{"meta":{"query_hash":"0b96b08e201c","filters":{"venue":"Journal of Bioinformatics and Computational Biology"},"cohort_total":95,"direct_labels_cover":0,"predictions_cover":95,"exported":95,"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/0b96b08e201c","api":"https://metacan.xera.ac/api/v1/cohort?venue=Journal+of+Bioinformatics+and+Computational+Biology"},"results":[{"id":"W1966875588","doi":"10.1142/s0219720006002168","title":"ON THE INFERENCE OF PARSIMONIOUS INDEL EVOLUTIONARY SCENARIOS","year":2006,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","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":"McGill University","funders":"","keywords":"Indel; Divide and conquer algorithms; Inference; Phylogenetic tree; Set (abstract data type); Computer science; Tree (set theory); Multiple sequence alignment; Algorithm; Mathematics; Theoretical computer science; Sequence alignment; Biology; Artificial intelligence; Combinatorics; Genetics; Gene","score_opus":0.008418978132512304,"score_gpt":0.22929757187566646,"score_spread":0.22087859374315416,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1966875588","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":[0.117230296,0.00096900563,0.8736627,0.0029262747,0.000044596447,0.00019116313,0.00083049876,0.0007397144,0.0034057917],"genre_scores_gemma":[0.4716325,0.0007711693,0.5230741,0.00063963374,0.00027119726,0.00026950706,0.0015324391,0.00021426192,0.0015951328],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9977665,0.0011731915,0.00009086377,0.00056883873,0.00023951542,0.00016111464],"domain_scores_gemma":[0.9501616,0.04714704,0.0012390981,0.00062122545,0.00043291607,0.0003981073],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0059246826,0.0011644674,0.0019434866,0.0031527756,0.0016842723,0.0029854022,0.0032790254,0.0036547987,0.004523321],"category_scores_gemma":[0.049102414,0.0013595944,0.0015953409,0.002938863,0.002610371,0.005504788,0.001935798,0.0030197217,0.00058115175],"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":[0.00038199514,0.00017048106,0.005338122,0.00021597963,0.00009822233,0.0003750832,0.0002381672,0.898697,0.00068936736,0.036761794,0.002069849,0.05496383],"study_design_scores_gemma":[0.00006375809,0.000027142709,0.00032926162,0.00003218951,0.000018425673,0.00011784228,0.000070991955,0.9186361,0.00036640387,0.07982325,0.00050017657,0.000014432194],"about_ca_topic_score_codex":0.005179384,"about_ca_topic_score_gemma":0.004964718,"teacher_disagreement_score":0.0059246826,"about_ca_system_score_codex":0.001375644,"about_ca_system_score_gemma":0.0018084929,"threshold_uncertainty_score":0.03133309},"labels":[],"label_agreement":null},{"id":"W1969299022","doi":"10.1142/s0219720010004987","title":"PROTEIN SECONDARY STRUCTURE PREDICTION USING NMR CHEMICAL SHIFT DATA","year":2010,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Protein Structure and Dynamics","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 Waterloo","funders":"","keywords":"Computer science; k-nearest neighbors algorithm; Software; Data mining; Key (lock); Sequence (biology); Limit (mathematics); Algorithm; Pattern recognition (psychology); Artificial intelligence; Mathematics; Chemistry","score_opus":0.008312882145458547,"score_gpt":0.2507371658138717,"score_spread":0.24242428366841312,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1969299022","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":[0.11975116,0.0015771446,0.8147178,0.00040546744,0.00015779336,0.00031774252,0.014661025,0.043588832,0.0048230276],"genre_scores_gemma":[0.2112123,0.0010422101,0.7562205,0.00013068858,0.000089600646,0.00021248609,0.027487805,0.0010682174,0.002536265],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9988863,0.00022310395,0.00011163504,0.00039027198,0.00034299088,0.000045695975],"domain_scores_gemma":[0.99699116,0.0010208652,0.00043061568,0.0004688799,0.0009547142,0.00013374283],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002861082,0.0015201205,0.0018873897,0.004270379,0.000760701,0.0012660815,0.0011943147,0.0010845507,0.0036663082],"category_scores_gemma":[0.0072874143,0.0007520972,0.0013389699,0.0023583085,0.00040231086,0.0017570257,0.0008701191,0.001163337,0.0049010664],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001480242,0.00040693764,0.023957787,0.0018344725,0.0004078223,0.00067336,0.00028342634,0.056267634,0.3012128,0.0048309676,0.035177764,0.5734669],"study_design_scores_gemma":[0.0001251944,0.00023920617,0.0106390035,0.00008894103,0.00012661159,0.00051837927,0.00011331696,0.8026891,0.15429756,0.012398048,0.01865028,0.000114370945],"about_ca_topic_score_codex":0.0014579396,"about_ca_topic_score_gemma":0.0023644608,"teacher_disagreement_score":0.004270379,"about_ca_system_score_codex":0.0005892068,"about_ca_system_score_gemma":0.0009837099,"threshold_uncertainty_score":0.015131056},"labels":[],"label_agreement":null},{"id":"W1971269462","doi":"10.1142/s0219720006001722","title":"THE USE OF FUNCTIONAL DOMAINS TO IMPROVE TRANSMEMBRANE PROTEIN TOPOLOGY PREDICTION","year":2006,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Machine Learning in Bioinformatics","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 Waterloo; Caprion (Canada); University of Calgary","funders":"Natural Sciences and Engineering Research Council of Canada; University of Calgary","keywords":"Transmembrane protein; Topology (electrical circuits); Transmembrane domain; Membrane protein; Computational biology; Computer science; Biology; Mathematics; Gene; Genetics; Membrane","score_opus":0.008049881864178819,"score_gpt":0.22718057682381404,"score_spread":0.21913069495963522,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1971269462","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":[0.31212875,0.00068064645,0.6730653,0.00035597503,0.00004959181,0.00007315535,0.0011762772,0.010332756,0.0021375478],"genre_scores_gemma":[0.7200267,0.000238692,0.27742177,0.00010612611,0.000028134353,0.00005023097,0.0013498245,0.0002137653,0.0005647725],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9994137,0.0002486367,0.000047626247,0.00013042829,0.00011509478,0.000044478442],"domain_scores_gemma":[0.9975581,0.0011696706,0.0002100199,0.0003496937,0.00062440673,0.00008810901],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0019007364,0.00062248023,0.00048210745,0.001364733,0.00039960738,0.00044549545,0.00070341653,0.0007672003,0.00092675514],"category_scores_gemma":[0.005455315,0.00028905942,0.00054902973,0.00091256195,0.0002565249,0.0011924863,0.00060425943,0.0004803494,0.0009032122],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009817405,0.00026188052,0.036911972,0.0004658278,0.00019110407,0.00066534727,0.00029326457,0.200441,0.15658246,0.00457065,0.004955121,0.59367967],"study_design_scores_gemma":[0.000023327493,0.0000797107,0.004840493,0.000016720513,0.000033807202,0.0003549678,0.00002141092,0.95975596,0.029466003,0.0033122313,0.0020627838,0.000032568205],"about_ca_topic_score_codex":0.0015348793,"about_ca_topic_score_gemma":0.0016709531,"teacher_disagreement_score":0.0019007364,"about_ca_system_score_codex":0.00030756768,"about_ca_system_score_gemma":0.00042602426,"threshold_uncertainty_score":0.0100521445},"labels":[],"label_agreement":null},{"id":"W1974873233","doi":"10.1142/s0219720013410059","title":"ENHANCING GENOMICS INFORMATION RETRIEVAL THROUGH DIMENSIONAL ANALYSIS","year":2013,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Advanced Graph Neural Networks","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":"York University","funders":"Université de Neuchâtel; University of Melbourne","keywords":"Computer science; Linear subspace; Information retrieval; Dimension (graph theory); Rank (graph theory); Homogeneity (statistics); Graph; Data mining; Set (abstract data type); Genomics; Theoretical computer science; Machine learning; Mathematics; Genome","score_opus":0.00768950690829641,"score_gpt":0.23559019145499488,"score_spread":0.22790068454669846,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1974873233","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":[0.03266321,0.001573845,0.96173614,0.0004007382,0.00005896335,0.00011040331,0.00036730486,0.0015740776,0.0015152706],"genre_scores_gemma":[0.31701517,0.0014664518,0.67755115,0.00023209187,0.00023793841,0.0002689604,0.0016367582,0.00011934949,0.001472145],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9979005,0.0007080584,0.0001962002,0.00030438375,0.00075276307,0.00013815354],"domain_scores_gemma":[0.99819833,0.00057629275,0.00023758774,0.00034415853,0.00058676506,0.0000568046],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014650848,0.0010271246,0.0016774302,0.005694883,0.0008050304,0.0020609305,0.0010095825,0.00074952364,0.0009966703],"category_scores_gemma":[0.004782307,0.0003229666,0.0012771325,0.005916039,0.00050300395,0.0032281994,0.0019936129,0.0008224032,0.0008026701],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00029957358,0.0003362484,0.0057223174,0.0004293605,0.00028066165,0.00017144873,0.0004913805,0.049486108,0.04260005,0.019870419,0.0083129825,0.8719995],"study_design_scores_gemma":[0.000052357966,0.00023793266,0.0039538,0.000039618968,0.00021381136,0.0004092433,0.00034556442,0.9125256,0.026962088,0.043042857,0.012038538,0.0001785667],"about_ca_topic_score_codex":0.002242022,"about_ca_topic_score_gemma":0.0024807658,"teacher_disagreement_score":0.005694883,"about_ca_system_score_codex":0.00064481935,"about_ca_system_score_gemma":0.0009876399,"threshold_uncertainty_score":0.0077481866},"labels":[],"label_agreement":null},{"id":"W1975762434","doi":"10.1142/s0219720007002606","title":"A SYSTEMS BIOLOGY APPROACH TO THE STUDY OF CISPLATIN DRUG RESISTANCE IN OVARIAN CANCERS","year":2007,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","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":"McGill University; Indiana University-Purdue University Indianapolis","keywords":"Proteomics; Proteome; Computational biology; Cisplatin; Biology; Systems biology; Mechanism (biology); Drug resistance; Quantitative proteomics; Bioinformatics; Cell biology; Gene; Biochemistry; Genetics","score_opus":0.009068232204207032,"score_gpt":0.24920682124571908,"score_spread":0.24013858904151206,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1975762434","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":[0.18327145,0.0023398267,0.8084911,0.0012562923,0.00009531704,0.00024438772,0.00067366153,0.0004289903,0.0031988989],"genre_scores_gemma":[0.76686954,0.0021581259,0.22845614,0.00019940744,0.00006670471,0.0004956945,0.0006383747,0.000042354994,0.001073578],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99964213,0.00019811475,0.000018627521,0.000072401985,0.000048035898,0.00002067175],"domain_scores_gemma":[0.99955183,0.0003158248,0.00004363088,0.00004263277,0.000028260407,0.000017879662],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006084905,0.00044158535,0.0005572591,0.0006365818,0.00037585615,0.0007561028,0.00040378707,0.00033816314,0.00080946623],"category_scores_gemma":[0.0011588399,0.00025040272,0.00079556805,0.00050507946,0.0006979918,0.0006262106,0.00061881484,0.00073949754,0.000098632554],"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":[0.00019478034,0.00023495666,0.010124808,0.0008957182,0.000826211,0.0004880939,0.00031319738,0.6103244,0.18766294,0.14150417,0.00096149667,0.046469245],"study_design_scores_gemma":[0.00003621714,0.00028422102,0.0059365616,0.000031494998,0.00016102164,0.00016545109,0.00011134491,0.8986204,0.023744078,0.06565795,0.0052154106,0.000035741814],"about_ca_topic_score_codex":0.0014854507,"about_ca_topic_score_gemma":0.0014929887,"teacher_disagreement_score":0.0014854507,"about_ca_system_score_codex":0.0010265949,"about_ca_system_score_gemma":0.00077961973,"threshold_uncertainty_score":0.007448554},"labels":[],"label_agreement":null},{"id":"W1978319862","doi":"10.1142/s0219720006002478","title":"CHARACTERIZATION OF THE EXISTENCE OF GALLED-TREE NETWORKS","year":2006,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Algorithms and Data Compression","field":"Computer 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":"Simon Fraser University","funders":"Simon Fraser University; Slovenská Akadémia Vied","keywords":"Mathematics; Tree (set theory); Simple (philosophy); Characterization (materials science); Root (linguistics); Convexity; Tree structure; Combinatorics; Discrete mathematics; Binary tree","score_opus":0.006882667899141399,"score_gpt":0.21145192726509057,"score_spread":0.20456925936594916,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1978319862","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":[0.15624891,0.0009786546,0.8307549,0.0009441434,0.000052544656,0.00013546608,0.0007787516,0.00039181588,0.009714749],"genre_scores_gemma":[0.7716764,0.0011318049,0.22012694,0.00039792314,0.000086992804,0.00028381628,0.0018346915,0.00020826022,0.004253142],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99857175,0.00035109618,0.0001171802,0.00035754713,0.00040482898,0.00019765344],"domain_scores_gemma":[0.9847845,0.009625132,0.0014247568,0.0013136845,0.0018714675,0.0009804713],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0019763862,0.0004678649,0.0008621727,0.0025300088,0.0011889996,0.0020869502,0.0012530361,0.0014841851,0.0043285172],"category_scores_gemma":[0.01927651,0.00069026876,0.00076240033,0.0011687052,0.001858595,0.0061119124,0.003383301,0.0018045488,0.0008214614],"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":[0.00049407705,0.00011205106,0.009834197,0.00068128813,0.00018055698,0.0015870015,0.0014335498,0.11895861,0.032703627,0.74903095,0.0066677863,0.07831628],"study_design_scores_gemma":[0.00003295904,0.00010866153,0.0024836832,0.00014621658,0.0000836542,0.0012880227,0.00046413278,0.46161914,0.01784195,0.50483376,0.01102359,0.00007418083],"about_ca_topic_score_codex":0.0004520859,"about_ca_topic_score_gemma":0.00057106983,"teacher_disagreement_score":0.0043285172,"about_ca_system_score_codex":0.0006812146,"about_ca_system_score_gemma":0.0005599649,"threshold_uncertainty_score":0.014480352},"labels":[],"label_agreement":null},{"id":"W1978894908","doi":"10.1142/s021972000700262x","title":"GASA: A GRAPH-BASED AUTOMATED NMR BACKBONE RESONANCE SEQUENTIAL ASSIGNMENT PROGRAM","year":2007,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Protein Structure and 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 Alberta","funders":"","keywords":"Chaining; Assignment problem; Computer science; Algorithm; Graph; Theoretical computer science; Mathematics; Mathematical optimization","score_opus":0.007024527546030541,"score_gpt":0.2742543475937394,"score_spread":0.26722982004770884,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1978894908","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.014593214,0.00019227654,0.90810126,0.00019423309,0.000065186745,0.0002274404,0.0014909043,0.073566064,0.0015694015],"genre_scores_gemma":[0.084592685,0.00017922904,0.90422493,0.00018352483,0.000039573013,0.00039020786,0.0055602924,0.002452025,0.002377425],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9995474,0.000099860394,0.000019770483,0.00016402011,0.00012811682,0.00004078685],"domain_scores_gemma":[0.9993087,0.00030367708,0.00007841904,0.00011966888,0.00013525001,0.000054394663],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00062524504,0.0017805488,0.0008728981,0.0018863345,0.0008992793,0.000742492,0.0020628546,0.0010200089,0.0073651355],"category_scores_gemma":[0.0018732623,0.00064078,0.0014656731,0.0013159028,0.0006425949,0.0013335358,0.0013493457,0.0014192909,0.0022909706],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00089602143,0.00042050984,0.0027873039,0.0006407449,0.00022538728,0.00043389175,0.00025132264,0.2830668,0.031856537,0.013637727,0.05522113,0.6105627],"study_design_scores_gemma":[0.000086451335,0.00007514089,0.00030906903,0.0000105674535,0.000023231985,0.00007525161,0.000033687807,0.9770005,0.005567446,0.008715282,0.008081725,0.000021685819],"about_ca_topic_score_codex":0.0063785217,"about_ca_topic_score_gemma":0.0077426494,"teacher_disagreement_score":0.0073651355,"about_ca_system_score_codex":0.00068582955,"about_ca_system_score_gemma":0.0019971374,"threshold_uncertainty_score":0.024638832},"labels":[],"label_agreement":null},{"id":"W1979548025","doi":"10.1142/s0219720011005355","title":"IMPROVED