106,0.0015725052,0.00096894306,0.0049244966,0.08975204,0.009147656,0.72160935,0.14522387],"study_design_scores_gemma":[0.0005498394,0.0005000227,0.011983145,0.00064337475,0.00054245826,0.0024425017,0.00021279743,0.017043024,0.05215144,0.005568441,0.90790814,0.00045484537],"about_ca_topic_score_codex":0.0037521184,"about_ca_topic_score_gemma":0.0034488693,"teacher_disagreement_score":0.019887762,"about_ca_system_score_codex":0.0012766472,"about_ca_system_score_gemma":0.003293687,"threshold_uncertainty_score":0.06653124},"labels":[],"label_agreement":null},{"id":"W4410812090","doi":"10.1093/bioinformatics/btaf325","title":"HSDSnake: a user-friendly SnakeMake pipeline for analysis of duplicate genes in eukaryotic genomes","year":2025,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University; Dalhousie University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Computer science; Pipeline (software); Workflow; Genome; Novelty; Gene duplication; Gene; KEGG; Computational biology; Data mining; Similarity (geometry); Information retrieval; Database; Biology; Genetics; Gene ontology; Artificial intelligence; Programming language","score_opus":0.012807122185724927,"score_gpt":0.2637024476158325,"score_spread":0.25089532543010756,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4410812090","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.010330652,0.0005461543,0.21040124,0.0004142065,0.0002867103,0.00036159926,0.07672283,0.69715416,0.0037824006],"genre_scores_gemma":[0.067235366,0.00063848833,0.57678413,0.0010396068,0.0001091424,0.0017049427,0.20006844,0.14158869,0.01083117],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9991586,0.000121106816,0.00007611497,0.00035258126,0.00019834854,0.00009324877],"domain_scores_gemma":[0.9989478,0.00045757974,0.0001049339,0.00020370174,0.00018924758,0.00009682483],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0023061761,0.0021482755,0.0016050624,0.001948317,0.0014331739,0.0024444666,0.0037995598,0.0013634833,0.045888066],"category_scores_gemma":[0.0064011095,0.0020730589,0.003039164,0.0013367758,0.00067602645,0.003082008,0.003435344,0.00295444,0.032918617],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0021432466,0.00015401866,0.008993817,0.004739241,0.0010771074,0.0007813214,0.0013458104,0.0074357702,0.059111744,0.007958041,0.7742198,0.13204013],"study_design_scores_gemma":[0.0012279159,0.000439324,0.019917062,0.0006791349,0.00060132827,0.0017952629,0.0006739347,0.1235772,0.120164834,0.036723014,0.6931281,0.0010728892],"about_ca_topic_score_codex":0.0031741671,"about_ca_topic_score_gemma":0.004963067,"teacher_disagreement_score":0.045888066,"about_ca_system_score_codex":0.00084107,"about_ca_system_score_gemma":0.0019457085,"threshold_uncertainty_score":0.15351087},"labels":[],"label_agreement":null},{"id":"W4411279248","doi":"10.1093/bioinformatics/btaf344","title":"PhyClone: accurate Bayesian reconstruction of cancer phylogenies from bulk sequencing","year":2025,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Cancer Genomics and Diagnostics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia; BC Cancer Agency","funders":"Natural Sciences and Engineering Research Council of Canada; BC Cancer Foundation; Michael Smith Health Research BC","keywords":"Bayesian probability; Computer science; Computational biology; DNA sequencing; Biology; Artificial intelligence; Genetics; Gene","score_opus":0.009649039802931908,"score_gpt":0.252707724374754,"score_spread":0.24305868457182206,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4411279248","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.022422276,0.0010249256,0.9525243,0.0007899154,0.00010091795,0.0001316542,0.008886368,0.012732235,0.0013874101],"genre_scores_gemma":[0.2150253,0.0012615548,0.7278222,0.0014106481,0.00029325244,0.0006934136,0.04548402,0.004440842,0.0035687338],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99854773,0.0005850318,0.00007166517,0.00040146883,0.00030245804,0.00009160759],"domain_scores_gemma":[0.99408454,0.0040367874,0.0003978746,0.00068835786,0.000550931,0.00024158381],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.006117995,0.0014509652,0.0019131901,0.00227912,0.0011282687,0.0027072409,0.0036486,0.0030882484,0.006724362],"category_scores_gemma":[0.022338556,0.0019234795,0.0026316752,0.0024484978,0.0015308041,0.0026400855,0.0025325625,0.0036657911,0.004386368],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010626533,0.00020358124,0.027593423,0.0011215156,0.0010109368,0.00066115183,0.0004332374,0.71982217,0.013177823,0.025556332,0.061797448,0.14755973],"study_design_scores_gemma":[0.00010201142,0.000040892977,0.0014704627,0.0000651168,0.000073420095,0.0003096371,0.000028888848,0.96223015,0.002449152,0.027064636,0.006117648,0.0000479469],"about_ca_topic_score_codex":0.008980636,"about_ca_topic_score_gemma":0.012754095,"teacher_disagreement_score":0.008980636,"about_ca_system_score_codex":0.0012696898,"about_ca_system_score_gemma":0.0030151645,"threshold_uncertainty_score":0.032355487},"labels":[],"label_agreement":null},{"id":"W4411297467","doi":"10.1093/bioinformatics/btaf345","title":"Securing diagonal integration of multimodal single-cell data against ambiguous mapping","year":2025,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Single-cell and spatial transcriptomics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Diagonal; Computer science; Software; Artificial intelligence; Computational biology; Mathematics; Biology; Programming language; Geometry","score_opus":0.030690999028713445,"score_gpt":0.24716306605043412,"score_spread":0.21647206702172067,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4411297467","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.023035865,0.00058272335,0.961891,0.0005270006,0.00019094434,0.00015233619,0.0009763398,0.011074205,0.0015695475],"genre_scores_gemma":[0.14976017,0.00039643305,0.8400329,0.000767231,0.000098943325,0.00033714503,0.0046074684,0.0025310717,0.0014686193],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9940837,0.0012251551,0.00055406836,0.0016415524,0.0021687492,0.0003267102],"domain_scores_gemma":[0.9855295,0.0051198485,0.0015361003,0.0038636047,0.0034985044,0.00045257344],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008722802,0.0018672985,0.0016260613,0.002323914,0.0013998769,0.0035682332,0.0031959128,0.0021680798,0.0032206192],"category_scores_gemma":[0.036185183,0.0009544952,0.0019885916,0.0016754095,0.0024812792,0.0031810815,0.007528969,0.0026875136,0.0033796418],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00083811325,0.0002622741,0.039026465,0.0018734583,0.0006131352,0.0015900971,0.0030184283,0.08104205,0.20701888,0.047262926,0.031637073,0.58581716],"study_design_scores_gemma":[0.00008924652,0.0002259402,0.0067590987,0.0003159077,0.00017161653,0.0012638248,0.0011064357,0.6654408,0.20775245,0.07672518,0.039958697,0.00019076226],"about_ca_topic_score_codex":0.0028441497,"about_ca_topic_score_gemma":0.004271365,"teacher_disagreement_score":0.008722802,"about_ca_system_score_codex":0.0010141139,"about_ca_system_score_gemma":0.0032342426,"threshold_uncertainty_score":0.046131134},"labels":[],"label_agreement":null},{"id":"W4411705236","doi":"10.1093/bioinformatics/btaf375","title":"Benchmarking peak calling methods for CUT&amp;RUN","year":2025,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Princess Margaret Cancer Centre","funders":"Canadian