SEQUENCE-BASED PREDICTION OF STRAND RESIDUES","year":2010,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","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":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Artificial intelligence; Computational biology; Classifier (UML); Pattern recognition (psychology); Computer science; Biological system; Biology; Bioinformatics","score_opus":0.009067655613722058,"score_gpt":0.25938986886793786,"score_spread":0.2503222132542158,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1979548025","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":[0.6438922,0.0023812994,0.3369226,0.00029356242,0.00012429574,0.00008653845,0.0029108615,0.008926727,0.0044619525],"genre_scores_gemma":[0.7898315,0.00062821043,0.19987915,0.00012531849,0.00006915645,0.00004577034,0.0064700483,0.000262556,0.002688244],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99958605,0.000101246485,0.000025642124,0.00009910109,0.00015571859,0.000032260457],"domain_scores_gemma":[0.9986578,0.0003770735,0.0002446752,0.00019210842,0.00044125077,0.00008717294],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000907095,0.00071796443,0.00054202374,0.0010701973,0.00020914197,0.000498548,0.0005568476,0.00047026892,0.0017947936],"category_scores_gemma":[0.002293914,0.0003557108,0.00046571487,0.0005265915,0.00018116918,0.0010254699,0.0005415497,0.0006793442,0.001712736],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0017945925,0.000338067,0.08238005,0.00057739665,0.00021482412,0.0004954726,0.00015635115,0.11931743,0.3267696,0.0029346235,0.0137228565,0.45129868],"study_design_scores_gemma":[0.00011055748,0.00041835482,0.019582912,0.000042050542,0.00008724266,0.0008239872,0.000040486728,0.84348375,0.12319461,0.0029486031,0.009199231,0.000068109824],"about_ca_topic_score_codex":0.00095793215,"about_ca_topic_score_gemma":0.0018271192,"teacher_disagreement_score":0.0017947936,"about_ca_system_score_codex":0.0002359229,"about_ca_system_score_gemma":0.0004857794,"threshold_uncertainty_score":0.0060041547},"labels":[],"label_agreement":null},{"id":"W1980611895","doi":"10.1142/s0219720006002302","title":"ITERATED LOCAL LEAST SQUARES MICROARRAY MISSING VALUE IMPUTATION","year":2006,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":87,"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; Chinese Academy of Sciences; University of Alberta","keywords":"Imputation (statistics); Missing data; Iterated function; Data mining; Microarray analysis techniques; Computer science; Mathematics; Statistics; Algorithm; Gene; Biology; Gene expression; Genetics","score_opus":0.006049518886673946,"score_gpt":0.24466432822405573,"score_spread":0.23861480933738177,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1980611895","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.006933752,0.00015973223,0.9915987,0.0000972358,0.00002574466,0.00003208032,0.0001625357,0.00077880506,0.00021128354],"genre_scores_gemma":[0.20003274,0.0003246528,0.79378325,0.00029089328,0.000109511355,0.0003776119,0.0021824976,0.00026152196,0.002637348],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9945252,0.0028627068,0.0002768797,0.001147606,0.00088436797,0.0003033785],"domain_scores_gemma":[0.98809135,0.0063591865,0.0008734543,0.0029902132,0.0014971421,0.00018876465],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008816703,0.00097603275,0.0027821774,0.0011392784,0.0008772614,0.001416005,0.004636539,0.0017154417,0.0027682802],"category_scores_gemma":[0.020378157,0.00078752334,0.0026021814,0.0020706386,0.00075574004,0.0013733709,0.0018603578,0.0028528948,0.0016009521],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0012333074,0.0002976915,0.020326635,0.0005506266,0.001291826,0.0007363931,0.0005863579,0.35323367,0.015063402,0.018803727,0.010401707,0.57747465],"study_design_scores_gemma":[0.00005931625,0.00012230327,0.0022761775,0.00002652357,0.00009666463,0.00028069658,0.000044403205,0.9746063,0.0069148885,0.01276683,0.002754779,0.000051096857],"about_ca_topic_score_codex":0.0018824395,"about_ca_topic_score_gemma":0.002366625,"teacher_disagreement_score":0.008816703,"about_ca_system_score_codex":0.0008381607,"about_ca_system_score_gemma":0.0017823101,"threshold_uncertainty_score":0.04662776},"labels":[],"label_agreement":null},{"id":"W1981321222","doi":"10.1142/s0219720009004333","title":"ASYMPTOTICS OF CANONICAL AND SATURATED RNA SECONDARY STRUCTURES","year":2009,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"RNA and protein synthesis mechanisms","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":"Carleton University","funders":"","keywords":"Pseudoknot; Protein secondary structure; Nucleic acid secondary structure; Base (topology); Canonical form; Class (philosophy); Computation; RNA; Existential quantification","score_opus":0.0062512840611390185,"score_gpt":0.2390686562936064,"score_spread":0.2328173722324674,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1981321222","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":[0.785583,0.0013068136,0.16722484,0.00063863874,0.00013516881,0.000049586648,0.0003518479,0.00079591403,0.043914225],"genre_scores_gemma":[0.97210914,0.00051361305,0.020622801,0.00018014811,0.00013636381,0.00011639659,0.000653172,0.0002618469,0.0054064826],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9983884,0.00030302498,0.00006544943,0.00042072154,0.00053923734,0.0002831429],"domain_scores_gemma":[0.98820937,0.007265479,0.0010552022,0.0010645711,0.0014397892,0.00096564565],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015621069,0.00038689995,0.0005072269,0.0020124433,0.00080795464,0.0015525616,0.0011689222,0.0006851984,0.0054976554],"category_scores_gemma":[0.020209948,0.00045685473,0.00079123286,0.0006266742,0.0023988753,0.0032445646,0.0015621597,0.0012015284,0.00070996705],"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":[0.00032575885,0.0001204341,0.022940937,0.00032627015,0.00006190342,0.0006633392,0.00044302735,0.053509384,0.017021013,0.8601967,0.003990234,0.040401075],"study_design_scores_gemma":[0.00002628324,0.00012743758,0.010140193,0.000078668214,0.000033871194,0.0011531345,0.0001973969,0.313372,0.00819277,0.6614492,0.0051581226,0.00007098682],"about_ca_topic_score_codex":0.0008720198,"about_ca_topic_score_gemma":0.0008216246,"teacher_disagreement_score":0.0054976554,"about_ca_system_score_codex":0.001782943,"about_ca_system_score_gemma":0.00080426905,"threshold_uncertainty_score":0.01839149},"labels":[],"label_agreement":null},{"id":"W1984673676","doi":"10.1142/s0219720008003291","title":"COMPLEXITIES AND ALGORITHMS FOR GLYCAN SEQUENCING USING TANDEM MASS SPECTROMETRY","year":2008,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Glycosylation and Glycoproteins Research","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","funders":"Center for Advanced Study, University of Illinois at Urbana-Champaign; Natural Sciences and Engineering Research Council of Canada; Tsinghua University","keywords":"Glycan; Tandem mass spectrometry; Computational biology; Glycopeptide; Computer science; Mass spectrometry; Proteomics; Algorithm; Biology; Chemistry; Biochemistry; Gene; Chromatography","score_opus":0.05674675018465132,"score_gpt":0.31674568914313656,"score_spread":0.25999893895848525,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1984673676","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":[0.003353582,0.00036328915,0.99337834,0.00043476952,0.00004283073,0.00010483752,0.00013725951,0.00085384236,0.0013313098],"genre_scores_gemma":[0.033166096,0.00053824054,0.9637533,0.00013653349,0.00009499541,0.00038643178,0.0005860336,0.00024607996,0.0010922424],"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99649566,0.0012509843,0.00027199357,0.0005482811,0.0012264067,0.00020669794],"domain_scores_gemma":[0.98473847,0.011419886,0.000797341,0.0014221368,0.0013833587,0.0002388155],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0035933584,0.0019319115,0.0016269013,0.002431248,0.0018165513,0.0030699363,0.0037952343,0.0020513993,0.00497456],"category_scores_gemma":[0.020818941,0.0015085499,0.0022038512,0.0037499478,0.0015162908,0.004641412,0.003075772,0.0034000392,0.0023371552],"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":[0.00036320926,0.00020462697,0.00156558,0.000656078,0.00013746144,0.00045948758,0.0002542348,0.5632086,0.004474307,0.08266254,0.011042564,0.33497128],"study_design_scores_gemma":[0.00006825514,0.00002922851,0.00024786865,0.00003033808,0.00002097815,0.00016881265,0.00004694979,0.85743684,0.001504864,0.13660297,0.0038148644,0.000028063649],"about_ca_topic_score_codex":0.0045498163,"about_ca_topic_score_gemma":0.0038442763,"teacher_disagreement_score":0.00497456,"about_ca_system_score_codex":0.0017715616,"about_ca_system_score_gemma":0.0029172238,"threshold_uncertainty_score":0.01900369},"labels":[],"label_agreement":null},{"id":"W1988405029","doi":"10.1142/s0219720009004102","title":"GENOME HALVING WITH DOUBLE CUT AND JOIN","year":2009,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Genome Rearrangement Algorithms","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","funders":"Natural Sciences and Engineering Research Council of Canada; Canada Research Chairs; Royal Society of Canada","keywords":"Join (topology); Reciprocal; Genome; Context (archaeology); Computer science; Block (permutation group theory); Combinatorics; Mathematics; Algorithm; Genetics; Biology; Gene","score_opus":0.007831224378292838,"score_gpt":0.231936087837947,"score_spread":0.22410486345965416,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1988405029","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":[0.041279126,0.00048548708,0.93969166,0.00044973497,0.00021380701,0.00038951542,0.0022427253,0.01273672,0.0025111777],"genre_scores_gemma":[0.06802652,0.00010302635,0.9214744,0.00015845061,0.000041043397,0.00029745893,0.0061949817,0.001368651,0.0023354178],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9977744,0.00047451712,0.00015752498,0.0008173411,0.0005512506,0.00022487591],"domain_scores_gemma":[0.99490225,0.0020061478,0.00022154271,0.002238279,0.00044824253,0.00018359075],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0031582953,0.001561773,0.0020197183,0.0022200455,0.0016428063,0.0023425126,0.004189081,0.0024906183,0.010633572],"category_scores_gemma":[0.009420035,0.0008485116,0.002329997,0.0028796198,0.0012998936,0.0029084287,0.0045289174,0.0030219879,0.0037080473],"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":[0.002346661,0.0005014622,0.00455268,0.000794201,0.00041779428,0.00046093398,0.00070967677,0.15750274,0.02991259,0.05165251,0.043578554,0.70757014],"study_design_scores_gemma":[0.0006459417,0.00031360972,0.0010607804,0.00005618591,0.0001200826,0.00067423895,0.00040580073,0.8249719,0.04629456,0.093670905,0.03170373,0.00008231971],"about_ca_topic_score_codex":0.001966744,"about_ca_topic_score_gemma":0.0026284857,"teacher_disagreement_score":0.010633572,"about_ca_system_score_codex":0.0007156054,"about_ca_system_score_gemma":0.0019392236,"threshold_uncertainty_score":0.035572767},"labels":[],"label_agreement":null},{"id":"W1989027782","doi":"10.1142/s0219720012410016","title":"IDENTIFICATION OF SHORTENED 3′ UNTRANSLATED REGIONS FROM EXPRESSION ARRAYS","year":2012,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"RNA Research and Splicing","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":"Institute of Cancer Research; Ligue Contre le Cancer; Institut National Du Cancer; European Commission","keywords":"RefSeq; Computational biology; Biology; DNA microarray; Genetics; Gene; Untranslated region; splice; Alternative splicing; Exon; Three prime untranslated region; Gene expression; Genome; Messenger RNA","score_opus":0.01353372281945734,"score_gpt":0.27811291777150116,"score_spread":0.2645791949520438,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1989027782","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":[0.33815986,0.00068272796,0.64518464,0.00014936665,0.00006487055,0.00027279533,0.010048934,0.0037301409,0.0017067783],"genre_scores_gemma":[0.31874815,0.00061700033,0.64098173,0.00021136433,0.000057166882,0.0010814931,0.035598446,0.0006482967,0.0020564492],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9991547,0.00014846327,0.00008625662,0.00030183338,0.00023717462,0.00007160534],"domain_scores_gemma":[0.9987206,0.0006581093,0.00017220076,0.0002259783,0.00018923556,0.00003392286],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00092054956,0.00064240996,0.0010232687,0.0010877795,0.00034115062,0.00069214747,0.00060772744,0.0003449487,0.0014727967],"category_scores_gemma":[0.0029777836,0.00041183044,0.0013156183,0.0015289213,0.00027736695,0.00035088457,0.0005366154,0.00081192475,0.001010221],"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":[0.0010059959,0.00012237062,0.015785128,0.00057729945,0.00029268587,0.00037697263,0.00018878971,0.028819775,0.812169,0.0026269052,0.0017491803,0.13628581],"study_design_scores_gemma":[0.0001529758,0.0008224604,0.07182117,0.000055078057,0.00058378355,0.00065549405,0.00017290773,0.3213164,0.57209355,0.009370363,0.022845177,0.00011059564],"about_ca_topic_score_codex":0.00070062897,"about_ca_topic_score_gemma":0.0015807902,"teacher_disagreement_score":0.0014727967,"about_ca_system_score_codex":0.00032378544,"about_ca_system_score_gemma":0.0009610627,"threshold_uncertainty_score":0.004927039},"labels":[],"label_agreement":null},{"id":"W1990343223","doi":"10.1142/s0219720006002442","title":"MULTIRESOLUTION ANALYSIS UNCOVERS HIDDEN CONSERVATION OF PROPERTIES IN STRUCTURALLY AND FUNCTIONALLY SIMILAR PROTEINS","year":2006,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","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":"McGill University","keywords":"Computational biology; Conserved sequence; Biology; Function (biology); Protein structure; Functional analysis; Biological system; Genetics; Peptide sequence; Computer science; Mathematics; Biochemistry; Gene","score_opus":0.005488743083350428,"score_gpt":0.20737772079378472,"score_spread":0.2018889777104343,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1990343223","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":[0.8938875,0.00045117142,0.10456412,0.0001158653,0.000012744084,0.00000856185,0.00009465078,0.0002988535,0.00056648586],"genre_scores_gemma":[0.9818215,0.00017352625,0.017627068,0.000013746962,0.000013388758,0.0000040416767,0.00017668131,0.000024199835,0.00014597696],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99967194,0.00009132303,0.000020177475,0.0000759116,0.000088951405,0.00005177595],"domain_scores_gemma":[0.9985879,0.0005510202,0.00035897648,0.00028376182,0.0001270575,0.00009124732],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009733764,0.00030610734,0.00057879306,0.0017290497,0.00027762147,0.0006201569,0.00030676322,0.00046066384,0.00040615862],"category_scores_gemma":[0.002346202,0.00018875534,0.00055001024,0.00077735004,0.0003203511,0.00076924864,0.0005580537,0.000495775,0.00022540193],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0012183401,0.00038285804,0.07797872,0.00031279898,0.00035793555,0.0011539954,0.0007852909,0.059104647,0.5700792,0.0050512953,0.0011783133,0.2823966],"study_design_scores_gemma":[0.00004112125,0.0004569202,0.1178841,0.000028936605,0.00018197265,0.0011942761,0.000355029,0.8036944,0.06799679,0.006180498,0.0019298736,0.00005604016],"about_ca_topic_score_codex":0.00073714124,"about_ca_topic_score_gemma":0.00048518382,"teacher_disagreement_score":0.0017290497,"about_ca_system_score_codex":0.00018278912,"about_ca_system_score_gemma":0.00016642727,"threshold_uncertainty_score":0.005147755},"labels":[],"label_agreement":null},{"id":"W1991265331","doi":"10.1142/s0219720012400082","title":"A