Institutes of Health Research","keywords":"Benchmarking; Computer science; Chromatin; Chromatin immunoprecipitation; Histone; Micrococcal nuclease; Identification (biology); Bottleneck; Data mining; Benchmark (surveying); Computational biology; Biology; Nucleosome; DNA; Genetics; Gene; Gene expression","score_opus":0.015096092663729303,"score_gpt":0.319743551025407,"score_spread":0.3046474583616777,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4411705236","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":"methods","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":"methods","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.15648943,0.004932469,0.59305304,0.0009939163,0.0009857481,0.0010551214,0.046604086,0.1882159,0.00767023],"genre_scores_gemma":[0.12561469,0.0011748867,0.69484025,0.0007641117,0.00012857497,0.0017800456,0.1420579,0.030437242,0.0032024018],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.98847055,0.0019973735,0.0012502713,0.003646471,0.004027021,0.0006084121],"domain_scores_gemma":[0.97856873,0.010301568,0.001187998,0.003688572,0.005530621,0.00072252244],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.016836032,0.0034703708,0.001661201,0.0049733184,0.002125754,0.0043887035,0.005092367,0.0019465467,0.006356681],"category_scores_gemma":[0.035160743,0.0012151026,0.0028560346,0.004481063,0.0013288358,0.0024194932,0.0032612884,0.002881933,0.007655026],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.010987072,0.00127396,0.058907095,0.008629835,0.004489609,0.0010637565,0.0025493416,0.08181668,0.22681291,0.01031899,0.14319624,0.4499545],"study_design_scores_gemma":[0.0005696389,0.0011083565,0.039974056,0.00048378092,0.000711523,0.0016133168,0.0007016003,0.37278882,0.44927204,0.013272,0.11866294,0.0008420251],"about_ca_topic_score_codex":0.0045081982,"about_ca_topic_score_gemma":0.0049604815,"teacher_disagreement_score":0.98316395,"about_ca_system_score_codex":0.0014438948,"about_ca_system_score_gemma":0.0026132057,"threshold_uncertainty_score":0.08903849},"labels":[],"label_agreement":null},{"id":"W4411803876","doi":"10.1093/bioinformatics/btaf382","title":"AutoPM3: enhancing variant interpretation via LLM-driven PM3 evidence extraction from scientific literature","year":2025,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Rare Diseases","field":"Biochemistry, Genetics and Molecular Biology","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Genome Canada","funders":"","keywords":"Interpretation (philosophy); Extraction (chemistry); Computer science; Scientific evidence; Epistemology; Artificial intelligence; Data science; Chemistry; Chromatography; Programming language; Philosophy","score_opus":0.00672845336697882,"score_gpt":0.257173390281499,"score_spread":0.2504449369145202,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4411803876","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.024491426,0.008857735,0.5851824,0.004839581,0.00077511015,0.0012611817,0.17986241,0.18651469,0.008215495],"genre_scores_gemma":[0.09207424,0.0021062996,0.70124346,0.0018651048,0.0004662522,0.0009841365,0.19038905,0.0074643195,0.0034072287],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99658895,0.00091181806,0.00059629383,0.00094868935,0.0008101951,0.00014406166],"domain_scores_gemma":[0.9841195,0.010869066,0.00096604746,0.001979774,0.0017013951,0.0003641785],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004882439,0.0023173497,0.0011555293,0.010849118,0.0009813327,0.0038592205,0.0025239415,0.0014957429,0.0186641],"category_scores_gemma":[0.03722819,0.0008669953,0.0035391252,0.0057111494,0.0007216178,0.0024914583,0.0050920458,0.0014565927,0.010298566],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011180954,0.00021409335,0.023354854,0.01508611,0.0016535452,0.003146481,0.0011962175,0.012836971,0.03237334,0.0105613,0.3055619,0.5928971],"study_design_scores_gemma":[0.0010952112,0.0005313781,0.018135715,0.0027204622,0.0022100648,0.007045498,0.0009964564,0.25560835,0.08098541,0.092929736,0.53730994,0.0004317556],"about_ca_topic_score_codex":0.0046303556,"about_ca_topic_score_gemma":0.011415336,"teacher_disagreement_score":0.0186641,"about_ca_system_score_codex":0.0009416644,"about_ca_system_score_gemma":0.0050397036,"threshold_uncertainty_score":0.062437594},"labels":[],"label_agreement":null},{"id":"W4412347897","doi":"10.1093/bioinformatics/btaf401","title":"shinyDSP: a Shiny application for interactive analysis and visualization of NanoString GeoMx Whole Transcriptome Atlas data","year":2025,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Single-cell and spatial transcriptomics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"Physicians' Services Incorporated Foundation; Academic Medical Organization of Southwestern Ontario","keywords":"Bioconductor; MIT License; Computer science; Personalization; Normalization (sociology); Visualization; Profiling (computer programming); Open source; Transcriptome; Software; World Wide Web; Information retrieval; Data mining; Operating system; Biology; Gene; Gene expression","score_opus":0.015490822766732453,"score_gpt":0.28644973302687343,"score_spread":0.27095891026014096,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4412347897","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0060996953,0.00053036807,0.18344809,0.00042353058,0.00037076225,0.00037775087,0.15778688,0.6449934,0.005969433],"genre_scores_gemma":[0.056421388,0.0011132809,0.39779732,0.0014426097,0.00017463863,0.0040579867,0.2689285,0.2509363,0.01912804],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99914396,0.00010678882,0.00008856702,0.00026055737,0.00028489897,0.00011515],"domain_scores_gemma":[0.9984263,0.0006177535,0.00013762822,0.00030455753,0.00038723336,0.00012661144],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0022920172,0.0018821376,0.0015007543,0.0023055505,0.00092277786,0.0017680235,0.0026478183,0.0009862586,0.11747832],"category_scores_gemma":[0.004779428,0.0013811785,0.0015956964,0.0019916943,0.0005511932,0.001839404,0.0036644521,0.002028801,0.04354361],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008145974,0.000048267,0.0024207588,0.0019160217,0.0002843681,0.00050213415,0.0006646151,0.0011272824,0.027019428,0.0027738498,0.89746374,0.06496498],"study_design_scores_gemma":[0.0009308202,0.00021516834,0.015799953,0.00075235474,0.0002279116,0.0013798354,0.00053173176,0.04791443,0.08445726,0.020345796,0.8269935,0.00045113504],"about_ca_topic_score_codex":0.0048059723,"about_ca_topic_score_gemma":0.008788505,"teacher_disagreement_score":0.11747832,"about_ca_system_score_codex":0.00097518135,"about_ca_system_score_gemma":0.002007627,"threshold_uncertainty_score":0.39300412},"labels":[],"label_agreement":null},{"id":"W4412454105","doi":"10.1093/bioinformatics/btaf259","title":"ADME-drug-likeness: enriching molecular foundation models via pharmacokinetics-guided multi-task learning for drug-likeness prediction","year":2025,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Computational Drug Discovery Methods","field":"Computer Science","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Artificial Intelligence in Medicine (Canada)","funders":"Ministry of Science and ICT, South Korea; Seoul National University; National Research Foundation; AIGENDRUG","keywords":"ADME; Computer science; Drug discovery; Drug development; Drug; Machine learning; Artificial intelligence; Computational