FAST METHOD FOR LARGE-SCALE MULTICHROMOSOMAL BREAKPOINT MEDIAN PROBLEMS","year":2011,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Genome Rearrangement Algorithms","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 Ottawa","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Breakpoint; Computer science; Scale (ratio); Algorithm; Mathematics; Biology; Chromosome; Genetics; Cartography; Geography","score_opus":0.015277791115516364,"score_gpt":0.2636110825953793,"score_spread":0.24833329147986294,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1991265331","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":[0.0007827696,0.000057276207,0.99802184,0.000085424435,0.000028148748,0.00003619769,0.000049581853,0.00040554634,0.00053327176],"genre_scores_gemma":[0.026793567,0.00012164533,0.9701217,0.00008719581,0.00007862804,0.00032919503,0.0003119221,0.00034183986,0.0018144586],"study_design_codex":"simulation_or_modeling","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9990502,0.000261979,0.000052890104,0.00017224558,0.00037255962,0.00009013511],"domain_scores_gemma":[0.9968606,0.001963105,0.00017448557,0.00035436268,0.00050025934,0.00014720901],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002056938,0.0010919597,0.0012519534,0.0017166075,0.0011628169,0.0014071367,0.0027802836,0.0018788,0.0106415665],"category_scores_gemma":[0.007647647,0.00063234114,0.0014602551,0.002002217,0.00083469565,0.0023122402,0.003257705,0.003920838,0.0027031181],"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":[0.0001791644,0.0001421212,0.0007437016,0.00035063285,0.000104953724,0.0002310384,0.00018709096,0.5166972,0.008260273,0.108788736,0.015713608,0.34860158],"study_design_scores_gemma":[0.0000669264,0.00003807134,0.00012252785,0.000019829722,0.000012008928,0.0001069944,0.000028012275,0.91777635,0.001457516,0.07251944,0.00783023,0.000022034],"about_ca_topic_score_codex":0.0022424336,"about_ca_topic_score_gemma":0.0039004008,"teacher_disagreement_score":0.0106415665,"about_ca_system_score_codex":0.0010478961,"about_ca_system_score_gemma":0.002199307,"threshold_uncertainty_score":0.03559959},"labels":[],"label_agreement":null},{"id":"W1991772956","doi":"10.1142/s0219720010004495","title":"A MULTI-STRATEGY APPROACH TO INFORMATIVE GENE IDENTIFICATION FROM GENE EXPRESSION DATA","year":2010,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","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":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Biotechnology Research Institute; National Research Council Canada","funders":"","keywords":"Identification (biology); Computer science; Selection (genetic algorithm); Gene prediction; Data mining; Gene; Computational biology; Genomics; Artificial intelligence; Biology; Genome; Genetics","score_opus":0.036577473547878755,"score_gpt":0.30692836221762937,"score_spread":0.2703508886697506,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1991772956","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":[0.008765259,0.00026860062,0.9896327,0.00006699018,0.000013012214,0.00014688498,0.00019437287,0.0006553269,0.0002567017],"genre_scores_gemma":[0.08879731,0.0001600868,0.90901417,0.00015253769,0.000026696203,0.00041204327,0.0007909514,0.00014153044,0.00050473405],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9963025,0.0013902069,0.00024547384,0.00074773637,0.0011464991,0.0001675697],"domain_scores_gemma":[0.9957054,0.00272957,0.00032469624,0.0005183018,0.00059223775,0.00012987632],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0056095175,0.0015897676,0.0015946621,0.0071427347,0.0007809071,0.00154505,0.0015404761,0.0009722729,0.0010430317],"category_scores_gemma":[0.0077011473,0.00057435216,0.002697886,0.0034946632,0.00085168367,0.0009763095,0.0015244924,0.0014825804,0.000648224],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006345874,0.0005960695,0.018431108,0.0010080669,0.001654893,0.0009619601,0.00077844353,0.051308814,0.21420434,0.018108087,0.0035633808,0.6887502],"study_design_scores_gemma":[0.00015712965,0.0008533256,0.020323947,0.00010771436,0.00054853206,0.0021409888,0.00024661236,0.8159371,0.10050468,0.046605505,0.012302903,0.0002715671],"about_ca_topic_score_codex":0.00144168,"about_ca_topic_score_gemma":0.0026383195,"teacher_disagreement_score":0.0071427347,"about_ca_system_score_codex":0.00070249196,"about_ca_system_score_gemma":0.0014975222,"threshold_uncertainty_score":0.029666305},"labels":[],"label_agreement":null},{"id":"W1993568797","doi":"10.1142/s0219720015500134","title":"Identification of biomarker genes for resistance to a pathogen by a novel method for meta-analysis of single-channel microarray datasets","year":2015,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Mycotoxins in Agriculture and Food","field":"Agricultural and Biological Sciences","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":"Agriculture and Agri-Food Canada","funders":"","keywords":"Microarray analysis techniques; DNA microarray; Microarray; Gene; Biology; Computational biology; Gene chip analysis; Microarray databases; Identification (biology); Data mining; Computer science; Genetics; Gene expression; Botany","score_opus":0.0949890221889506,"score_gpt":0.3118014570270975,"score_spread":0.21681243483814688,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1993568797","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.053814273,0.0007757358,0.9401076,0.0001535555,0.000051821793,0.0002771622,0.0023368406,0.002147328,0.0003356805],"genre_scores_gemma":[0.18829165,0.00032475745,0.80672497,0.000093522205,0.000043613396,0.0007218065,0.0033370948,0.0001556626,0.00030690015],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9981323,0.00042291454,0.00015584768,0.0006635571,0.00054290815,0.00008245957],"domain_scores_gemma":[0.99822074,0.0009164499,0.00026110825,0.00026969053,0.0002663755,0.00006558394],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0031314462,0.0012851445,0.0018226252,0.005194767,0.00042214897,0.0016520544,0.001094128,0.00072118745,0.0005031756],"category_scores_gemma":[0.0036649588,0.0006034295,0.0031713697,0.0034860855,0.00036691173,0.0008274067,0.00088152173,0.0012912023,0.00024757796],"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":[0.0009478079,0.0008785876,0.03618273,0.0018783652,0.0070637553,0.00043315088,0.00039251035,0.05294608,0.650302,0.0026575064,0.0015796579,0.24473797],"study_design_scores_gemma":[0.00016258447,0.00079929974,0.046822526,0.00008353359,0.0023324816,0.0007473564,0.00029861595,0.79743,0.13053006,0.012176771,0.008355129,0.00026163555],"about_ca_topic_score_codex":0.0010653868,"about_ca_topic_score_gemma":0.0018252782,"teacher_disagreement_score":0.005194767,"about_ca_system_score_codex":0.000646609,"about_ca_system_score_gemma":0.0010000531,"threshold_uncertainty_score":0.016560912},"labels":[],"label_agreement":null},{"id":"W1995229312","doi":"10.1142/s0219720005001247","title":"SPIDER: SOFTWARE FOR PROTEIN IDENTIFICATION FROM SEQUENCE TAGS WITH <i>DE NOVO</i> SEQUENCING ERROR","year":2005,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Advanced Proteomics Techniques and Applications","field":"Chemistry","cited_by":219,"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":"Software; Identification (biology); Sequence (biology); Computational biology; Computer science; Sequence assembly; Protein sequencing; DNA sequencing; Protein methods; Biology; Peptide sequence; Genetics; Programming language; Gene; Transcriptome","score_opus":0.021062959777406715,"score_gpt":0.2886165250035087,"score_spread":0.26755356522610196,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1995229312","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":[0.0061976817,0.00037460407,0.60312927,0.00012564837,0.00014813145,0.0002566404,0.0062029264,0.38191608,0.001648957],"genre_scores_gemma":[0.05123097,0.0006986186,0.8288671,0.00041172514,0.000085495616,0.0012883828,0.037608713,0.06891741,0.010891594],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99865156,0.00021866495,0.00019520821,0.00027133993,0.0005497871,0.000113505834],"domain_scores_gemma":[0.9975789,0.0010724403,0.00035577515,0.0003556958,0.0005004627,0.00013679282],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0029008149,0.001995193,0.001447852,0.0022797554,0.00075353374,0.0015484042,0.0026561702,0.0010420881,0.019230224],"category_scores_gemma":[0.00675602,0.0015804922,0.0014302417,0.0015252176,0.0006465529,0.0022218905,0.0026818768,0.0021872679,0.016741265],"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":[0.0027978807,0.0002848749,0.010841208,0.0030140178,0.000841692,0.0013012129,0.0008144359,0.015165367,0.06970793,0.014753525,0.40012765,0.48035026],"study_design_scores_gemma":[0.00087591767,0.0004735568,0.010915401,0.00049321406,0.000308871,0.0044395,0.00024159484,0.3196637,0.20032711,0.04327009,0.418408,0.0005829489],"about_ca_topic_score_codex":0.0012009162,"about_ca_topic_score_gemma":0.0010897913,"teacher_disagreement_score":0.019230224,"about_ca_system_score_codex":0.00063028274,"about_ca_system_score_gemma":0.00135954,"threshold_uncertainty_score":0.06433147},"labels":[],"label_agreement":null},{"id":"W1996146483","doi":"10.1142/s0219720005001314","title":"DYNAMIC MODEL-BASED CLUSTERING FOR TIME–COURSE GENE EXPRESSION DATA","year":2005,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Gene expression and cancer classification","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 Saskatchewan","funders":"Atomic Energy of Canada Limited; Natural Sciences and Engineering Research Council of Canada; Technische Universiteit Delft; University of Saskatchewan","keywords":"Cluster analysis; Computer science; Data mining; Bootstrapping (finance); Expression (computer science); Hierarchical clustering; Artificial intelligence; Mathematics","score_opus":0.018436637173491813,"score_gpt":0.3028210396366478,"score_spread":0.284384402463156,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1996146483","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0062896293,0.00023835943,0.99242264,0.00007971053,0.000021702468,0.00004567954,0.000108741086,0.00048131033,0.00031223803],"genre_scores_gemma":[0.28331465,0.0010323222,0.70876664,0.00017718038,0.00009462227,0.0006122315,0.0029622384,0.00039300884,0.0026470704],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99835396,0.00043266208,0.00009288266,0.00056602055,0.00044579132,0.000108679524],"domain_scores_gemma":[0.9982278,0.00089341763,0.00023445702,0.00022638717,0.0003505892,0.00006740601],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0020675017,0.0011162284,0.0017383811,0.0028404843,0.001040962,0.0011441644,0.0022686569,0.0012232861,0.00091202045],"category_scores_gemma":[0.0050378772,0.00055747,0.0023793583,0.0027205106,0.0006920066,0.0012002988,0.0010073569,0.0018503943,0.0006387862],"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":[0.00010313736,0.00009788232,0.0016335159,0.00014998185,0.00018601073,0.00010829772,0.00016886134,0.8783326,0.0066969357,0.013201859,0.0017626861,0.09755816],"study_design_scores_gemma":[0.0000034258037,0.00001142999,0.0003434237,0.0000036831896,0.000009932745,0.000024804467,0.000012610456,0.99358034,0.00070840464,0.0046101236,0.0006777457,0.00001403468],"about_ca_topic_score_codex":0.013053979,"about_ca_topic_score_gemma":0.010923528,"teacher_disagreement_score":0.013053979,"about_ca_system_score_codex":0.0019381165,"about_ca_system_score_gemma":0.0015831385,"threshold_uncertainty_score":0.025955975},"labels":[],"label_agreement":null},{"id":"W1996806342","doi":"10.1142/s0219720007003107","title":"COMPUTING THE BREAKPOINT DISTANCE BETWEEN PARTIALLY ORDERED GENOMES","year":2007,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Genome Rearrangement Algorithms","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":"Tsinghua University; McMaster University; University of Minnesota; National Science Foundation","keywords":"Breakpoint; Genome; Nondeterministic algorithm; Chromosome; Computational biology; Heuristic; Biology; Algorithm; Genetics; Computer science; Gene; Artificial intelligence","score_opus":0.010050303129952113,"score_gpt":0.2603520993494437,"score_spread":0.2503017962194916,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1996806342","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":[0.45250252,0.0003502045,0.54304665,0.00030452435,0.000030266398,0.00009619723,0.0011594713,0.0013792095,0.0011309198],"genre_scores_gemma":[0.66788167,0.00014419405,0.3280809,0.00005074776,0.00001104385,0.000077903336,0.002884929,0.00013279828,0.000735753],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9991842,0.00014755157,0.000063823965,0.000315138,0.0001993245,0.00009000124],"domain_scores_gemma":[0.99564844,0.0031122186,0.00040757467,0.0003611187,0.000299072,0.00017149138],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007079671,0.00048708302,0.00087399327,0.0020727946,0.0006946083,0.0011837414,0.0016064721,0.0012209424,0.002218449],"category_scores_gemma":[0.008276398,0.00045763166,0.00070383254,0.0012679271,0.0005143753,0.0019610026,0.0011316463,0.0010034293,0.00031703207],"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":[0.0009780694,0.00020726652,0.017127184,0.0004023442,0.00015424553,0.000403017,0.0003800852,0.7010577,0.021946032,0.026072819,0.0020653636,0.22920586],"study_design_scores_gemma":[0.00005679223,0.00011705996,0.0026675698,0.00001861784,0.000028004291,0.00021210387,0.0001519965,0.94598466,0.009351668,0.04004478,0.0013411258,0.00002558833],"about_ca_topic_score_codex":0.0036694568,"about_ca_topic_score_gemma":0.0032662165,"teacher_disagreement_score":0.0036694568,"about_ca_system_score_codex":0.0011637405,"about_ca_system_score_gemma":0.0010098679,"threshold_uncertainty_score":0.008443534},"labels":[],"label_agreement":null},{"id":"W2007567771","doi":"10.1142/s0219720009004199","title":"GENE LOSS UNDER NEIGHBORHOOD SELECTION FOLLOWING WHOLE GENOME DUPLICATION AND THE RECONSTRUCTION OF THE ANCESTRAL POPULUS GENOME","year":2009,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","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":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Ottawa","funders":"","keywords":"Genome; Gene duplication; Biology; Populus trichocarpa; Genetics; Gene; Chromosome; Genome evolution; Segmental duplication; Phylogenomics; Gene family; Computational biology; Phylogenetic tree","score_opus":0.006985693473040731,"score_gpt":0.22958404739665136,"score_spread":0.22259835392361063,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2007567771","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":[0.96580094,0.000050007602,0.033735212,0.000026622334,0.0000016900082,0.000008676148,0.000029655494,0.00008321318,0.00026393944],"genre_scores_gemma":[0.9811624,0.000038720525,0.018413689,0.000012905371,0.000002958615,0.000014519561,0.0001298248,0.000031668027,0.0001933386],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9995889,0.0001582648,0.000017870232,0.000095554766,0.00009970362,0.00003965036],"domain_scores_gemma":[0.9977564,0.0014677856,0.0002751779,0.0002054672,0.00018289224,0.00011214608],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015174529,0.00014136013,0.0004347749,0.00064597913,0.0003076548,0.0004954037,0.0005443305,0.00022217432,0.00036959597],"category_scores_gemma":[0.005760282,0.00014493603,0.00030989174,0.00041170735,0.00045389688,0.00035885,0.00047973215,0.0002904653,0.000058660757],"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":[0.0016758287,0.00012874385,0.2596561,0.00018630586,0.0002875244,0.00074951485,0.00063408475,0.45669332,0.1355143,0.012160964,0.00036668326,0.1319466],"study_design_scores_gemma":[0.00003685196,0.0002175109,0.05858998,0.000012712302,0.000085486674,0.00043811416,0.00016063306,0.8981016,0.036051355,0.0056707705,0.0006084597,0.00002650055],"about_ca_topic_score_codex":0.0015689027,"about_ca_topic_score_gemma":0.0027799031,"teacher_disagreement_score":0.0015689027,"about_ca_system_score_codex":0.00040437584,"about_ca_system_score_gemma":0.00044104393,"threshold_uncertainty_score":0.008025169},"labels":[],"label_agreement":null},{"id":"W2014066195","doi":"10.1142/s021972000800345x","title":"SEMIPARAMETRIC CLUSTERING METHOD FOR MICROARRAY DATA ANALYSIS","year":2008,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","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":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"National Center for Research Resources","keywords":"Cluster analysis; Nonparametric statistics; Parametric statistics; Robustness (evolution); Mixture model; Computer science; Data mining; Bayesian probability; Bayesian information criterion; Mathematics; Artificial intelligence; Statistics","score_opus":0.04351846861240303,"score_gpt":0.3385774862596929,"score_spread":0.29505901764728987,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2014066195","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0004603502,0.00017411767,0.9985845,0.00007756879,0.000028368275,0.000032177795,0.00007635353,0.00027900768,0.0002876456],"genre_scores_gemma":[0.04739763,0.0007217666,0.9473137,0.00025011378,0.0001878631,0.0008762455,0.00091358525,0.00026709415,0.002071911],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9956013,0.0020365927,0.00023088603,0.00080629735,0.0011853798,0.00013959664],"domain_scores_gemma":[0.9957196,0.0023254442,0.00040403416,0.0006736738,0.00078091846,0.00009634759],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0040694186,0.0011121379,0.0021270874,0.0023653202,0.00092917733,0.0014219804,0.0027711461,0.0016541035,0.003776224],"category_scores_gemma":[0.010998182,0.00064533553,0.0020672677,0.0032613454,0.0011007494,0.0013558808,0.0021370822,0.0029876104,0.0026135424],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00024011331,0.00013750533,0.0021135646,0.0009853737,0.0006206361,0.00029476619,0.0005164572,0.31143472,0.015611202,0.20534927,0.022286667,0.4404098],"study_design_scores_gemma":[0.000023314544,0.000043550823,0.001108517,0.000052685296,0.00005935652,0.00029389674,0.000052614247,0.8921099,0.003310344,0.08342851,0.019419,0.00009837635],"about_ca_topic_score_codex":0.0018217127,"about_ca_topic_score_gemma":0.0016351513,"teacher_disagreement_score":0.0040694186,"about_ca_system_score_codex":0.001344377,"about_ca_system_score_gemma":0.0018363546,"threshold_uncertainty_score":0.02152139},"labels":[],"label_agreement":null},{"id":"W2015820505","doi":"10.1142/s0219720006001898","title":"SELECTIVE TARGETING OF INDEL-INFERRED DIFFERENCES IN SPATIAL STRUCTURES OF HOMOLOGOUS PROTEINS","year":2006,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Research on Leishmaniasis Studies","field":"Medicine","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":"Canada's Michael Smith Genome Sciences Centre; Michael Smith Health Research BC; University of British Columbia","funders":"Michael Smith Health Research BC","keywords":"Biology; Virulence; In silico; Peptide sequence; Genetics; Human pathogen; Computational biology; Sequence alignment; Protein sequencing; Protein superfamily; Gene","score_opus":0.01524377962744824,"score_gpt":0.2806208448044124,"score_spread":0.2653770651769642,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2015820505","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":[0.98799974,0.00086718344,0.009735173,0.000065627675,0.000014333862,0.000043770528,0.000120418146,0.00015015365,0.0010035291],"genre_scores_gemma":[0.9898359,0.00044608692,0.00894269,0.000053334217,0.000006518583,0.000020615165,0.00022268629,0.000023859455,0.00044836212],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9998683,0.000032442764,0.000007883209,0.000030929128,0.000041835377,0.000018585131],"domain_scores_gemma":[0.99967515,0.00016376267,0.00008574992,0.000032180764,0.000018775812,0.000024329545],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00022467734,0.00021128338,0.00044684453,0.00021703145,0.0001402281,0.00029032578,0.0002961273,0.00024330014,0.0011085179],"category_scores_gemma":[0.0006292438,0.00016471327,0.00019839162,0.00019192533,0.00020895439,0.00019779154,0.00026418295,0.00043353252,0.00028420697],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005455147,0.00009057425,0.0013533345,0.00014877292,0.000025263827,0.00026224606,0.00002892515,0.00324804,0.9813945,0.00034552146,0.00006527677,0.012492014],"study_design_scores_gemma":[0.00007540751,0.0008247502,0.0038385554,0.000010907851,0.00008590153,0.0009233523,0.00004628373,0.020758504,0.9707592,0.00028393266,0.0023762742,0.000016944623],"about_ca_topic_score_codex":0.00019742225,"about_ca_topic_score_gemma":0.00060461083,"teacher_disagreement_score":0.0011085179,"about_ca_system_score_codex":0.00015461085,"about_ca_system_score_gemma":0.00014433816,"threshold_uncertainty_score":0.0037084222},"labels":[],"label_agreement":null},{"id":"W2019662118","doi":"10.1142/s0219720010005294","title":"ADVANCES IN GENOME INFORMATICS","year":2010,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"Global Institute for Water Security, University of Saskatchewan","keywords":"Chinese academy of sciences; Library science; China; Informatics; Computer science; Geography; Engineering","score_opus":0.0044127069985803865,"score_gpt":0.23686196601772613,"score_spread":0.23244925901914576,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2019662118","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":[0.009817261,0.35096824,0.3698624,0.07672998,0.015743334,0.00018319722,0.002050138,0.005431041,0.16921431],"genre_scores_gemma":[0.10152167,0.39859116,0.39989772,0.015291834,0.018295996,0.00020758765,0.0060494733,0.0011028809,0.059041765],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99881506,0.0003581266,0.00007628635,0.00022753209,0.00043759987,0.00008536474],"domain_scores_gemma":[0.99552023,0.0019258166,0.00012665607,0.0011402804,0.0009562052,0.00033091372],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0044876877,0.0007450013,0.0009646592,0.0036884714,0.0007269827,0.0041769785,0.001696933,0.0013455496,0.019483365],"category_scores_gemma":[0.006763508,0.0005544696,0.00061145826,0.0047297426,0.0019308047,0.008372452,0.0024142738,0.004149151,0.0089932745],"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":[0.00007744912,0.0001462994,0.0008771362,0.00077942753,0.000046225417,0.00007342364,0.000195551,0.00094838045,0.0040454837,0.2150555,0.10274949,0.6750056],"study_design_scores_gemma":[0.00001795073,0.00003225753,0.0011771255,0.00035975905,0.000027008988,0.00024526747,0.000104860424,0.0041428083,0.0022558891,0.117379524,0.8742372,0.000020448595],"about_ca_topic_score_codex":0.0011509248,"about_ca_topic_score_gemma":0.0010738126,"teacher_disagreement_score":0.019483365,"about_ca_system_score_codex":0.0014202327,"about_ca_system_score_gemma":0.0019749685,"threshold_uncertainty_score":0.065178335},"labels":[],"label_agreement":null},{"id":"W2021405914","doi":"10.1142/s0219720012500011","title":"A METHOD FOR DISCOVERING COMMON PATTERNS FROM TWO RNA SECONDARY STRUCTURES AND ITS APPLICATION TO STRUCTURAL REPEAT DETECTION","year":2011,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","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":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"National Institutes of Health; Cancer Research Institute","keywords":"Computer science; Nucleic acid secondary structure; Computational biology; RNA; Protein secondary structure; Java; Source code; Tree (set theory); Theoretical computer science; Artificial intelligence; Data mining; Biology; Gene; Genetics; Mathematics; Programming language; Combinatorics","score_opus":0.012618896130908556,"score_gpt":0.27712885254702485,"score_spread":0.2645099564161163,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2021405914","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0054060854,0.0000823631,0.9919911,0.00005466739,0.000013401701,0.000050826173,0.00015115527,0.001966049,0.00028439114],"genre_scores_gemma":[0.029167037,0.00007068873,0.9693607,0.000036850328,0.000016105734,0.000108741035,0.00039989993,0.00022567989,0.00061416626],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99844414,0.00021071345,0.00010270128,0.0005607222,0.0005968856,0.000084878375],"domain_scores_gemma":[0.9981913,0.0009260346,0.00022054704,0.00025310137,0.00029278986,0.00011622233],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011472945,0.0010322547,0.0010678973,0.0042283274,0.0011530514,0.0011937114,0.0016815356,0.0011648024,0.0030978646],"category_scores_gemma":[0.003545223,0.0009954628,0.0018597523,0.0024952793,0.0011753953,0.0016872253,0.0012433869,0.0012858944,0.0014117971],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00025720708,0.000325228,0.0073234295,0.0006295078,0.00029508493,0.0009166651,0.0004909476,0.02910367,0.12160991,0.021416672,0.0062200003,0.81141174],"study_design_scores_gemma":[0.00016361258,0.0003026403,0.0046738707,0.000070107795,0.00014081708,0.004681057,0.00016869516,0.8484955,0.06841547,0.051350784,0.021326639,0.00021085105],"about_ca_topic_score_codex":0.0015533336,"about_ca_topic_score_gemma":0.0017655711,"teacher_disagreement_score":0.0042283274,"about_ca_system_score_codex":0.0004400581,"about_ca_system_score_gemma":0.001315577,"threshold_uncertainty_score":0.0103634},"labels":[],"label_agreement":null},{"id":"W2023250395","doi":"10.1142/s0219720010005099","title":"ADEPTS: ADVANCED PEPTIDE <i>DE NOVO</i> SEQUENCING WITH A PAIR OF TANDEM MASS SPECTRA","year":2010,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Advanced Proteomics Techniques and Applications","field":"Chemistry","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 Waterloo","funders":"University of Waterloo; Natural Sciences and Engineering Research Council of Canada; Mitacs","keywords":"Computational biology; DNA sequencing; Tandem mass spectrometry; Peptide; Proteomics; Biology; Genetics; Chemistry; Mass spectrometry; Biochemistry; Gene; Chromatography","score_opus":0.007141073480087647,"score_gpt":0.24843712025055886,"score_spread":0.24129604677047123,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2023250395","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":[0.013981186,0.00093828235,0.9525881,0.00041904696,0.00049789954,0.00039000102,0.0016165548,0.021770427,0.007798533],"genre_scores_gemma":[0.047957823,0.0004526508,0.9359393,0.00038610696,0.00021955726,0.0002700276,0.002816731,0.0012850836,0.010672606],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9986532,0.00028775155,0.000096291995,0.00040526857,0.00047839666,0.00007914519],"domain_scores_gemma":[0.9987562,0.00027698805,0.00017384559,0.00036973273,0.0002576436,0.0001656606],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0018787053,0.0013926385,0.00091116497,0.0015690087,0.00074429566,0.0017171417,0.0018816934,0.0010090609,0.01327172],"category_scores_gemma":[0.0024684325,0.00077670696,0.00077736226,0.0011638486,0.0005157632,0.001754607,0.0019086284,0.0017136004,0.007923314],"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":[0.0015758387,0.00039997057,0.0050366423,0.00091144507,0.00027186857,0.0010932928,0.00018169671,0.009401026,0.12867463,0.01905208,0.056323998,0.7770774],"study_design_scores_gemma":[0.00083390233,0.0007215006,0.0044406364,0.00014220581,0.00023964564,0.00614219,0.00011475374,0.40358686,0.28970587,0.054556906,0.23925297,0.00026257648],"about_ca_topic_score_codex":0.0005378469,"about_ca_topic_score_gemma":0.00092362973,"teacher_disagreement_score":0.01327172,"about_ca_system_score_codex":0.00047562734,"about_ca_system_score_gemma":0.0011576099,"threshold_uncertainty_score":0.044398308},"labels":[],"label_agreement":null},{"id":"W2030659385","doi":"10.1142/s0219720006001746","title":"PRIMA: PEPTIDE ROBUST IDENTIFICATION FROM MS/MS SPECTRA","year":2006,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Advanced Proteomics Techniques and Applications","field":"Chemistry","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 Waterloo; Western University; McGill University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Tandem mass spectrometry; Mascot; Computer science; Identification (biology); Software; Construct (python library); Ion trap; Proteomics; Peptide; Tandem; Artificial intelligence; Mass spectrometry; Data mining; Chemistry; Chromatography; Biology; Engineering","score_opus":0.009623334375525457,"score_gpt":0.2478373714365536,"score_spread":0.23821403706102814,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2030659385","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.009308604,0.0004057371,0.96330875,0.00011860565,0.000065822875,0.00013305797,0.00057798956,0.025417643,0.0006637897],"genre_scores_gemma":[0.07289635,0.00039409005,0.91924006,0.00017251699,0.00007473224,0.0002642375,0.002332216,0.0019363366,0.0026894014],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99830866,0.00038913634,0.00013367203,0.0003938383,0.00068883545,0.00008590656],"domain_scores_gemma":[0.9971691,0.00087833457,0.0004365464,0.00084952515,0.0005380921,0.00012846611],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0035470442,0.0017396758,0.001415384,0.0017756046,0.00080247613,0.0022861592,0.0026383232,0.00094499806,0.0045164037],"category_scores_gemma":[0.008112245,0.00090787275,0.0013981722,0.0014756416,0.0007575037,0.0020328735,0.003123647,0.0019237659,0.00658682],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000879317,0.0002543083,0.005102648,0.000950413,0.00039943994,0.00065748766,0.00015108303,0.039652135,0.10298762,0.009404259,0.021085609,0.81847566],"study_design_scores_gemma":[0.00006066117,0.00030536423,0.0023111736,0.000050194336,0.00007739955,0.0014272954,0.000043616335,0.8394868,0.11965603,0.016631128,0.019840242,0.0001101535],"about_ca_topic_score_codex":0.0005921571,"about_ca_topic_score_gemma":0.00071460597,"teacher_disagreement_score":0.0045164037,"about_ca_system_score_codex":0.0003456845,"about_ca_system_score_gemma":0.0011645843,"threshold_uncertainty_score":0.018758774},"labels":[],"label_agreement":null},{"id":"W2032713750","doi":"10.1142/s0219720007002941","title":"EVALUATION OF NORMALIZATION AND PRE-CLUSTERING ISSUES IN A NOVEL CLUSTERING APPROACH: GLOBAL OPTIMUM SEARCH WITH ENHANCED POSITIONING","year":2007,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Gene expression and cancer classification","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":"McGill University; Foundation for the National Institutes of Health; National Science Foundation","keywords":"Cluster analysis; Normalization (sociology); Data mining; Computer science; Correlation clustering; CURE data clustering algorithm; Outlier; Fuzzy clustering; Clustering high-dimensional data; Consensus clustering; Artificial intelligence","score_opus":0.022538950893690676,"score_gpt":0.3214746440522242,"score_spread":0.2989356931585335,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2032713750","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.064608395,0.00027961616,0.93316984,0.00018782426,0.000044280667,0.00007904858,0.000047704663,0.0007734621,0.0008098133],"genre_scores_gemma":[0.18438992,0.00016857518,0.813991,0.0000680469,0.00002160513,0.00012443606,0.00019792233,0.0003167148,0.00072175486],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99836963,0.00061241066,0.000096570024,0.00028401165,0.0005556733,0.00008177226],"domain_scores_gemma":[0.9952362,0.0026155182,0.0004194354,0.0005834663,0.0010459833,0.000099474164],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0047600903,0.0011834336,0.001244148,0.0009511744,0.0006811916,0.0011883598,0.0012250772,0.00150034,0.001321566],"category_scores_gemma":[0.012696853,0.00044730006,0.0009108198,0.0014474688,0.00080137077,0.001679494,0.0010324764,0.0011825479,0.00059775484],"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":[0.00075566856,0.00026290453,0.0026434273,0.00035212067,0.00020083445,0.00011676124,0.00026520126,0.71561414,0.04611878,0.0063946,0.0010847829,0.2261907],"study_design_scores_gemma":[0.000043354175,0.00030630676,0.0010951678,0.000012454938,0.0000389099,0.00007158729,0.00004137208,0.97128266,0.02486537,0.0013754022,0.0008348751,0.00003250015],"about_ca_topic_score_codex":0.0022836917,"about_ca_topic_score_gemma":0.003694613,"teacher_disagreement_score":0.0047600903,"about_ca_system_score_codex":0.0009977939,"about_ca_system_score_gemma":0.0011446691,"threshold_uncertainty_score":0.025174022},"labels":[],"label_agreement":null},{"id":"W2036814280","doi":"10.1142/s0219720008003813","title":"UBIQUITOUS