biology; Pharmacology; Bioinformatics; Medicine; Biology","score_opus":0.027931166727851352,"score_gpt":0.321388499425966,"score_spread":0.29345733269811464,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4412454105","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0850166,0.0020150745,0.9029647,0.0013902647,0.00013961001,0.00013788485,0.0007780865,0.005151671,0.0024062148],"genre_scores_gemma":[0.7689972,0.0006750013,0.22135307,0.0013006376,0.00017209274,0.00028203175,0.0029363784,0.00044747168,0.0038361603],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9993255,0.00023370104,0.00003940047,0.00021696067,0.00012166654,0.00006281376],"domain_scores_gemma":[0.99781895,0.0013022533,0.00023851545,0.0002360453,0.00025337326,0.0001508716],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0023325146,0.0014467238,0.0012724838,0.0010029471,0.0004309365,0.0011194209,0.0016417925,0.001798523,0.0028979243],"category_scores_gemma":[0.0051277825,0.0006396137,0.0018742684,0.0006514902,0.00076467666,0.0016631729,0.0015229159,0.0028692398,0.000978413],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003220474,0.0003208509,0.0032324742,0.00021711201,0.00016639112,0.00015878682,0.000056847453,0.8479946,0.0061417976,0.004569571,0.005236701,0.13158278],"study_design_scores_gemma":[0.000009454562,0.000043573724,0.00008226138,0.000004949758,0.000008407922,0.000013729752,0.00000228791,0.9961265,0.00082300184,0.0025888828,0.00029153095,0.0000054700595],"about_ca_topic_score_codex":0.0024056355,"about_ca_topic_score_gemma":0.0036652826,"teacher_disagreement_score":0.0028979243,"about_ca_system_score_codex":0.0009806458,"about_ca_system_score_gemma":0.0014619401,"threshold_uncertainty_score":0.012335658},"labels":[],"label_agreement":null},{"id":"W4412454224","doi":"10.1093/bioinformatics/btaf238","title":"MixingDTA: improved drug–target affinity prediction by extending mixup with guilt-by-association","year":2025,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Computational Drug Discovery Methods","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Artificial Intelligence in Medicine (Canada)","funders":"Institute for Information and Communications Technology Promotion; Seoul National University; National Research Foundation; Ministry of Science and ICT, South Korea; Neurosciences Research Foundation; AIGENDRUG","keywords":"Computer science; Drug target; Binary classification; Generalizability theory; Benchmark (surveying); Machine learning; Artificial intelligence; Drug discovery; Task (project management); Chemical space; Regression; Data mining; Support vector machine; Bioinformatics; Mathematics; Statistics; Medicine","score_opus":0.005990640469809734,"score_gpt":0.24150881920899692,"score_spread":0.23551817873918718,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4412454224","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.12953445,0.0033109053,0.8407032,0.0018660559,0.0003576095,0.00020682109,0.0025371907,0.018105602,0.0033781978],"genre_scores_gemma":[0.72894603,0.0006234241,0.25602183,0.0013141193,0.00023530031,0.00029222335,0.0065476377,0.0006806524,0.005338769],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9989072,0.0003675033,0.00007750349,0.00033824495,0.00021142849,0.00009798171],"domain_scores_gemma":[0.99723047,0.0014172683,0.0002396252,0.00056292675,0.0003508413,0.00019880694],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0024709783,0.002085091,0.0017225344,0.0014151003,0.0006602892,0.0012644582,0.0025573156,0.0019140927,0.0029206644],"category_scores_gemma":[0.0048068343,0.00073283643,0.0021329802,0.0010042393,0.00088827923,0.002786882,0.002708619,0.0037260638,0.0020651154],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00078021997,0.00060268334,0.009697433,0.0003135072,0.00040856513,0.0003191637,0.00020565763,0.6263693,0.010778496,0.007668459,0.018964887,0.3238917],"study_design_scores_gemma":[0.000011820479,0.000057529764,0.0001592364,0.0000055453747,0.000013032072,0.00002759903,0.000008438833,0.99479735,0.0013936672,0.0028967785,0.0006198064,0.000009102496],"about_ca_topic_score_codex":0.004432528,"about_ca_topic_score_gemma":0.006753515,"teacher_disagreement_score":0.004432528,"about_ca_system_score_codex":0.0009718047,"about_ca_system_score_gemma":0.0014649894,"threshold_uncertainty_score":0.013067961},"labels":[],"label_agreement":null},{"id":"W4412492686","doi":"10.1093/bioinformatics/btaf407","title":"<tt>AdDeam</tt> : a fast and scalable tool for estimating and clustering reference-level damage profiles","year":2025,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Forensic and Genetic Research","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université Laval","funders":"Novo Nordisk","keywords":"Cluster analysis; Computer science; Scalability; Software; Data mining; Artificial intelligence; Database; Programming language","score_opus":0.028539724186112926,"score_gpt":0.29874605660229886,"score_spread":0.2702063324161859,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4412492686","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0033991798,0.00040537945,0.19011718,0.00034181718,0.0003166902,0.00031160025,0.06728789,0.7355033,0.0023169063],"genre_scores_gemma":[0.041451912,0.0006426564,0.54594475,0.0011973111,0.00027619372,0.0025656256,0.27157363,0.12929694,0.0070510013],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99661523,0.00038596478,0.0002835001,0.0012442265,0.001200932,0.00027016862],"domain_scores_gemma":[0.9946713,0.0021667967,0.0006047793,0.0011751724,0.0010112637,0.00037065241],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003774658,0.006993513,0.00300481,0.007274015,0.0028154172,0.005479012,0.0073095323,0.0033598167,0.046149146],"category_scores_gemma":[0.020123202,0.0030947644,0.0045170602,0.0051175226,0.0014267232,0.0048181796,0.0068593198,0.004185701,0.04042243],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0018064023,0.0003312537,0.012330954,0.0042881253,0.0013343064,0.0011676995,0.0010660798,0.023509737,0.022221142,0.0069976766,0.7120963,0.21285027],"study_design_scores_gemma":[0.0012056587,0.00035898876,0.015844382,0.0008891983,0.0004340754,0.0012016497,0.00048457636,0.39633414,0.0734705,0.052851845,0.45574722,0.001177728],"about_ca_topic_score_codex":0.0083131045,"about_ca_topic_score_gemma":0.012127063,"teacher_disagreement_score":0.046149146,"about_ca_system_score_codex":0.0015666232,"about_ca_system_score_gemma":0.0024907023,"threshold_uncertainty_score":0.15438426},"labels":[],"label_agreement":null},{"id":"W4413396779","doi":"10.1093/bioinformatics/btaf460","title":"PDP-Miner: an AI/ML tool to detect prophage tail proteins with depolymerase domains across thousands of bacterial genomes","year":2025,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Machine Learning in Bioinformatics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université Laval","funders":"","keywords":"Prophage; Genome; Bacteriophage; Bacterial genome size; Source code; Computational biology; Computer science; Biology; Software; Gene; Genetics; Programming language; Escherichia