REASSORTMENTS IN INFLUENZA A VIRUSES","year":2008,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Influenza Virus Research Studies","field":"Medicine","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 Alberta","funders":"","keywords":"Reassortment; Virology; Biology; Neuraminidase; Nucleoprotein; Hemagglutinin (influenza); Influenza A virus subtype H5N1; RNA; H5N1 genetic structure; Virus; Gene; Genetics; Coronavirus disease 2019 (COVID-19)","score_opus":0.106559983716119,"score_gpt":0.3925495000729928,"score_spread":0.28598951635687375,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2036814280","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":[0.7730806,0.00144364,0.22057895,0.00004263521,0.000046983696,0.00012739426,0.00066277984,0.0017007348,0.00231633],"genre_scores_gemma":[0.83003986,0.0005164503,0.16598046,0.000034644952,0.00002341949,0.000060087837,0.002162267,0.00019040066,0.0009923168],"study_design_codex":"bench_or_experimental","study_design_gemma":"observational","domain_scores_codex":[0.9992663,0.00010679621,0.00009714228,0.00030658982,0.00016821314,0.000054973767],"domain_scores_gemma":[0.9991605,0.00030458707,0.0002542201,0.00009883413,0.00011565155,0.00006619034],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006670945,0.00041140153,0.00034680447,0.0018352973,0.00077207194,0.0007475092,0.0004355739,0.00035512177,0.0014133098],"category_scores_gemma":[0.0016803353,0.00027439737,0.0006613021,0.0016186174,0.00027511513,0.00062313676,0.0007149401,0.00028176582,0.00047810783],"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":[0.0011536336,0.00026302054,0.26659808,0.0008733837,0.00049053814,0.0014440643,0.0019864808,0.025478274,0.4437774,0.0058191493,0.0018256451,0.2502903],"study_design_scores_gemma":[0.0001021089,0.0009759048,0.30806157,0.00021694352,0.00066987163,0.011126356,0.001698956,0.44372725,0.19333708,0.016054511,0.023824973,0.00020447669],"about_ca_topic_score_codex":0.0012624689,"about_ca_topic_score_gemma":0.0018001855,"teacher_disagreement_score":0.0018352973,"about_ca_system_score_codex":0.0002886037,"about_ca_system_score_gemma":0.00019793077,"threshold_uncertainty_score":0.0047280192},"labels":[],"label_agreement":null},{"id":"W2037625634","doi":"10.1142/s0219720006002399","title":"PROTEOMIC BIOMARKER IDENTIFICATION FOR DIAGNOSIS OF EARLY RELAPSE IN OVARIAN CANCER","year":2006,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Advanced Proteomics Techniques and Applications","field":"Chemistry","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":"Women's Health Research Institute","funders":"National Science Foundation","keywords":"Markov blanket; Feature selection; Support vector machine; Artificial intelligence; Computer science; Feature (linguistics); Pattern recognition (psychology); Feature vector; Biomarker discovery; Biomarker; Curse of dimensionality; Machine learning; Identification (biology); Markov chain; Proteomics; Markov model; Biology","score_opus":0.013561879053840334,"score_gpt":0.29218493878624746,"score_spread":0.27862305973240714,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2037625634","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":[0.94960034,0.004049665,0.044579282,0.00032399708,0.000037680453,0.00010183244,0.00031503392,0.00040877584,0.00058327185],"genre_scores_gemma":[0.9525061,0.00074408355,0.045861755,0.000049955335,0.000021270698,0.000044690674,0.00034868007,0.000010700139,0.00041277584],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9997416,0.00009611238,0.000024338704,0.000031364798,0.000077287485,0.000029228762],"domain_scores_gemma":[0.9995782,0.00018706749,0.00007701558,0.000025918363,0.000105094594,0.000026778209],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007544298,0.00028405094,0.00060304714,0.0011580454,0.00017599661,0.00038060735,0.00025558597,0.00030530497,0.0003918038],"category_scores_gemma":[0.0021237917,0.00013476255,0.00027130125,0.00063041766,0.0001362391,0.00024461336,0.00018419421,0.00028632075,0.00022722333],"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":[0.0018632481,0.00051075075,0.18722887,0.00039006033,0.00015489338,0.00051930035,0.0001329478,0.008167645,0.25179812,0.0005078148,0.0020450305,0.54668134],"study_design_scores_gemma":[0.00023525926,0.0027046006,0.35690337,0.00007676687,0.00040279102,0.0044171414,0.00026616312,0.26689416,0.3575,0.0030387174,0.007475641,0.00008542086],"about_ca_topic_score_codex":0.00043554045,"about_ca_topic_score_gemma":0.0006328796,"teacher_disagreement_score":0.0011580454,"about_ca_system_score_codex":0.00021151034,"about_ca_system_score_gemma":0.0003789208,"threshold_uncertainty_score":0.0039898753},"labels":[],"label_agreement":null},{"id":"W2041368329","doi":"10.1142/s0219720005001570","title":"HEURISTIC SEARCH IN CONSTRAINED BIPARTITE MATCHING WITH APPLICATIONS TO PROTEIN NMR BACKBONE RESONANCE ASSIGNMENT","year":2005,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Machine Learning and Algorithms","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":"University of Alberta","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Bipartite graph; Combinatorics; Matching (statistics); Heuristics; Assignment problem; Complete bipartite graph; Mathematics; Heuristic; Combinatorial optimization; Local search (optimization); Algorithm; Computer science; Graph; Mathematical optimization","score_opus":0.010647841950632267,"score_gpt":0.2710084611486648,"score_spread":0.2603606191980325,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2041368329","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":[0.0196547,0.001231298,0.9723808,0.00044431473,0.000056842135,0.00011336051,0.00007916564,0.00041776703,0.0056217103],"genre_scores_gemma":[0.35335153,0.0010072095,0.6416661,0.00033212517,0.00008660967,0.00035822106,0.00021568323,0.00013244236,0.002850049],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99892956,0.000719833,0.000034103094,0.00011179276,0.00011777237,0.00008694406],"domain_scores_gemma":[0.99748707,0.0020917477,0.00014259017,0.00010200145,0.000112482296,0.000064109554],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00175519,0.00070983986,0.0012452246,0.0010800215,0.0006899662,0.0008888291,0.0011347434,0.0012685227,0.0022482825],"category_scores_gemma":[0.0054283924,0.00047958968,0.000548934,0.0023620473,0.00096836797,0.0009075172,0.0009726773,0.000907174,0.00033097208],"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":[0.00010528617,0.00011059358,0.00029514384,0.00012337463,0.000039736493,0.00007287383,0.000047259822,0.8916191,0.0007050041,0.046606787,0.0022982068,0.057976663],"study_design_scores_gemma":[0.000036463673,0.000034465444,0.000053562362,0.000009408787,0.000008125143,0.000029099225,0.000016185395,0.9659348,0.00031353103,0.03244242,0.0011158399,0.000006121218],"about_ca_topic_score_codex":0.004888972,"about_ca_topic_score_gemma":0.0040703793,"teacher_disagreement_score":0.004888972,"about_ca_system_score_codex":0.0010546292,"about_ca_system_score_gemma":0.0014790229,"threshold_uncertainty_score":0.009721041},"labels":[],"label_agreement":null},{"id":"W2041576045","doi":"10.1142/s0219720014500243","title":"A new efficient algorithm for inferring explicit hybridization networks following the Neighbor-Joining principle","year":2014,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Genetic diversity and population structure","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":"Université du Québec à Montréal","funders":"","keywords":"Phylogenetic network; Phylogenetic tree; Tree (set theory); Algorithm; Heuristic; Metric (unit); Computer science; Tree rearrangement; Distance matrix; Mathematics; Artificial intelligence; Biology; Combinatorics","score_opus":0.008442928588421848,"score_gpt":0.24694638730283322,"score_spread":0.23850345871441136,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2041576045","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0013425257,0.00006817843,0.9975993,0.000034528453,0.000020782958,0.000048868584,0.00005015181,0.00057163497,0.0002640793],"genre_scores_gemma":[0.010402667,0.000050571183,0.9884179,0.000035517194,0.000018021517,0.00016151895,0.00026727165,0.00010603557,0.00054058473],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99840564,0.00034203802,0.00010980177,0.00053557096,0.0005090847,0.00009785606],"domain_scores_gemma":[0.99820185,0.0009062855,0.00015952601,0.00021444596,0.00043724157,0.00008069518],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0017675061,0.0015373256,0.0016922614,0.002969259,0.0014396495,0.0016392524,0.0030694257,0.0017893814,0.0049481476],"category_scores_gemma":[0.0062402654,0.0010853657,0.0013983313,0.0030540163,0.0008079766,0.0028069853,0.0020381943,0.0023923614,0.0021476408],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001484019,0.00015538196,0.0015129574,0.00033129664,0.00023626086,0.00016519027,0.00034523397,0.21683188,0.01478164,0.023570394,0.0069135753,0.7350079],"study_design_scores_gemma":[0.00006882448,0.0000624477,0.00032798972,0.000023132376,0.000041555366,0.00022951516,0.000073587,0.9630104,0.0033679276,0.025217019,0.0075424155,0.00003520676],"about_ca_topic_score_codex":0.0050587333,"about_ca_topic_score_gemma":0.0057543535,"teacher_disagreement_score":0.0050587333,"about_ca_system_score_codex":0.0009982849,"about_ca_system_score_gemma":0.002556367,"threshold_uncertainty_score":0.016553164},"labels":[],"label_agreement":null},{"id":"W2048065755","doi":"10.1142/s0219720009004242","title":"A TUTORIAL OF TECHNIQUES FOR IMPROVING STANDARD HIDDEN MARKOV MODEL ALGORITHMS","year":2009,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Algorithms and Data Compression","field":"Computer Science","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":"University of Waterloo","funders":"","keywords":"Viterbi algorithm; Computer science; Hidden Markov model; Logarithm; Forward algorithm; Heuristics; Algorithm; Factor (programming language); Heuristic; Markov model; Sequence (biology); Markov chain; Space (punctuation); Parallel computing; Artificial intelligence; Mathematics; Variable-order Markov model; Machine learning; Programming language","score_opus":0.011313900673213523,"score_gpt":0.27365696638997694,"score_spread":0.2623430657167634,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2048065755","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00039756062,0.009653324,0.9788801,0.00046530037,0.00083039113,0.00007890682,0.0004092473,0.0033166802,0.00596855],"genre_scores_gemma":[0.0073784054,0.01702475,0.96089315,0.0008225787,0.0013998789,0.00039430062,0.0017514889,0.001933495,0.008401826],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.996608,0.0008185963,0.00034931253,0.0005089202,0.00155028,0.00016488942],"domain_scores_gemma":[0.99589586,0.0026727438,0.00014586223,0.0005326735,0.00068624294,0.00006664712],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0028971375,0.0033410725,0.0015713926,0.003247167,0.0007583209,0.0021655476,0.0033441957,0.002163655,0.030628715],"category_scores_gemma":[0.016708408,0.0020000364,0.0028574318,0.0048225373,0.0009956729,0.0057285675,0.0023084318,0.0052074455,0.02355619],"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":[0.00011233123,0.00020126585,0.00045775468,0.0020886604,0.0002158205,0.00025792397,0.00018646065,0.028897043,0.0050145257,0.11579044,0.09811266,0.7486651],"study_design_scores_gemma":[0.00008179383,0.00015297977,0.00069707865,0.00064595765,0.000121885,0.0010460608,0.000068476744,0.18406634,0.007849984,0.38404578,0.42106676,0.00015707732],"about_ca_topic_score_codex":0.0014896081,"about_ca_topic_score_gemma":0.0015828821,"teacher_disagreement_score":0.030628715,"about_ca_system_score_codex":0.0011526863,"about_ca_system_score_gemma":0.0012553141,"threshold_uncertainty_score":0.102463245},"labels":[],"label_agreement":null},{"id":"W2051186751","doi":"10.1142/s0219720003000083","title":"Computational Assignment of Protein Backbone NMR Peaks by Efficient Bounding and Filtering","year":2003,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","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 Alberta","funders":"","keywords":"Bounding overwatch; Computer science; Algorithm; Assignment problem; Bipartite graph; Computation; Greedy algorithm; Weapon target assignment problem; Process (computing); Matching (statistics); Computational complexity theory; Approximation algorithm; Generalized assignment problem; Mathematical optimization; Theoretical computer science; Mathematics; Optimization problem; Artificial intelligence; Graph","score_opus":0.004713577335423098,"score_gpt":0.2258257784558294,"score_spread":0.2211122011204063,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2051186751","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.011876778,0.00006303956,0.9865954,0.000087905726,0.000008926095,0.000037723355,0.000029017769,0.00073802064,0.00056316267],"genre_scores_gemma":[0.1054394,0.00010121499,0.8929118,0.00006860138,0.000022255279,0.00020203179,0.0002866794,0.00015911905,0.00080891093],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.998497,0.0005398005,0.00008821978,0.00025756104,0.00043224604,0.00018523229],"domain_scores_gemma":[0.99581015,0.0030595888,0.00027608784,0.0004893685,0.00025713953,0.000107610416],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0027248452,0.0012509854,0.0018605925,0.0012994969,0.001062898,0.0014764561,0.0021650933,0.0016629029,0.0024427336],"category_scores_gemma":[0.008234158,0.0008761169,0.00106835,0.0019537292,0.0011677287,0.002696234,0.0016797155,0.0013809182,0.0007608917],"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":[0.00028471125,0.0002282859,0.0010937145,0.00012746439,0.000050624567,0.00006237633,0.00010697015,0.7171891,0.009070187,0.020272456,0.002379372,0.24913467],"study_design_scores_gemma":[0.000020260608,0.000016044545,0.00009446895,0.0000030577714,0.0000067206115,0.000012579702,0.000008308731,0.9922259,0.0016520395,0.0055884332,0.0003665239,0.0000057054935],"about_ca_topic_score_codex":0.006418401,"about_ca_topic_score_gemma":0.005806558,"teacher_disagreement_score":0.006418401,"about_ca_system_score_codex":0.0015072483,"about_ca_system_score_gemma":0.002961543,"threshold_uncertainty_score":0.014410496},"labels":[],"label_agreement":null},{"id":"W2053279888","doi":"10.1142/s0219720004000843","title":"HISTOGRAM-BASED SCORING SCHEMES FOR PROTEIN NMR RESONANCE ASSIGNMENT","year":2004,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","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 Alberta","funders":"University of Alberta","keywords":"Histogram; Chemical shift; Spin (aerodynamics); Adjacency list; Biological system; Computer science; Key (lock); Signature (topology); Pattern recognition (psychology); Algorithm; Data mining; Chemistry; Artificial intelligence; Mathematics; Physics; Nuclear magnetic resonance; Biology; Image (mathematics)","score_opus":0.00786710609598433,"score_gpt":0.2470661261058875,"score_spread":0.23919902000990317,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2053279888","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.012629411,0.00019381079,0.98548245,0.00009663752,0.000053035827,0.0001382295,0.00011019587,0.00074283365,0.00055344764],"genre_scores_gemma":[0.30516154,0.00028246053,0.69079554,0.00010823448,0.0001098381,0.000335483,0.0006636849,0.00015750816,0.0023857593],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9955781,0.0018581894,0.00029601846,0.00030709623,0.0017433597,0.00021731654],"domain_scores_gemma":[0.9872966,0.006641426,0.000688412,0.0017967328,0.0031557214,0.00042112573],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0074390643,0.0008509058,0.0009562169,0.002521453,0.0009871841,0.00126767,0.0026339402,0.0011859292,0.003110908],"category_scores_gemma":[0.02508321,0.000412878,0.00046596065,0.002558941,0.0010560644,0.002966543,0.0019914042,0.0014459341,0.0014394452],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005388346,0.00031150755,0.0063650073,0.00020960529,0.000113509,0.000118367505,0.0002959892,0.2504551,0.008397272,0.08699167,0.00656239,0.63964075],"study_design_scores_gemma":[0.00005436249,0.00009556157,0.0012348301,0.000015886819,0.000016018721,0.00008468466,0.00004169993,0.9533966,0.004017581,0.038778927,0.002180687,0.00008324113],"about_ca_topic_score_codex":0.0033727025,"about_ca_topic_score_gemma":0.00348642,"teacher_disagreement_score":0.0074390643,"about_ca_system_score_codex":0.0011147212,"about_ca_system_score_gemma":0.0014724861,"threshold_uncertainty_score":0.039341986},"labels":[],"label_agreement":null},{"id":"W2057771742","doi":"10.1142/s0219720006002314","title":"OPTIMALLY-CONNECTED HIDDEN MARKOV MODELS FOR PREDICTING MHC-BINDING PEPTIDES","year":2006,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"vaccines and immunoinformatics approaches","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 Saskatchewan","funders":"Max-Planck-Gesellschaft; Genome Canada","keywords":"Hidden Markov model; Computation; Computer science; Human leukocyte antigen; Pattern recognition (psychology); Artificial intelligence; Merge (version control); Heuristic; Markov chain; Computational biology; Algorithm; Machine learning; Biology; Genetics; Antigen","score_opus":0.011809212372534465,"score_gpt":0.23368368039645088,"score_spread":0.22187446802391642,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2057771742","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0797281,0.0004880742,0.9163074,0.0002667976,0.00004491942,0.000068218775,0.0004948495,0.0012224923,0.0013791871],"genre_scores_gemma":[0.7063725,0.00058042275,0.28930715,0.000111697576,0.00005469391,0.00024335335,0.0013692263,0.00014953647,0.001811421],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99967074,0.00016605396,0.000016932236,0.000066268556,0.000052892476,0.00002714603],"domain_scores_gemma":[0.9988801,0.00087242544,0.00007603884,0.00008290266,0.000056840854,0.00003171787],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006706627,0.00057773537,0.00056412735,0.0008066783,0.00042178313,0.000513127,0.0008972124,0.00079577055,0.001575443],"category_scores_gemma":[0.00305333,0.00060618145,0.00064232026,0.0007263432,0.00036375804,0.0011332046,0.0004899255,0.0008503809,0.0004994051],"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":[0.00009580923,0.000029129847,0.0012044797,0.00003147703,0.000033479682,0.00005365561,0.00004062269,0.9666167,0.0011183957,0.005095793,0.00041836698,0.025262136],"study_design_scores_gemma":[0.0000038271273,0.0000065532254,0.00009851409,0.0000025695858,0.0000047186395,0.000006485085,0.0000035664418,0.99515414,0.00021672739,0.0043615173,0.00013793331,0.0000035009946],"about_ca_topic_score_codex":0.005530004,"about_ca_topic_score_gemma":0.008059452,"teacher_disagreement_score":0.005530004,"about_ca_system_score_codex":0.00061423244,"about_ca_system_score_gemma":0.0007648213,"threshold_uncertainty_score":0.010995626},"labels":[],"label_agreement":null},{"id":"W2060355037","doi":"10.1142/s0219720006002363","title":"RNALL: AN EFFICIENT ALGORITHM FOR PREDICTING RNA LOCAL SECONDARY STRUCTURAL LANDSCAPE IN GENOMES","year":2006,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","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":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Natural Sciences and Engineering Research Council of Canada; U.S. Department of Energy","keywords":"RNA; Computational biology; Nucleic acid secondary structure; Energy landscape; Computer science; Genome; Algorithm; Sliding window protocol; Bioinformatics; Data mining; Gene; Window (computing); Biology; Genetics","score_opus":0.005349491978337566,"score_gpt":0.23176467682351212,"score_spread":0.22641518484517456,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2060355037","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.026280412,0.0005324812,0.9215327,0.0002807195,0.000057235735,0.00023412888,0.002475376,0.04699252,0.0016145259],"genre_scores_gemma":[0.09732723,0.0002244847,0.89057696,0.00024306889,0.000038347527,0.0008424622,0.006172794,0.0030163962,0.0015583129],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9993494,0.00016996746,0.00004305885,0.00022806301,0.00013828026,0.000071280905],"domain_scores_gemma":[0.99885845,0.0007626785,0.00010817509,0.00009552683,0.00011017135,0.000065128115],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011656448,0.0021911678,0.0016977974,0.0033375223,0.0010524733,0.0014952787,0.0030425468,0.0017639432,0.008483919],"category_scores_gemma":[0.004347046,0.0011557714,0.0016661135,0.0019753708,0.0006988794,0.0026075467,0.001891086,0.0015185905,0.0032630512],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008879495,0.00037649993,0.008641737,0.0012443879,0.0003911677,0.00072618364,0.00044057786,0.3407298,0.02420685,0.011627646,0.039835118,0.5708921],"study_design_scores_gemma":[0.00011881805,0.000054721153,0.00049132365,0.000030518313,0.00002970029,0.00012533012,0.00004781529,0.9815461,0.0031884066,0.009284626,0.0050561856,0.000026490809],"about_ca_topic_score_codex":0.0035537602,"about_ca_topic_score_gemma":0.00563875,"teacher_disagreement_score":0.008483919,"about_ca_system_score_codex":0.0012085612,"about_ca_system_score_gemma":0.001408187,"threshold_uncertainty_score":0.028381526},"labels":[],"label_agreement":null},{"id":"W2066123704","doi":"10.1142/s0219720006002119","title":"IMPROVING THE SENSITIVITY AND SPECIFICITY OF PROTEIN HOMOLOGY SEARCH BY INCORPORATING PREDICTED SECONDARY STRUCTURES","year":2006,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","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":"Western University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Homology (biology); Protein secondary structure; Protein structure prediction; Homology modeling; Computer science; Computational biology; Sensitivity (control systems); Protein structure; Machine learning; Artificial intelligence; Biology; Genetics; Amino acid; Engineering; Biochemistry","score_opus":0.005744006339003689,"score_gpt":0.21144460315384553,"score_spread":0.20570059681484185,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2066123704","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.3555527,0.003945131,0.6291074,0.0005838588,0.00016729096,0.00015701298,0.00022591732,0.0064531732,0.0038075426],"genre_scores_gemma":[0.7981943,0.0008674146,0.19848616,0.00031133302,0.000100159035,0.000047217574,0.0005504058,0.0002975813,0.0011453684],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9938061,0.0029978892,0.00046201967,0.0008350548,0.0015770015,0.00032194954],"domain_scores_gemma":[0.963874,0.029384509,0.00097388664,0.0026393915,0.0027445043,0.00038367556],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.011354781,0.0012727161,0.0020568878,0.0022496455,0.0004269624,0.0019765932,0.0013215572,0.0015278341,0.0015414541],"category_scores_gemma":[0.032521605,0.0006961869,0.0009901819,0.0010806785,0.0006557282,0.0026126215,0.001778621,0.0014564104,0.0019891018],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0018186049,0.0005889786,0.06186967,0.0010086715,0.0007133954,0.0007868813,0.00026700515,0.11023908,0.29796582,0.004493786,0.0018924553,0.5183556],"study_design_scores_gemma":[0.00008112301,0.00047026642,0.013928167,0.000055691522,0.00031599862,0.0018759136,0.00005808541,0.8050347,0.17043026,0.0036692787,0.0039509614,0.00012966548],"about_ca_topic_score_codex":0.0007743004,"about_ca_topic_score_gemma":0.00093648717,"teacher_disagreement_score":0.011354781,"about_ca_system_score_codex":0.00041224106,"about_ca_system_score_gemma":0.00069786253,"threshold_uncertainty_score":0.060050488},"labels":[],"label_agreement":null},{"id":"W2066349684","doi":"10.1142/s0219720005001296","title":"MULTIPLE RNA STRUCTURE ALIGNMENT","year":2005,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"RNA and protein synthesis mechanisms","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":"Western University; University Health Network","funders":"","keywords":"RNA; String (physics); Algorithm; Structural alignment; Nucleic acid structure; Computer science; Sequence (biology); Multiple sequence alignment; Mathematics; Sequence alignment; Gene; Genetics; Biology; Peptide sequence","score_opus":0.007012230245182748,"score_gpt":0.23166254141168083,"score_spread":0.22465031116649808,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2066349684","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.009139219,0.000756782,0.9741795,0.00021212873,0.00029512733,0.00016384409,0.00095899653,0.0062786397,0.008015815],"genre_scores_gemma":[0.08427759,0.0005625116,0.89749706,0.00018660113,0.00012897851,0.00022024785,0.0057329116,0.0016121401,0.009781968],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9962463,0.00064451277,0.00028340274,0.0015450666,0.0009976337,0.00028319034],"domain_scores_gemma":[0.9966102,0.0007203306,0.00041148035,0.0011527471,0.0009582144,0.00014705445],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016018235,0.0017342072,0.0024003282,0.0029070405,0.0016212951,0.0022764534,0.0030427666,0.0022126336,0.03086752],"category_scores_gemma":[0.006983605,0.0009215746,0.0016092545,0.003435282,0.0005885067,0.0033010854,0.0029683597,0.002124389,0.015459176],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006305644,0.0002745179,0.0025167072,0.0010008469,0.0003019955,0.000887714,0.00037153592,0.03885897,0.10430071,0.07854202,0.037884116,0.7344303],"study_design_scores_gemma":[0.0001241897,0.00033413398,0.0019331792,0.00012896258,0.00012815317,0.0022849585,0.000392567,0.55651045,0.14801823,0.11401637,0.17601638,0.00011252276],"about_ca_topic_score_codex":0.0004084037,"about_ca_topic_score_gemma":0.0008591835,"teacher_disagreement_score":0.03086752,"about_ca_system_score_codex":0.00080164993,"about_ca_system_score_gemma":0.00110681,"threshold_uncertainty_score":0.10326213},"labels":[],"label_agreement":null},{"id":"W2067201268","doi":"10.1142/s0219720004000788","title":"IDENTIFYING UNIFORMLY MUTATED SEGMENTS WITHIN REPEATS","year":2004,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Algorithms and Data Compression","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":"Pacific Institute for the Mathematical Sciences; Simon Fraser University","funders":"","keywords":"Mathematics; String (physics); Coin flipping; Combinatorics; Set (abstract data type); Mutation; Prior probability; Algorithm; Mutation rate; Shuffling; Discrete mathematics; Statistics; Computer science; Genetics; Biology; Bayesian probability","score_opus":0.017156221096612675,"score_gpt":0.2722669738252626,"score_spread":0.2551107527286499,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2067201268","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":[0.24484745,0.00015640576,0.74912816,0.0001536642,0.000025743098,0.00017309417,0.0004439963,0.0032096247,0.0018618795],"genre_scores_gemma":[0.47429213,0.00008319754,0.5215225,0.000095256444,0.000018078143,0.0001652228,0.0011571308,0.0003338221,0.0023327377],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9984819,0.0003028771,0.00009269864,0.0006037238,0.00034841764,0.00017032078],"domain_scores_gemma":[0.9955094,0.002475505,0.00084870594,0.000497071,0.00048751725,0.00018179632],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013159517,0.0005599692,0.0011054886,0.0017877606,0.00055456103,0.001421001,0.0017466698,0.0018637904,0.0022516993],"category_scores_gemma":[0.010870622,0.0005894596,0.0007587383,0.0010279528,0.0007311748,0.0010509478,0.0012364195,0.0008353481,0.0011722624],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0021356987,0.0002833189,0.05975987,0.00038051928,0.00027434164,0.0018824778,0.00093315437,0.25612676,0.17751276,0.03187162,0.0030798514,0.46575966],"study_design_scores_gemma":[0.000041207113,0.0001017135,0.0041830274,0.000026556183,0.000033185483,0.0004527557,0.000117766904,0.93449265,0.03791112,0.021046756,0.0015521317,0.000041061885],"about_ca_topic_score_codex":0.0020781495,"about_ca_topic_score_gemma":0.0022364522,"teacher_disagreement_score":0.0022516993,"about_ca_system_score_codex":0.0009995907,"about_ca_system_score_gemma":0.0011236425,"threshold_uncertainty_score":0.007532656},"labels":[],"label_agreement":null},{"id":"W2070718329","doi":"10.1142/s0219720004000892","title":"IMPROVING GENE NETWORK INFERENCE BY COMPARING EXPRESSION TIME-SERIES ACROSS SPECIES, DEVELOPMENTAL STAGES OR TISSUES","year":2004,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Gene Regulatory Network Analysis","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 Ottawa; Université de Montréal","funders":"","keywords":"Inference; Gene regulatory network; Computer science; Graph; Time series; Series (stratigraphy); Gene; Data mining; Computational biology; Gene expression; Biology; Artificial intelligence; Machine learning; Theoretical computer science; Genetics","score_opus":0.01041361670249012,"score_gpt":0.2526595291995905,"score_spread":0.2422459124971004,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2070718329","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":[0.010947661,0.0000467885,0.98774135,0.000060865445,0.000011010534,0.000016206926,0.00009916704,0.000939739,0.00013713965],"genre_scores_gemma":[0.18037178,0.00012700567,0.81712043,0.00010340723,0.00004798282,0.00008793743,0.001192779,0.00036114428,0.000587548],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9973068,0.0013307603,0.00013898648,0.0006952566,0.0004381258,0.00009013025],"domain_scores_gemma":[0.9811317,0.013046277,0.0012936203,0.0028710174,0.0014536827,0.000203789],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0069767237,0.0008763173,0.0010354877,0.0027718404,0.00062767515,0.001133671,0.0018632866,0.0008247187,0.0014742167],"category_scores_gemma":[0.036440942,0.0005735535,0.0013373239,0.0017332103,0.000676388,0.0017953491,0.0011440057,0.0020080684,0.00075080286],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00026460734,0.00023715178,0.022032436,0.0002260026,0.00072587474,0.00017511925,0.00038237107,0.38904253,0.037701666,0.012616537,0.0025793589,0.53401643],"study_design_scores_gemma":[0.000015346734,0.00004618359,0.0025488879,0.000008698295,0.00006380846,0.00006738277,0.000029990677,0.97246766,0.0059864353,0.017651297,0.0010913359,0.000022910554],"about_ca_topic_score_codex":0.0062988554,"about_ca_topic_score_gemma":0.010307712,"teacher_disagreement_score":0.0069767237,"about_ca_system_score_codex":0.000615191,"about_ca_system_score_gemma":0.001204536,"threshold_uncertainty_score":0.036896884},"labels":[],"label_agreement":null},{"id":"W2071777796","doi":"10.1142/s021972000300023x","title":"BIOINFORMATICS MEETS PROTEOMICS — BRIDGING THE GAP BETWEEN MASS SPECTROMETRY DATA ANALYSIS AND CELL BIOLOGY","year":2003,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Advanced Proteomics Techniques and Applications","field":"Chemistry","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":"Caprion (Canada)","funders":"","keywords":"Proteomics; Proteome; Computational biology; Context (archaeology); Bridging (networking); Mass spectrometry; Identification (biology); Systems biology; Biomarker discovery; Biology; Bioinformatics; Computer science; Data