coli","score_opus":0.00454705534616463,"score_gpt":0.2597397804878549,"score_spread":0.2551927251416903,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413396779","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.020229142,0.0010991001,0.24513221,0.0007389995,0.00022243537,0.00048380354,0.06394197,0.66518533,0.0029670554],"genre_scores_gemma":[0.10139239,0.00095654174,0.6786502,0.00092282053,0.00016402289,0.0019037841,0.17407045,0.03713542,0.0048043206],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99860877,0.00023279982,0.00015797578,0.0005587598,0.00035822834,0.00008357025],"domain_scores_gemma":[0.9967957,0.0021592379,0.00033238053,0.0002872498,0.00024551316,0.00017994293],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0028974297,0.0036260672,0.0014702331,0.0034808265,0.0011702825,0.0021038875,0.003939617,0.0016223637,0.013621309],"category_scores_gemma":[0.009196552,0.0016703416,0.0018802707,0.0020969282,0.00076743757,0.0022793282,0.0031174112,0.0025657315,0.009669773],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0032988642,0.000721762,0.027709855,0.0061248345,0.0016289591,0.0016471512,0.0007986457,0.033849,0.06444778,0.0056047807,0.47792563,0.37624273],"study_design_scores_gemma":[0.0013749333,0.0006121404,0.010880602,0.00044493214,0.00050220377,0.0014100566,0.00033277765,0.6555301,0.09593874,0.02417365,0.20851578,0.00028409334],"about_ca_topic_score_codex":0.0013466022,"about_ca_topic_score_gemma":0.0022204027,"teacher_disagreement_score":0.013621309,"about_ca_system_score_codex":0.0005601019,"about_ca_system_score_gemma":0.0017996638,"threshold_uncertainty_score":0.04556781},"labels":[],"label_agreement":null},{"id":"W4413445904","doi":"10.1093/bioinformatics/btaf465","title":"Privacy-preserving federated unsupervised domain adaptation with application to age prediction from DNA methylation data","year":2025,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Privacy-Preserving Technologies in Data","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Privy Council Office","funders":"Bundesministerium für Bildung und Forschung","keywords":"Computer science; Adaptation (eye); DNA methylation; Domain (mathematical analysis); Data mining; Computational biology; Biology; Genetics; Gene; Mathematics","score_opus":0.03585036117643044,"score_gpt":0.27262643966513717,"score_spread":0.23677607848870674,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413445904","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.01839278,0.00018833447,0.97802585,0.00044342826,0.000052008276,0.000050057908,0.00025087383,0.00202088,0.00057580956],"genre_scores_gemma":[0.5860102,0.00033395042,0.40792224,0.00075548945,0.0001662064,0.00028794736,0.0012254259,0.00029974245,0.0029988182],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9970477,0.001304786,0.00013166526,0.00075206393,0.00055019755,0.00021357831],"domain_scores_gemma":[0.9936991,0.002795875,0.00042903092,0.0021642442,0.00069514546,0.00021662647],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004958381,0.00061773445,0.0010358578,0.0006554239,0.00067910145,0.0012766734,0.0016608451,0.0012778102,0.0011778448],"category_scores_gemma":[0.015145017,0.0002921517,0.001109483,0.0010096822,0.0013356545,0.0019276224,0.0027940753,0.0023493895,0.0007169334],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00068722514,0.0004269113,0.0081376815,0.0001376482,0.00020102902,0.0004695541,0.0003974766,0.55264014,0.014780095,0.023986904,0.0096608745,0.3884745],"study_design_scores_gemma":[0.000027767783,0.000040787552,0.0006028726,0.000010366903,0.000012520445,0.00013221319,0.000034832156,0.96490353,0.006740825,0.025790105,0.0016840782,0.000020065894],"about_ca_topic_score_codex":0.0028223593,"about_ca_topic_score_gemma":0.0029899508,"teacher_disagreement_score":0.004958381,"about_ca_system_score_codex":0.0009421966,"about_ca_system_score_gemma":0.0019206892,"threshold_uncertainty_score":0.026222706},"labels":[],"label_agreement":null},{"id":"W4413491758","doi":"10.1093/bioinformatics/btaf458","title":"An interpretable alphabet for local protein structure search based on amino acid neighborhoods","year":2025,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Protein Structure and Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Canadian Aeronautics and Space Institute; Wellcome Trust","keywords":"Alphabet; Computer science; Amino acid; Computational biology; Artificial intelligence; Biology; Genetics","score_opus":0.004914227286936982,"score_gpt":0.25773394991352166,"score_spread":0.2528197226265847,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413491758","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.07335699,0.0008363401,0.88799393,0.00050378556,0.00023196501,0.00024050352,0.0038654793,0.019491674,0.013479303],"genre_scores_gemma":[0.36813238,0.00033495048,0.6217182,0.0003334897,0.00005504021,0.0004735891,0.004719842,0.001601831,0.0026305788],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9989573,0.00027569017,0.00015069725,0.00020859046,0.00033222252,0.00007553147],"domain_scores_gemma":[0.9973532,0.0012369982,0.00020412468,0.0005758553,0.000456394,0.00017343888],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011688427,0.0005008598,0.0008306899,0.0013573903,0.00064520387,0.0014247559,0.0018204143,0.0008745465,0.00610383],"category_scores_gemma":[0.007208938,0.00035105736,0.0005488707,0.0014212527,0.0007676438,0.0016253393,0.0015784664,0.000963629,0.001995666],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0021987257,0.0005715973,0.014616788,0.0013707697,0.000083954925,0.00075625896,0.00074201723,0.19140245,0.06858584,0.13512419,0.045176063,0.5393714],"study_design_scores_gemma":[0.00021119861,0.0003020554,0.0014703805,0.00016487978,0.00004193748,0.0004564506,0.0001721816,0.8173048,0.026934694,0.11554227,0.037328962,0.000070156355],"about_ca_topic_score_codex":0.0012758718,"about_ca_topic_score_gemma":0.0023650604,"teacher_disagreement_score":0.00610383,"about_ca_system_score_codex":0.0010511115,"about_ca_system_score_gemma":0.001363478,"threshold_uncertainty_score":0.020419419},"labels":[],"label_agreement":null},{"id":"W4413814807","doi":"10.1093/bioinformatics/btaf474","title":"Autocycler: long-read consensus assembly for bacterial genomes","year":2025,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":43,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Health and Medical Research Council; Institute of Infection and Immunity; Centers for Disease Control and Prevention; State Government of Victoria","keywords":"Computer science; Contig; Bacterial genome size; Sequence assembly; Nanopore sequencing; Scalability; De Bruijn graph; Genome; Automation; De Bruijn sequence; Software; Data mining; Levenshtein distance; Graph; Computational biology; Artificial intelligence; Theoretical computer science; Programming language; Biology; Database; Genetics; Engineering; Gene","score_opus":0.013589816077252527,"score_gpt":0.25762647453241855,"score_spread":0.24403665845516603,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413814807","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.014143017,0.002233523,0.74654686,0.0003685378,0.00033419763,0.0006710404,0.011706183,0.22034192,0.003654683],"genre_scores_gemma":[0.048346214,0.0009425368,0.8888788,0.00038129464,0.00011714158,0.0014298188,0.027867196,0.028853798,0.0031830983],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99595225,0.0012810243,0.00038271668,0.0011951921,0.0009821228,0.00020668532],"domain_scores_gemma":[0.995765,0.0017604736,0.0005881642,0.0007164695,0.0009045421,0.00026539952],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0052106725,0.0027623773,0.0018221871,0.0019805904,0.0014527044,0.0017861683,0.0030379957,0.0015690984,0.012050816],"category_scores_gemma":[0.011927227,0.0017249895,0.0019673093,0.0017171327,0.00091866916,0.0018572216,0.0020600753,0.003492442,0.017621113],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0028355608,0.00030318642,0.0054251854,0.0083419755,0.0007708228,0.0008357669,0.0016718215,0.04781623,0.3153249,0.017136622,0.24343589,0.356102],"study_design_scores_gemma":[0.0007178618,0.0009965217,0.004437896,0.0011133029,0.00024687705,0.0014291848,0.0002308438,0.34760344,0.35756505,0.026919222,0.25806072,0.00067907316],"about_ca_topic_score_codex":0.0019708541,"about_ca_topic_score_gemma":0.0019657642,"teacher_disagreement_score":0.012050816,"about_ca_system_score_codex":0.00084597146,"about_ca_system_score_gemma":0.001876941,"threshold_uncertainty_score":0.04031396},"labels":[],"label_agreement":null},{"id":"W4414084902","doi":"10.1093/bioinformatics/btaf497","title":"PEtab.jl: advancing the efficiency and utility of dynamic modelling","year":2025,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Gene Regulatory Network Analysis","field":"Biochemistry, Genetics and Molecular Biology","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Artificial Intelligence in Medicine (Canada)","funders":"Medical Research Council; Stiftelsen för Strategisk Forskning; Deutsche Forschungsgemeinschaft; Wellcome Trust; Francis Crick Institute; Vetenskapsrådet; Cancer Research UK","keywords":"R package; Software package; Software; Dynamic programming","score_opus":0.005400617714703178,"score_gpt":0.22796683724358083,"score_spread":0.22256621952887765,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4414084902","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0032113458,0.0002858282,0.89900833,0.00040739062,0.00016136607,0.0000726588,0.00399794,0.08659708,0.006258057],"genre_scores_gemma":[0.0686768,0.00096845254,0.84862566,0.0005687731,0.00013251785,0.00072333415,0.01822693,0.055233575,0.0068438794],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99780124,0.0006655704,0.0001995152,0.00041742102,0.0007196188,0.00019658636],"domain_scores_gemma":[0.9920397,0.004193491,0.00044567572,0.0018211283,0.0012534647,0.0002464978],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005129452,0.0022426748,0.0016107345,0.0015682926,0.0008962823,0.004067249,0.0048246775,0.0018350589,0.030344576],"category_scores_gemma":[0.017687619,0.0019848656,0.0036117104,0.0011209293,0.0010589785,0.003094132,0.004496294,0.0042364285,0.017543467],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007192308,0.00024776996,0.0048863594,0.0024302753,0.0006504858,0.00077354064,0.00058166654,0.43618134,0.019228255,0.14093225,0.20084976,0.19251908],"study_design_scores_gemma":[0.00009028445,0.000043091342,0.00033814157,0.00016275117,0.00006761942,0.0001791144,0.000033145003,0.866735,0.008344534,0.038037512,0.08589041,0.0000784544],"about_ca_topic_score_codex":0.0070132827,"about_ca_topic_score_gemma":0.0053032176,"teacher_disagreement_score":0.030344576,"about_ca_system_score_codex":0.001306302,"about_ca_system_score_gemma":0.0033275052,"threshold_uncertainty_score":0.10151273},"labels":[],"label_agreement":null},{"id":"W4414380127","doi":"10.1093/bioinformatics/btaf501","title":"ChemBounce: a computational framework for scaffold hopping in drug discovery","year":2025,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Computational Drug Discovery Methods","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Artificial Intelligence in Medicine (Canada)","funders":"National Research Foundation of Korea; Korea Research Institute of Chemical Technology","keywords":"Drug discovery; Source code; Scaffold; Code (set theory); Software","score_opus":0.019675415154723378,"score_gpt":0.3165883645346435,"score_spread":0.29691294937992013,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4414380127","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.01632244,0.0026233897,0.8532059,0.0015909055,0.00037041816,0.00079968636,0.015324034,0.08279906,0.026964169],"genre_scores_gemma":[0.15892391,0.0021978256,0.7990371,0.0010283975,0.00009424847,0.0030160644,0.020697514,0.00919593,0.0058089974],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9994677,0.0002048827,0.000029924033,0.00008151194,0.00015660826,0.000059544993],"domain_scores_gemma":[0.9986186,0.0010198514,0.000048747926,0.00010102963,0.00010710292,0.000104742176],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0017823491,0.0014858887,0.0012381615,0.0012640541,0.0007857651,0.0018026455,0.0034170907,0.0012178841,0.021446364],"category_scores_gemma":[0.0044921082,0.00072833314,0.0014729461,0.0012885169,0.0008333728,0.0012204556,0.0023392732,0.0019402945,0.0038253828],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013525124,0.00048147788,0.0026579758,0.0034440763,0.00072127074,0.0005040338,0.00034605945,0.5634996,0.0052634003,0.10743789,0.14618343,0.16810821],"study_design_scores_gemma":[0.00048710452,0.00006353413,0.00025879647,0.00012891264,0.00006824835,0.00008488495,0.000044202712,0.90613544,0.0024025955,0.03459705,0.055681553,0.000047645804],"about_ca_topic_score_codex":0.008050903,"about_ca_topic_score_gemma":0.014492439,"teacher_disagreement_score":0.021446364,"about_ca_system_score_codex":0.001235476,"about_ca_system_score_gemma":0.0029100662,"threshold_uncertainty_score":0.07174528},"labels":[],"label_agreement":null},{"id":"W4414422676","doi":"10.1093/bioinformatics/btaf530","title":"DiSTect: a Bayesian model for disease-associated gene discovery and prediction in spatial transcriptomics","year":2025,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Single-cell and spatial transcriptomics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; McGill University Health Centre","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Source code; Bayesian probability; Code (set theory); Software; Bayesian network; Transcriptome","score_opus":0.009850950237346955,"score_gpt":0.219404878304263,"score_spread":0.20955392806691606,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4414422676","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.007988417,0.00050365494,0.98746246,0.00076893374,0.000043371358,0.00009596733,0.0017594926,0.0007219473,0.0006558333],"genre_scores_gemma":[0.32412833,0.0019834794,0.65093255,0.0012803084,0.00038661773,0.0020775034,0.010275602,0.00061379,0.008321846],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99786514,0.0011807642,0.000095705815,0.0004195855,0.00030068352,0.00013808212],"domain_scores_gemma":[0.99099684,0.0073870732,0.00043034207,0.0003447471,0.0005923612,0.00024861578],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00837499,0.0011405515,0.001987342,0.0014505743,0.0007084514,0.0021541019,0.003963742,0.0023439396,0.0038761492],"category_scores_gemma":[0.018175924,0.001221022,0.0016295407,0.0022839971,0.0016935718,0.0019671987,0.0021434277,0.003453051,0.0013836675],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005110992,0.000091804286,0.005166615,0.00023273237,0.00018542219,0.00018996313,0.0001617875,0.8577022,0.0013283836,0.06295238,0.007372062,0.06410553],"study_design_scores_gemma":[0.00004152028,0.000024971505,0.0003329526,0.000018390334,0.000020990508,0.000038969967,0.000009305185,0.96404994,0.00026285206,0.03348829,0.001693035,0.000018821207],"about_ca_topic_score_codex":0.012604827,"about_ca_topic_score_gemma":0.016625477,"teacher_disagreement_score":0.012604827,"about_ca_system_score_codex":0.0020266552,"about_ca_system_score_gemma":0.002615863,"threshold_uncertainty_score":0.044291675},"labels":[],"label_agreement":null},{"id":"W4414458008","doi":"10.1093/bioinformatics/btaf531","title":"SPAED: harnessing AlphaFold output for accurate segmentation of phage endolysin domains","year":2025,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Machine Learning in Bioinformatics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université Laval","funders":"Fonds de recherche du Québec – Nature et technologies; Natural Sciences and Engineering Research Council of Canada; Vlaamse regering; Universiteit Gent; Alliance de recherche numérique du Canada; European Society of Clinical Microbiology and Infectious Diseases; Mitacs; Fonds de Recherche du Québec - Santé; Fonds Wetenschappelijk Onderzoek","keywords":"Lysin; Segmentation; Bacteriophage; Software; Pattern recognition (psychology)","score_opus":0.012178475071931668,"score_gpt":0.2942372479139277,"score_spread":0.28205877284199604,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4414458008","genre_codex":"software","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.055213273,0.0012617504,0.20942856,0.0005394597,0.0005268586,0.00020918949,0.059822764,0.6678522,0.0051459046],"genre_scores_gemma":[0.23058593,0.00093352003,0.45086434,0.0007098714,0.00017550394,0.00056705985,0.246946,0.06395377,0.005263979],"study_design_codex":"not_applicable","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9986827,0.00019868012,0.000098832286,0.0005409789,0.00035347143,0.00012534985],"domain_scores_gemma":[0.9980003,0.00073386845,0.00016162924,0.0005077023,0.0004181985,0.0001782225],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0031322306,0.0023264992,0.0013660444,0.0027463343,0.0010400172,0.0026506293,0.0022763168,0.0013753951,0.014295862],"category_scores_gemma":[0.0062836343,0.00090095215,0.0019978285,0.001526478,0.0007074947,0.002086233,0.0027635572,0.0019648185,0.012182163],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0041721566,0.00058943994,0.02365164,0.0033643348,0.0010182734,0.0007482651,0.00083355326,0.04781171,0.059951708,0.006565561,0.5335306,0.3177627],"study_design_scores_gemma":[0.00085653656,0.0004527949,0.010286717,0.00030487456,0.00017033177,0.000687126,0.00031002847,0.7386619,0.0822482,0.0209285,0.14486559,0.00022744616],"about_ca_topic_score_codex":0.0027994064,"about_ca_topic_score_gemma":0.0047073355,"teacher_disagreement_score":0.014295862,"about_ca_system_score_codex":0.00096311874,"about_ca_system_score_gemma":0.0017925866,"threshold_uncertainty_score":0.047824442},"labels":[],"label_agreement":null},{"id":"W4414474889","doi":"10.1093/bioinformatics/btag366","title":"LCR-modules: a collection of workflows for cancer genome analysis","year":2025,"lang":"en","type":"preprint","venue":"Bioinformatics","topic":"Genetics, Bioinformatics, and Biomedical Research","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"BC Cancer Agency; Spinal Cord Injury BC; Canada's Michael Smith Genome Sciences Centre; Simon Fraser University","funders":"Michael Smith Health Research BC","keywords":"Workflow; Genomics; Exome sequencing; Suite; Genome; Scalability; Flexibility (engineering); Exome; Pipeline (software)","score_opus":0.02317833863942167,"score_gpt":0.3131928350839846,"score_spread":0.29001449644456295,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4414474889","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0043067806,0.0010878567,0.39725873,0.00062632584,0.0001901169,0.0012206514,0.0687011,0.52076393,0.005844539],"genre_scores_gemma":[0.028547412,0.0013719685,0.5590123,0.0010694748,0.00015949534,0.0031910297,0.32469857,0.076188534,0.0057611945],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9960317,0.0008840844,0.0005180493,0.001069152,0.0011655093,0.00033152828],"domain_scores_gemma":[0.99356365,0.001833636,0.0004407589,0.0022226057,0.0012890466,0.00065045775],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0077425097,0.0037389107,0.0015173214,0.0047086608,0.0013088675,0.00309656,0.003806224,0.0012279296,0.01590431],"category_scores_gemma":[0.012161042,0.0021582772,0.003686646,0.0042873854,0.0009684283,0.0020955738,0.005555491,0.0033435582,0.02193085],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0016418329,0.0002796031,0.0072846273,0.0033487864,0.0008612912,0.0005863324,0.0009122977,0.021834403,0.034209926,0.017003082,0.6143039,0.29773393],"study_design_scores_gemma":[0.001065916,0.0003432666,0.012946478,0.0005960459,0.00038066585,0.0013753992,0.00020596397,0.093100525,0.08070057,0.05766224,0.7509373,0.00068561855],"about_ca_topic_score_codex":0.005463279,"about_ca_topic_score_gemma":0.006655255,"teacher_disagreement_score":0.01590431,"about_ca_system_score_codex":0.0018167105,"about_ca_system_score_gemma":0.005612588,"threshold_uncertainty_score":0.053205192},"labels":[],"label_agreement":null},{"id":"W4414874493","doi":"10.1093/bioinformatics/btaf555","title":"Doblin: inferring dominant clonal lineages from high-resolution DNA barcoding time series","year":2025,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Evolution and Genetic Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"Canadian Institutes of Health Research; Université de Montréal; Natural Sciences and Engineering Research Council of Canada; Canada Research Chairs","keywords":"DNA barcoding; Series (stratigraphy); DNA; DNA sequencing; Time series","score_opus":0.005501050349470936,"score_gpt":0.22789125407750205,"score_spread":0.22239020372803112,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4414874493","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.08903027,0.0020897868,0.81158763,0.001074517,0.00021207605,0.0001933099,0.033541918,0.05894094,0.003329459],"genre_scores_gemma":[0.32854754,0.00082136464,0.6112046,0.00083567685,0.00010711884,0.00060128584,0.0485692,0.0068901805,0.002423041],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99870336,0.0003841763,0.000074435426,0.0004905505,0.00027087826,0.000076627286],"domain_scores_gemma":[0.993954,0.0037988268,0.0008845308,0.0006447156,0.0004562152,0.0002617002],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003315215,0.0013640533,0.0011178853,0.0029395532,0.0013010064,0.0021000267,0.0022165363,0.0015938734,0.005037148],"category_scores_gemma":[0.019782819,0.00083665736,0.0009093033,0.00217062,0.0010267807,0.0015040282,0.0017707109,0.0016956297,0.0028329163],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013162058,0.00037344208,0.13730824,0.0034691824,0.0012500109,0.0011346511,0.0018620642,0.392677,0.05655159,0.027727589,0.089258045,0.28707203],"study_design_scores_gemma":[0.00011258162,0.000081587816,0.008262607,0.00016663576,0.00009201272,0.00044339345,0.00015396891,0.9332688,0.011798432,0.019372663,0.026148312,0.00009907439],"about_ca_topic_score_codex":0.009399598,"about_ca_topic_score_gemma":0.016350148,"teacher_disagreement_score":0.009399598,"about_ca_system_score_codex":0.0011850101,"about_ca_system_score_gemma":0.0019002986,"threshold_uncertainty_score":0.018689752},"labels":[],"label_agreement":null},{"id":"W4415283170","doi":"10.1093/bioinformatics/btaf578","title":"Sparse polygenic risk score inference with the spike-and-slab LASSO","year":2025,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"McGill Genome Centre; McGill University","funders":"Natural Sciences and Engineering Research Council of Canada; Canada Research