science; Chemistry; Biochemistry; Chromatography","score_opus":0.02296125878695445,"score_gpt":0.2901927437328728,"score_spread":0.26723148494591836,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2071777796","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":[0.0062375795,0.051935017,0.86059487,0.06628458,0.0019031564,0.00018258936,0.0006589693,0.0056568617,0.0065463637],"genre_scores_gemma":[0.05302979,0.06313216,0.8589207,0.013028365,0.0049951305,0.00026509477,0.0019350452,0.0018380317,0.0028556446],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.98672706,0.0067806914,0.001142216,0.0010959194,0.0038629747,0.00039114986],"domain_scores_gemma":[0.9559509,0.031177519,0.0014881202,0.0048143,0.0036314747,0.00293772],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.021273965,0.0012458296,0.0029317304,0.0040136543,0.0012392215,0.01056652,0.0032375536,0.0044899746,0.0039808913],"category_scores_gemma":[0.029721744,0.0013436505,0.0011317234,0.0064840843,0.0047753737,0.016658204,0.0068088956,0.006776876,0.0056330157],"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":[0.0007202066,0.00025576836,0.0037794418,0.0036927683,0.00044228227,0.0013119214,0.0013626061,0.005221739,0.03305173,0.22889265,0.07074319,0.6505257],"study_design_scores_gemma":[0.00010717382,0.00020215458,0.0025061285,0.000809514,0.000080606245,0.0023092714,0.00073836057,0.040576268,0.009631667,0.5753203,0.36757588,0.00014267502],"about_ca_topic_score_codex":0.0010392417,"about_ca_topic_score_gemma":0.00088930567,"teacher_disagreement_score":0.021273965,"about_ca_system_score_codex":0.0017004925,"about_ca_system_score_gemma":0.0044035735,"threshold_uncertainty_score":0.11250883},"labels":[],"label_agreement":null},{"id":"W2075341611","doi":"10.1142/s0219720005001065","title":"APPROXIMATE MATCHING OF STRUCTURED MOTIFS IN DNA SEQUENCES","year":2005,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","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":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Computer Research Institute of Montréal; Université de Montréal","funders":"","keywords":"Regular expression; Pattern matching; Automaton; Theoretical computer science; Algorithm; Genome; Mathematics; Computer science; Time complexity; Pushdown automaton; Computational biology; Combinatorics; Biology; Genetics; Artificial intelligence; Gene","score_opus":0.008234771360779138,"score_gpt":0.2438183279726577,"score_spread":0.23558355661187858,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2075341611","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.09434583,0.00020979429,0.903499,0.00008546588,0.0000131050665,0.000057009052,0.00010882396,0.00072643015,0.00095457985],"genre_scores_gemma":[0.47797945,0.00021276168,0.51919883,0.00007884196,0.000015938205,0.00013943434,0.0007143747,0.00011970115,0.0015407276],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9987878,0.00034942062,0.000093290364,0.00028459757,0.00040015447,0.00008470055],"domain_scores_gemma":[0.99710625,0.0017237215,0.0003502257,0.00055743166,0.00020837042,0.000054067255],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00083582464,0.0003650072,0.00071269827,0.0011571876,0.00036732174,0.0009660637,0.0010060377,0.00090899406,0.0016891299],"category_scores_gemma":[0.0077958424,0.00039056552,0.0006056403,0.0014281798,0.00089654414,0.002300802,0.0011246743,0.00055820297,0.00037389077],"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":[0.00075062853,0.00018280282,0.0051192176,0.0005693954,0.00015264632,0.00055743736,0.0006053419,0.4766015,0.04987912,0.16233288,0.0017450664,0.30150393],"study_design_scores_gemma":[0.00002943502,0.000096794574,0.00051829405,0.000020089888,0.000017729466,0.00022625562,0.000076003125,0.8747699,0.012370597,0.10985198,0.0020091976,0.000013664855],"about_ca_topic_score_codex":0.0009993588,"about_ca_topic_score_gemma":0.0011924566,"teacher_disagreement_score":0.0016891299,"about_ca_system_score_codex":0.0006209556,"about_ca_system_score_gemma":0.0004900374,"threshold_uncertainty_score":0.005650699},"labels":[],"label_agreement":null},{"id":"W2076795608","doi":"10.1142/s0219720008003527","title":"MANGO: MULTIPLE ALIGNMENT WITH N GAPPED OLIGOS","year":2008,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","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":"University of Waterloo","funders":"Natural Sciences and Engineering Research Council of Canada; National Natural Science Foundation of China; Canada Research Chairs","keywords":"Multiple sequence alignment; Heuristics; Scalability; Computer science; Dynamic programming; Sequence (biology); Sequence alignment; Task (project management); State (computer science); Algorithm; Biology","score_opus":0.0105151779553291,"score_gpt":0.2176171667524678,"score_spread":0.2071019887971387,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2076795608","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":[0.06904059,0.003726095,0.7485019,0.0006204886,0.0007185199,0.00062559557,0.011256474,0.15458757,0.010922762],"genre_scores_gemma":[0.106065184,0.0008516893,0.86531246,0.00045460137,0.000065540204,0.00087545515,0.012839391,0.0083926795,0.0051430003],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99931073,0.00014314108,0.00007160423,0.0002861095,0.00013028123,0.00005815196],"domain_scores_gemma":[0.9994592,0.00023303497,0.000095040734,0.0001148521,0.00005031,0.000047573034],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008737444,0.0017914079,0.0010096957,0.00070144946,0.0008924068,0.0009875882,0.0014532343,0.0010138173,0.005902051],"category_scores_gemma":[0.0024557412,0.00095697155,0.00093828456,0.0009823439,0.0005901013,0.0017370725,0.0014424098,0.0015152985,0.0039305543],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0042406754,0.00039594425,0.00830255,0.005203562,0.0007337775,0.0013154612,0.00094683096,0.045565564,0.3387665,0.02640023,0.12877822,0.43935072],"study_design_scores_gemma":[0.00067198486,0.00080867216,0.0049316147,0.00034429153,0.00027768934,0.0015353967,0.0004122395,0.3711448,0.28808022,0.038036022,0.2934176,0.00033949016],"about_ca_topic_score_codex":0.0011390293,"about_ca_topic_score_gemma":0.0029040552,"teacher_disagreement_score":0.005902051,"about_ca_system_score_codex":0.00060641,"about_ca_system_score_gemma":0.0010518944,"threshold_uncertainty_score":0.019744337},"labels":[],"label_agreement":null},{"id":"W2082232137","doi":"10.1142/s0219720004000776","title":"DISCOVERY OF FUNCTIONAL GENES FOR SYSTEMIC ACQUIRED RESISTANCE IN<i>ARABIDOPSIS THALIANA</i>THROUGH INTEGRATED DATA MINING","year":2004,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Gene expression and cancer classification","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":"National Research Council Canada; Plant Biotechnology Institute","funders":"National Research Council Canada; Natural Sciences and Engineering Research Council of Canada","keywords":"Gene; Cluster analysis; Arabidopsis thaliana; Computational biology; Arabidopsis; Biology; Genetics; Candidate gene; Promoter; Data mining; Computer science; Artificial intelligence; Mutant; Gene expression","score_opus":0.034269504117523004,"score_gpt":0.2854363661434654,"score_spread":0.2511668620259424,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2082232137","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":[0.94379205,0.000769163,0.049318556,0.0001993276,0.000009599127,0.00009269355,0.0043872865,0.0008058969,0.0006254737],"genre_scores_gemma":[0.87805563,0.0003008988,0.11120493,0.000056767374,0.000009441363,0.00013043334,0.009803416,0.000042843312,0.00039571588],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99979955,0.000027242222,0.000018714825,0.00006964657,0.0000583299,0.000026625983],"domain_scores_gemma":[0.99961495,0.00014009874,0.000077919816,0.000029362453,0.00009467909,0.00004284309],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00045058117,0.00034769584,0.00059583253,0.0017237378,0.00022165442,0.00043970227,0.0003538381,0.00026677633,0.0004650282],"category_scores_gemma":[0.00068286725,0.000120967125,0.00067832915,0.0014612358,0.00018296925,0.00023001515,0.0002022451,0.00032679262,0.00021206241],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011863732,0.00048222556,0.141237,0.00075523247,0.0002457601,0.00090003747,0.0002564879,0.0067127743,0.68077034,0.0012278388,0.0012228385,0.1650031],"study_design_scores_gemma":[0.00014463555,0.0010917139,0.5175424,0.000109622124,0.00069533463,0.0022494537,0.00065653864,0.15523155,0.30976188,0.0039277324,0.008485974,0.00010314844],"about_ca_topic_score_codex":0.001057953,"about_ca_topic_score_gemma":0.0013730178,"teacher_disagreement_score":0.0017237378,"about_ca_system_score_codex":0.00036584662,"about_ca_system_score_gemma":0.0005253697,"threshold_uncertainty_score":0.0026544929},"labels":[],"label_agreement":null},{"id":"W2084252156","doi":"10.1142/s0219720004000983","title":"A TUTORIAL OF RECENT DEVELOPMENTS IN THE SEEDING OF LOCAL ALIGNMENT","year":2004,"lang":"en","type":"review","venue":"Journal of Bioinformatics and Computational Biology","topic":"Genomics and Phylogenetic Studies","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":"Western University; University of Waterloo","funders":"Natural Sciences and Engineering Research Council of Canada; Canada Research Chairs","keywords":"Seeding; Heuristic; Focus (optics); Sensitivity (control systems); Computer science; Artificial intelligence; Machine learning; Algorithm; Biology; Engineering; Physics","score_opus":0.03088546029394813,"score_gpt":0.30646730880305817,"score_spread":0.27558184850911005,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2084252156","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":[0.00042269053,0.9485661,0.04016864,0.0005810261,0.001909522,0.000059397786,0.00007920781,0.000407586,0.0078059053],"genre_scores_gemma":[0.0033210807,0.93293864,0.051666148,0.00144735,0.0017637317,0.00017288487,0.0002877421,0.00015559707,0.008246855],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99923766,0.00017242551,0.000076827986,0.00015219589,0.00031552627,0.00004536674],"domain_scores_gemma":[0.9986524,0.0007736842,0.0001156192,0.00008809293,0.0002871433,0.00008306403],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015059771,0.0015901347,0.00174061,0.0035957117,0.00060507806,0.0014829491,0.0025206448,0.0017029884,0.00836128],"category_scores_gemma":[0.00300554,0.00096746004,0.00083882705,0.00473882,0.0010306917,0.003547781,0.0010728387,0.0023674953,0.01271031],"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":[0.000065361055,0.000099500445,0.00019181702,0.007914758,0.000072173876,0.0004246628,0.000095874624,0.0028192736,0.006253403,0.015093219,0.056909315,0.91006064],"study_design_scores_gemma":[0.000015743675,0.00009378675,0.00032164343,0.0012568458,0.000039395105,0.0019701892,0.000033259854,0.0010884221,0.0028138778,0.008123757,0.9841957,0.00004748278],"about_ca_topic_score_codex":0.0009622189,"about_ca_topic_score_gemma":0.0013849753,"teacher_disagreement_score":0.00836128,"about_ca_system_score_codex":0.00097291736,"about_ca_system_score_gemma":0.00085107645,"threshold_uncertainty_score":0.027971268},"labels":[],"label_agreement":null},{"id":"W2084332942","doi":"10.1142/s0219720008003308","title":"FITTING PROTEIN CHAINS TO CUBIC LATTICE IS NP-COMPLETE","year":2008,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Protein Structure and Dynamics","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 Lethbridge; Simon Fraser University","funders":"","keywords":"Lattice (music); Combinatorics; Protein folding; Square lattice; Crystal structure; Mathematics; Chain (unit); Crystallography; Physics; Statistical physics; Chemistry; Quantum mechanics","score_opus":0.012204151468651761,"score_gpt":0.2475589559974527,"score_spread":0.23535480452880095,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2084332942","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":[0.30176434,0.002557711,0.6367725,0.012201718,0.00034610328,0.0006165277,0.0080321655,0.007611036,0.030097853],"genre_scores_gemma":[0.46320724,0.0014056074,0.5116663,0.001667804,0.0002216544,0.0005601316,0.0102113355,0.0008466209,0.010213306],"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99781775,0.00062499347,0.00014841698,0.00061923504,0.00056221394,0.00022724684],"domain_scores_gemma":[0.98854756,0.009105809,0.00064777123,0.00085209624,0.000516924,0.00032979908],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001211096,0.0010816848,0.0019439019,0.0006779462,0.0016941926,0.003111467,0.0024859537,0.0024597673,0.0069940556],"category_scores_gemma":[0.008494579,0.0010151356,0.0018470716,0.00259354,0.0016265367,0.0041815885,0.0022425682,0.0029595515,0.0016821144],"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":[0.00048656636,0.00050588866,0.0035125175,0.0010990739,0.00022867562,0.00075075944,0.00049125083,0.7770999,0.004967293,0.041435163,0.04746048,0.12196239],"study_design_scores_gemma":[0.00015966337,0.000071111484,0.00036864914,0.00005086676,0.000032513057,0.000374482,0.0003633012,0.8495735,0.0024803393,0.13743222,0.009063441,0.000029929217],"about_ca_topic_score_codex":0.008135529,"about_ca_topic_score_gemma":0.0079156635,"teacher_disagreement_score":0.008135529,"about_ca_system_score_codex":0.001454096,"about_ca_system_score_gemma":0.0022976932,"threshold_uncertainty_score":0.023397505},"labels":[],"label_agreement":null},{"id":"W2087799175","doi":"10.1142/s0219720008003801","title":"MODELING OF GLYCEROL-3-PHOSPHATE TRANSPORTER SUGGESTS A POTENTIAL 'TILT' MECHANISM INVOLVED IN ITS FUNCTION","year":2008,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Drug Transport and Resistance Mechanisms","field":"Medicine","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":"National Institute of Allergy and Infectious Diseases; National Institutes of Health; McGill University","keywords":"Major facilitator superfamily; Transporter; Permease; Transmembrane domain; Periplasmic space; Lactose permease; Membrane transport protein; Transport protein; Protein structure; Biology; Transmembrane protein; ATP-binding cassette transporter; Ligand (biochemistry); Biophysics; Chemistry; Biochemistry; Amino acid; Gene; Receptor","score_opus":0.015852783756237724,"score_gpt":0.23200887046751428,"score_spread":0.21615608671127656,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2087799175","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":[0.92854005,0.00032206214,0.05616928,0.00065417663,0.000033872366,0.000037085476,0.00041562394,0.00040210283,0.013425778],"genre_scores_gemma":[0.9909223,0.00019627145,0.0073640305,0.000051108113,0.0000081674325,0.000028645278,0.00021293078,0.000034963632,0.0011816726],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9999565,0.000009974489,0.0000014387865,0.0000064973983,0.000011408446,0.000014181328],"domain_scores_gemma":[0.9999664,0.000008601782,0.000004062167,0.000002987905,0.000008044542,0.000009998703],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00009581518,0.00031301784,0.0003520944,0.00014528443,0.0002994427,0.0007330089,0.00057831244,0.00072132284,0.002244938],"category_scores_gemma":[0.00018426517,0.0002242461,0.0003715673,0.0001857544,0.00030861452,0.00045612,0.0002434918,0.00032172535,0.00033446413],"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":[0.00017974545,0.000049454986,0.0037825024,0.00006211896,0.000033249027,0.00026214487,0.000100618265,0.9549179,0.026973411,0.008382722,0.0005917709,0.0046644816],"study_design_scores_gemma":[0.000021835516,0.00003507696,0.0009566366,0.0000041988646,0.000008900669,0.000040378913,0.000037568112,0.9953028,0.0013848633,0.001455724,0.00074327324,0.000008709856],"about_ca_topic_score_codex":0.0074181096,"about_ca_topic_score_gemma":0.004039823,"teacher_disagreement_score":0.0074181096,"about_ca_system_score_codex":0.00042577277,"about_ca_system_score_gemma":0.00078290066,"threshold_uncertainty_score":0.014749825},"labels":[],"label_agreement":null},{"id":"W2096487774","doi":"10.1142/s0219720010004872","title":"THE IMPORTANCE OF INHIBITORS FOR THE SIMULATION OF METABOLIC PROCESSES: <i>IN SILICO</i><font>Zn<sup>2+</sup></font> INHIBITION OF m-ACONITASE FROM ANALYSIS OF GLYCOLYSIS AND KREBS CYCLE KINETIC MODELS","year":2010,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Enzyme Structure and Function","field":"Materials 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":"National Research Council Canada; Mount Allison University; University of New Brunswick","funders":"","keywords":"Aconitase; Glycolysis; Citric