Chairs","keywords":"Inference; Lasso (programming language); R package; Polygenic risk score; Covariate; Elastic net regularization; Software","score_opus":0.011044334332258661,"score_gpt":0.2495141751595947,"score_spread":0.23846984082733605,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4415283170","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.013100275,0.00016289303,0.9839643,0.0005738476,0.000036188765,0.000040213476,0.0002441015,0.00082729425,0.0010508553],"genre_scores_gemma":[0.41536814,0.00027083844,0.57804644,0.0007284463,0.00018088074,0.000301483,0.0017270959,0.00043048125,0.002946194],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99828905,0.0009730738,0.00006754786,0.00025201045,0.0003149794,0.00010323893],"domain_scores_gemma":[0.9954626,0.003159087,0.00037573377,0.00045265615,0.00038233152,0.00016762942],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0038054069,0.0011538569,0.0012479139,0.0004050186,0.00053292426,0.0009128416,0.0016994632,0.0011354133,0.0026246905],"category_scores_gemma":[0.01128637,0.0005233385,0.0010438841,0.0006886133,0.0011374073,0.0010376143,0.0015633128,0.0030172535,0.0007297443],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001613636,0.00007128096,0.002688732,0.000091294074,0.00012584568,0.000113464856,0.000083596104,0.9268791,0.0014440083,0.017075395,0.005870317,0.04539564],"study_design_scores_gemma":[0.000015994974,0.000009916816,0.00012291904,0.0000048268844,0.0000036339354,0.000009925191,0.0000050692333,0.99449337,0.00020689811,0.0048539704,0.00026893217,0.0000045052393],"about_ca_topic_score_codex":0.009301661,"about_ca_topic_score_gemma":0.009510175,"teacher_disagreement_score":0.009301661,"about_ca_system_score_codex":0.00059456495,"about_ca_system_score_gemma":0.0020849174,"threshold_uncertainty_score":0.02012521},"labels":[],"label_agreement":null},{"id":"W4415295489","doi":"10.1093/bioinformatics/btag362","title":"Movi 2: Fast and Space-Efficient Queries on Pangenomes","year":2025,"lang":"en","type":"preprint","venue":"Bioinformatics","topic":"Advanced Graph Neural Networks","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Dalhousie University","funders":"","keywords":"Memory footprint; Search engine indexing; Space (punctuation); Footprint; Compression (physics); Variety (cybernetics)","score_opus":0.015049534014978837,"score_gpt":0.24618535397210797,"score_spread":0.23113581995712915,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4415295489","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.052050844,0.0024284434,0.6111247,0.002216955,0.0006414748,0.00068784243,0.03844696,0.27608216,0.016320638],"genre_scores_gemma":[0.16074945,0.0011665463,0.737072,0.00067524763,0.00038540518,0.0009697201,0.076380126,0.015354892,0.0072466605],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9982004,0.00027313398,0.00015508273,0.0005330547,0.0006694174,0.00016890436],"domain_scores_gemma":[0.9963457,0.0012648051,0.00021603193,0.0014626124,0.00050427346,0.00020650409],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0021941692,0.0018009199,0.0011564234,0.0031901076,0.001108123,0.0032598672,0.003519155,0.0013503914,0.011957232],"category_scores_gemma":[0.015246589,0.0011666868,0.0011779543,0.0051503475,0.0009902153,0.0052907835,0.0044583813,0.0018220919,0.005692327],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014675393,0.0003338951,0.007664745,0.0016242926,0.00042432535,0.0004469911,0.00095413584,0.026626267,0.053889558,0.043242984,0.32740602,0.53591925],"study_design_scores_gemma":[0.00089596485,0.00036425763,0.005789649,0.0002174576,0.00018463358,0.0009201839,0.0006452211,0.6390938,0.08265811,0.095012195,0.17395015,0.0002683901],"about_ca_topic_score_codex":0.005022975,"about_ca_topic_score_gemma":0.008447524,"teacher_disagreement_score":0.011957232,"about_ca_system_score_codex":0.0012828972,"about_ca_system_score_gemma":0.0023227315,"threshold_uncertainty_score":0.040000916},"labels":[],"label_agreement":null},{"id":"W4415405753","doi":"10.1093/bioinformatics/btaf574","title":"aPhyloGeo: a Python application for correlating genetic and climatic conditions","year":2025,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genetic diversity and population structure","field":"Biochemistry, Genetics and Molecular Biology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Sherbrooke","funders":"","keywords":"Python (programming language); Scripting language; Software; R package; Software package","score_opus":0.0074076524019816995,"score_gpt":0.24863392399308845,"score_spread":0.24122627159110674,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4415405753","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.013438496,0.0002983165,0.24799518,0.0005080435,0.00016536063,0.00044922507,0.08311589,0.6454598,0.008569736],"genre_scores_gemma":[0.15809701,0.0007912238,0.50496477,0.0018539818,0.0002027944,0.0038695098,0.1670319,0.14929684,0.013892024],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99940026,0.00008624451,0.0000553946,0.00020619725,0.00017006334,0.00008184067],"domain_scores_gemma":[0.9991861,0.00035369073,0.00011444541,0.00011954721,0.00012799285,0.00009809968],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015624632,0.0016464179,0.0010927275,0.0014705838,0.0007948466,0.001406389,0.0022996047,0.00059798814,0.030188296],"category_scores_gemma":[0.0040575624,0.0009805644,0.0014663845,0.0011681152,0.00077426014,0.001785027,0.0033666685,0.0016670759,0.012133508],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0015660175,0.00042024525,0.039151028,0.004532398,0.0010430701,0.0011449058,0.0015563851,0.023448562,0.03027979,0.017071221,0.6514982,0.22828816],"study_design_scores_gemma":[0.00084855297,0.00018531254,0.05071677,0.0005703436,0.000296219,0.0012962465,0.00033549382,0.32252246,0.037415523,0.06893432,0.5162399,0.00063883397],"about_ca_topic_score_codex":0.004459806,"about_ca_topic_score_gemma":0.0051590307,"teacher_disagreement_score":0.030188296,"about_ca_system_score_codex":0.0006066476,"about_ca_system_score_gemma":0.002210503,"threshold_uncertainty_score":0.10098988},"labels":[],"label_agreement":null},{"id":"W4415683348","doi":"10.1093/bioinformatics/btaf582","title":"Endowing protein language models with structural knowledge","year":2025,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Biomedical Text Mining and Ontologies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Instituto de Ciencias del Mar y Limnología, Universidad Nacional Autónoma de México; Institute for Catastrophic Loss Reduction","keywords":"Code (set theory); Software; Source code; Knowledge representation and reasoning; Natural language; Language model","score_opus":0.01077066642079481,"score_gpt":0.26765891738078107,"score_spread":0.25688825095998624,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4415683348","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.035015192,0.0003559481,0.9567283,0.0008369753,0.000051300052,0.000048971582,0.0006458897,0.004429854,0.0018875727],"genre_scores_gemma":[0.52001756,0.0009182527,0.4671672,0.0005394681,0.00012596315,0.00028738735,0.0046061925,0.0007124434,0.005625514],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9994338,0.00017909845,0.000034409964,0.00015573883,0.0001555165,0.00004141358],"domain_scores_gemma":[0.9979042,0.0012017165,0.00016079961,0.0003901209,0.00026131188,0.00008175651],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011133994,0.0009349791,0.00050449435,0.00065983593,0.00029659012,0.0009826598,0.0015112638,0.00083956594,0.0031349906],"category_scores_gemma":[0.0072638057,0.0005082093,0.0007776124,0.00080158963,0.0006785199,0.0035405515,0.002305134,0.0020457434,0.002859422],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00025055197,0.00023212547,0.0024347045,0.0004112076,0.00011079162,0.00019680212,0.000201916,0.691763,0.022148967,0.034581468,0.009201369,0.2384671],"study_design_scores_gemma":[0.000007658566,0.000021853371,0.00009505645,0.000009153392,0.000009446844,0.000025887946,0.000012588112,0.9751394,0.003494923,0.019714803,0.0014621943,0.0000070246647],"about_ca_topic_score_codex":0.0023843278,"about_ca_topic_score_gemma":0.004054407,"teacher_disagreement_score":0.0031349906,"about_ca_system_score_codex":0.00066057494,"about_ca_system_score_gemma":0.0010213417,"threshold_uncertainty_score":0.010487616},"labels":[],"label_agreement":null},{"id":"W4416888770","doi":"10.1093/bioinformatics/btaf647","title":"PICNIC web server for predicting proteins involved in biomolecular condensates","year":2025,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Machine Learning in Bioinformatics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Innovation Cluster (Canada)","funders":"H2020 European Research Council; NOMIS Stiftung; Max-Planck-Gesellschaft","keywords":"Picnic; Web server; Software; Clickstream; Server; The Internet","score_opus":0.008964446904368453,"score_gpt":0.26154384707350364,"score_spread":0.2525794001691352,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4416888770","genre_codex":"software","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.050353184,0.0024391983,0.14190884,0.0010773862,0.0005344589,0.00091674423,0.24130225,0.5421923,0.019275714],"genre_scores_gemma":[0.11811979,0.0011938965,0.18390642,0.0009492846,0.00023775335,0.0016291015,0.66696095,0.018549852,0.00845292],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9995204,0.00006977988,0.00004157237,0.0001520019,0.00015895398,0.000057327812],"domain_scores_gemma":[0.9989219,0.0003177553,0.00010781838,0.00017452218,0.00029738425,0.0001808077],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012313924,0.0019416063,0.001177473,0.0021356267,0.000711648,0.0015463643,0.002689933,0.0012920555,0.023586627],"category_scores_gemma":[0.0038748095,0.00067427853,0.0008964365,0.0018440973,0.00034377226,0.0014473874,0.0013649046,0.0012162733,0.022922168],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0024799197,0.00043462767,0.014063428,0.0024351575,0.00032253767,0.00059182313,0.0001692416,0.016698975,0.019577892,0.006838957,0.8403875,0.095999874],"study_design_scores_gemma":[0.0010703101,0.00048709585,0.020107359,0.00043110098,0.00033285926,0.0016624616,0.00016890894,0.56867474,0.054738883,0.02167686,0.33040434,0.00024503883],"about_ca_topic_score_codex":0.0017624291,"about_ca_topic_score_gemma":0.0021525116,"teacher_disagreement_score":0.023586627,"about_ca_system_score_codex":0.0010162861,"about_ca_system_score_gemma":0.001779206,"threshold_uncertainty_score":0.078905165},"labels":[],"label_agreement":null},{"id":"W4417091474","doi":"10.1093/bioinformatics/btaf652","title":"Columba: fast approximate pattern matching with optimized search schemes","year":2025,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Advanced Image and Video Retrieval Techniques","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Dalhousie University","funders":"Vlaamse regering; Fonds Wetenschappelijk Onderzoek","keywords":"Source code; Scripting language; Code (set theory); Pattern matching; Matching (statistics); Software; Pattern recognition (psychology)","score_opus":0.013772872712575615,"score_gpt":0.2763447204906703,"score_spread":0.2625718477780947,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4417091474","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.010786427,0.0010123324,0.9079888,0.00032321268,0.0002359203,0.00027826344,0.003268611,0.07045916,0.0056472146],"genre_scores_gemma":[0.07584622,0.0003182024,0.901891,0.00035283825,0.000063309664,0.00078466436,0.009621361,0.005758751,0.005363635],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9983797,0.00035787115,0.000102470614,0.00030737976,0.0007050952,0.00014754444],"domain_scores_gemma":[0.9984029,0.00072589016,0.00011671753,0.00041010624,0.00026924524,0.00007510666],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014496227,0.0016798067,0.0014210086,0.0017173282,0.001035396,0.002338187,0.0041432283,0.0021963322,0.024325619],"category_scores_gemma":[0.009484234,0.0009832423,0.0012393538,0.0033090382,0.0007378377,0.0032295084,0.0025959837,0.0018228482,0.012489047],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0020493919,0.0003807997,0.002335333,0.001559153,0.00030794708,0.0003579886,0.0002247091,0.17425194,0.027768508,0.040989734,0.16899739,0.5807771],"study_design_scores_gemma":[0.00038005062,0.00013095488,0.00046112257,0.0000509677,0.00003299458,0.00020418018,0.00005237079,0.9162865,0.009420016,0.03975676,0.033163514,0.00006060489],"about_ca_topic_score_codex":0.0054649916,"about_ca_topic_score_gemma":0.007940398,"teacher_disagreement_score":0.024325619,"about_ca_system_score_codex":0.0012786582,"about_ca_system_score_gemma":0.0026970804,"threshold_uncertainty_score":0.08137727},"labels":[],"label_agreement":null},{"id":"W7116865545","doi":"10.1093/bioinformatics/btaf669","title":"RISK: a next-generation tool for biological network annotation and visualization","year":2025,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"Canadian Institutes of Health Research; Canada Research Chairs","keywords":"Python (programming language); Visualization; Documentation; MIT License; Annotation; Software; Open source; License","score_opus":0.021530494849600415,"score_gpt":0.2663043655355657,"score_spread":0.2447738706859653,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7116865545","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.002190002,0.00036448397,0.42924878,0.00068571855,0.00028669924,0.0001841504,0.039881755,0.52221304,0.004945387],"genre_scores_gemma":[0.036392376,0.0011168978,0.70286095,0.0010898592,0.00023229283,0.0017394039,0.12988392,0.11584677,0.010837555],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9986305,0.0002201661,0.00013018498,0.0002991433,0.0006031947,0.00011675639],"domain_scores_gemma":[0.99693906,0.0012939317,0.00036813907,0.00060455,0.00047148747,0.00032287146],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0029557843,0.0022963567,0.0012378396,0.0037356596,0.0009281058,0.0030662972,0.0038004336,0.0013429188,0.07809943],"category_scores_gemma":[0.01011103,0.0014227553,0.0022991288,0.002360616,0.0006867563,0.004557217,0.005053015,0.0037046508,0.033219457],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00037294504,0.00010743387,0.0023714209,0.0017171472,0.00029028955,0.00053796737,0.0007160602,0.008870997,0.012823178,0.024406223,0.8004791,0.14730725],"study_design_scores_gemma":[0.0004717406,0.00010977115,0.0044508455,0.000601574,0.00016149569,0.0014253114,0.00018664476,0.14348884,0.025765704,0.11951626,0.70343447,0.00038741538],"about_ca_topic_score_codex":0.0028344183,"about_ca_topic_score_gemma":0.003829914,"teacher_disagreement_score":0.07809943,"about_ca_system_score_codex":0.00084627775,"about_ca_system_score_gemma":0.001961935,"threshold_uncertainty_score":0.26126862},"labels":[],"label_agreement":null}]}