acid cycle; Enzyme; Chemistry; Metabolite; Non-competitive inhibition; Metabolism; Biochemistry; Mixed inhibition; Biophysics; Biology","score_opus":0.008604154339040906,"score_gpt":0.24694955928707146,"score_spread":0.23834540494803055,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2096487774","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":[0.38957995,0.0010514358,0.5718821,0.0008957209,0.00021786692,0.0002503258,0.0021807821,0.0014081815,0.032533567],"genre_scores_gemma":[0.94044715,0.00075397186,0.0517367,0.00014887449,0.000029455103,0.00051513524,0.0008306847,0.00021964067,0.0053183623],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99986684,0.000049283473,0.000009347949,0.000023563603,0.00003130686,0.000019607816],"domain_scores_gemma":[0.999166,0.00062123395,0.000073457784,0.000038408474,0.00007993761,0.000020988442],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003341435,0.0006566566,0.0007193588,0.00020966188,0.00029746004,0.0007737649,0.0008147156,0.0010725326,0.0026439354],"category_scores_gemma":[0.0014528269,0.0003298636,0.00065161,0.0003023386,0.0003027632,0.0005235154,0.0003649373,0.0006914473,0.00048163158],"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":[0.000056420646,0.000026014692,0.00067714107,0.00007945645,0.000015589185,0.00004267196,0.000018680084,0.99192035,0.0037692443,0.001795791,0.00014351361,0.0014550986],"study_design_scores_gemma":[0.000009539836,0.000017006709,0.00015767195,0.0000048005645,0.000009482393,0.000009924648,0.0000061789724,0.99645823,0.0021466117,0.00045612972,0.00072001055,0.000004359981],"about_ca_topic_score_codex":0.008106062,"about_ca_topic_score_gemma":0.0041852817,"teacher_disagreement_score":0.008106062,"about_ca_system_score_codex":0.00067985134,"about_ca_system_score_gemma":0.00079061225,"threshold_uncertainty_score":0.016117752},"labels":[],"label_agreement":null},{"id":"W2097950066","doi":"10.1142/s0219720014410091","title":"<i>In silico</i>evolution of the<i>hunchback</i>gene indicates redundancy in cis-regulatory organization and spatial gene expression","year":2014,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Plant Molecular Biology Research","field":"Agricultural and Biological Sciences","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":"British Columbia Institute of Technology","funders":"National Institute of General Medical Sciences; National Institutes of Health; Russian Foundation for Basic Research","keywords":"Modularity (biology); Computational biology; Biology; In silico; Modular design; Redundancy (engineering); Gene regulatory network; Cis-regulatory module; Robustness (evolution); Gene; Genetics; Regulatory sequence; Expression (computer science); Regulation of gene expression; Gene expression; Function (biology); Computer science; Promoter","score_opus":0.005311822434021521,"score_gpt":0.2025287940153882,"score_spread":0.1972169715813667,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2097950066","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":[0.95313036,0.000042225653,0.043994512,0.00011667442,0.00001581856,0.00001673851,0.00016584375,0.00021959751,0.002298203],"genre_scores_gemma":[0.97994244,0.000043531807,0.019186677,0.000045670655,0.0000040287205,0.000035615387,0.00024651445,0.00005436896,0.00044124367],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9998965,0.00003735627,0.000004795778,0.000027978473,0.000018221086,0.000015108937],"domain_scores_gemma":[0.9995833,0.00027900326,0.00005603523,0.00003716472,0.000026937252,0.000017568218],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00036409963,0.00024393306,0.00028953713,0.00023088416,0.00035288814,0.00034937664,0.00050155615,0.0005200154,0.0012862191],"category_scores_gemma":[0.0010687918,0.00020194335,0.00047764598,0.0002486536,0.00038296237,0.00027900471,0.00018636361,0.00041194938,0.00013693681],"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":[0.00008317934,0.00006412936,0.008689356,0.000069861766,0.00006796576,0.00010779419,0.000041608175,0.9667109,0.016558642,0.0038584678,0.00019014081,0.003557928],"study_design_scores_gemma":[0.0000134144,0.000050367147,0.0015224344,0.0000035371756,0.000018666156,0.00004252741,0.000010980595,0.99222296,0.004630279,0.00123191,0.00024711585,0.0000058356723],"about_ca_topic_score_codex":0.003982291,"about_ca_topic_score_gemma":0.004320832,"teacher_disagreement_score":0.003982291,"about_ca_system_score_codex":0.000529485,"about_ca_system_score_gemma":0.00044767477,"threshold_uncertainty_score":0.007918239},"labels":[],"label_agreement":null},{"id":"W2107187602","doi":"10.1142/s0219720005001053","title":"CLUSTERING AND RE-CLUSTERING FOR PATTERN DISCOVERY IN GENE EXPRESSION DATA","year":2005,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","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":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Guelph","funders":"","keywords":"Cluster analysis; Data mining; Single-linkage clustering; Computational biology; Computer science; Correlation clustering; Expression (computer science); Gene; Pairwise comparison; CURE data clustering algorithm; Pattern recognition (psychology); Biology; Artificial intelligence; Genetics","score_opus":0.02993371998511228,"score_gpt":0.30038400889389766,"score_spread":0.27045028890878536,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2107187602","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":[0.0012226457,0.00034193273,0.9958235,0.00008737418,0.000048304402,0.00022766639,0.00014363944,0.0019070965,0.00019790641],"genre_scores_gemma":[0.0117342435,0.00025249124,0.9861145,0.000066141976,0.00003324242,0.00054258615,0.0006631865,0.0002604401,0.0003330922],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9878433,0.0041750073,0.0011812875,0.0032976938,0.0031666057,0.00033622925],"domain_scores_gemma":[0.98968494,0.0046468037,0.0010161585,0.0024051047,0.0020649221,0.00018213123],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007930526,0.0030060858,0.0037627118,0.007904905,0.0019852964,0.0026439442,0.004909215,0.0027658239,0.0021494892],"category_scores_gemma":[0.022644086,0.0015283957,0.0044184034,0.009844454,0.0018630887,0.0028308197,0.0027758598,0.004423226,0.0034468917],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0004667711,0.00030562296,0.002449723,0.0017282823,0.0010647497,0.00047696655,0.0010820272,0.18875225,0.025945228,0.032899193,0.010940012,0.7338892],"study_design_scores_gemma":[0.000052213356,0.00010387045,0.0015516636,0.00009581212,0.00010369012,0.00043314163,0.00016004844,0.93408436,0.011946503,0.03751905,0.013796913,0.00015275134],"about_ca_topic_score_codex":0.0052297963,"about_ca_topic_score_gemma":0.004830847,"teacher_disagreement_score":0.007930526,"about_ca_system_score_codex":0.0018231813,"about_ca_system_score_gemma":0.0026298335,"threshold_uncertainty_score":0.041941166},"labels":[],"label_agreement":null},{"id":"W2109628189","doi":"10.1142/s0219720013500078","title":"<i>DE NOVO</i> SEQUENCING WITH LIMITED NUMBER OF POST-TRANSLATIONAL MODIFICATIONS PER PEPTIDE","year":2013,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Advanced Proteomics Techniques and Applications","field":"Chemistry","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 Waterloo","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Computational biology; Posttranslational modification; Computer science; Proteomics; Peptide; Sequence (biology); DNA sequencing; Tandem mass spectrometry; Biology; Identification (biology); Chemistry; Genetics; Mass spectrometry; Biochemistry; Gene; Chromatography","score_opus":0.011816309144595085,"score_gpt":0.2614766387186723,"score_spread":0.24966032957407724,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2109628189","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.008013241,0.0002606909,0.9826151,0.000094136376,0.00008614901,0.00011311155,0.0004432007,0.0062025376,0.0021717781],"genre_scores_gemma":[0.034816265,0.00034120772,0.95826095,0.00016762021,0.000031355845,0.00017237544,0.001873976,0.0011050991,0.0032311676],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99926704,0.00010141534,0.00007385621,0.00031128153,0.00019025536,0.000056209992],"domain_scores_gemma":[0.9989139,0.00028266906,0.00016610508,0.0003276533,0.0002572102,0.000052456136],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013055488,0.0011346055,0.0009310285,0.0007134747,0.00082045177,0.0016782904,0.0018392113,0.0008732862,0.0048190625],"category_scores_gemma":[0.0032132524,0.00079533196,0.0010578316,0.00089843577,0.0005580388,0.001384028,0.0013439953,0.0013362952,0.004885345],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00083386176,0.00015178842,0.003286427,0.0010677197,0.00022121618,0.00047053065,0.00025985023,0.021029877,0.42240024,0.022463467,0.0225028,0.50531226],"study_design_scores_gemma":[0.00011196966,0.00024991258,0.0024071361,0.00010264287,0.00009377291,0.0015175104,0.00007392271,0.32340637,0.5413509,0.020331448,0.11022699,0.00012744276],"about_ca_topic_score_codex":0.0014150876,"about_ca_topic_score_gemma":0.0018987708,"teacher_disagreement_score":0.0048190625,"about_ca_system_score_codex":0.0006689685,"about_ca_system_score_gemma":0.0009394376,"threshold_uncertainty_score":0.016121387},"labels":[],"label_agreement":null},{"id":"W2110186020","doi":"10.1142/s0219720009004345","title":"PREDICTING LOCAL QUALITY OF A SEQUENCE–STRUCTURE ALIGNMENT","year":2009,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Protein Structure and 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":"University of Waterloo","funders":"Natural Sciences and Engineering Research Council of Canada; Ministry of Science and Technology of the People's Republic of China","keywords":"Protein structure prediction; Threading (protein sequence); Computer science; Support vector machine; CASP; Sequence (biology); Artificial intelligence; Quality (philosophy); Data mining; Homology modeling; Local structure; Similarity (geometry); Loop modeling; Smith–Waterman algorithm; Protein structure; Sequence alignment; Biology; Peptide sequence","score_opus":0.010224034197594269,"score_gpt":0.28141123862534545,"score_spread":0.2711872044277512,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2110186020","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":[0.852674,0.000614566,0.14297387,0.00007607402,0.000016478949,0.000042520423,0.0006069597,0.0022283127,0.0007671474],"genre_scores_gemma":[0.95268214,0.00013295832,0.045426954,0.000019873394,0.000012107534,0.00001759661,0.0013593667,0.00008541765,0.0002636301],"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99918884,0.0001301481,0.00007858766,0.00020353687,0.00032769475,0.0000712058],"domain_scores_gemma":[0.9954211,0.0015446093,0.0012062958,0.00038705728,0.0011262415,0.00031457483],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015526998,0.00045653235,0.0008225893,0.0028837123,0.00025864827,0.0006744495,0.00056402036,0.0005919856,0.0010623559],"category_scores_gemma":[0.0050314255,0.00024509543,0.0005695731,0.0013224877,0.00030791212,0.0011153765,0.00051228644,0.00048809714,0.0006701812],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0015619693,0.00042857305,0.34653264,0.00041154155,0.00041526408,0.0006822518,0.00029506112,0.20928879,0.20230213,0.0016323185,0.0034016303,0.2330478],"study_design_scores_gemma":[0.000023843957,0.00031756717,0.057957664,0.000017218437,0.0000476518,0.0002402885,0.00009211493,0.91406476,0.025812957,0.00086579745,0.0005297159,0.000030472786],"about_ca_topic_score_codex":0.002112957,"about_ca_topic_score_gemma":0.0027833395,"teacher_disagreement_score":0.0028837123,"about_ca_system_score_codex":0.0004266422,"about_ca_system_score_gemma":0.00045731268,"threshold_uncertainty_score":0.008211553},"labels":[],"label_agreement":null},{"id":"W2111107841","doi":"10.1142/s0219720004000545","title":"DERIVING HAPLOTYPES THROUGH RECOMBINATION AND GENE CONVERSION PATHWAYS","year":2004,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","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":"Université de Montréal","funders":"","keywords":"Haplotype; Biology; Genetics; Genome; Haplotype estimation; Gene; Evolutionary biology; Computational biology; Allele","score_opus":0.01155091853823511,"score_gpt":0.2252479685338379,"score_spread":0.2136970499956028,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2111107841","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":[0.23450682,0.00015874136,0.760497,0.0002319421,0.000014804601,0.00015110675,0.0003062281,0.0008030912,0.003330227],"genre_scores_gemma":[0.47647694,0.0002289093,0.51994544,0.000048354173,0.000010403469,0.00012862477,0.0009830204,0.00020600019,0.0019723761],"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99955076,0.0001251851,0.000031444393,0.00012898436,0.00008616565,0.00007742196],"domain_scores_gemma":[0.9986091,0.0007817296,0.00013305336,0.00025748851,0.00013486281,0.000083768726],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008515072,0.00045849686,0.00070249866,0.00097155763,0.0010443273,0.0011813388,0.0010710153,0.00074890815,0.0023436071],"category_scores_gemma":[0.0036197416,0.00041222223,0.0011011899,0.0010672727,0.0007921767,0.0015835087,0.0015355678,0.000893152,0.0005136519],"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":[0.00050372514,0.00036822932,0.017310072,0.0004906024,0.0001237023,0.0010924069,0.001242781,0.43839118,0.03633341,0.1351084,0.0028362672,0.36619923],"study_design_scores_gemma":[0.0000643685,0.00015485336,0.0018763986,0.00004743276,0.00007142046,0.00047069314,0.00029838068,0.72709614,0.019747686,0.24325785,0.0068701883,0.00004465629],"about_ca_topic_score_codex":0.0016993985,"about_ca_topic_score_gemma":0.0023388716,"teacher_disagreement_score":0.0023436071,"about_ca_system_score_codex":0.000710712,"about_ca_system_score_gemma":0.0010079531,"threshold_uncertainty_score":0.007840097},"labels":[],"label_agreement":null},{"id":"W2112332161","doi":"10.1142/s0219720004000764","title":"MODELING AND SIMULATION OF MOLECULAR BIOLOGY SYSTEMS USING PETRI NETS: MODELING GOALS OF VARIOUS APPROACHES","year":2004,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Gene Regulatory Network Analysis","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":"Polytechnique Montréal","funders":"","keywords":"Petri net; Computer science; Concurrency; Process architecture; Stochastic Petri net; Representation (politics); Synchronization (alternating current); Systems biology; Theoretical computer science; Modeling and simulation; Distributed computing; Event (particle physics); Programming language; Computational biology; Simulation; Biology","score_opus":0.026486863741333418,"score_gpt":0.2677644995817309,"score_spread":0.24127763584039746,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2112332161","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0033005024,0.0057569086,0.98576736,0.000989247,0.00008224776,0.00007901246,0.000059669383,0.00024164372,0.003723386],"genre_scores_gemma":[0.13015889,0.028214939,0.8369062,0.00038338793,0.0003316108,0.0004745365,0.00025425424,0.00015246186,0.0031237395],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9983374,0.00059440406,0.00015866563,0.0002035396,0.0006271937,0.00007880982],"domain_scores_gemma":[0.9987276,0.0007303829,0.00010313997,0.00019004679,0.00018427838,0.000064541506],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0030589362,0.0014520489,0.0011829997,0.0017533455,0.0006512453,0.0028797677,0.002286255,0.0016032105,0.00091424375],"category_scores_gemma":[0.003447031,0.0010555122,0.0021365038,0.0018477339,0.0024128787,0.004782325,0.001998575,0.0024387399,0.0004997784],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006190196,0.00008397232,0.0008015718,0.000735965,0.00010055119,0.00013979187,0.00035142835,0.31901148,0.0034513846,0.61695933,0.0007987012,0.057503868],"study_design_scores_gemma":[0.000023126946,0.00005739285,0.0001909586,0.00026349246,0.00007121828,0.00012673096,0.00010137112,0.60914433,0.00420355,0.35797778,0.027787255,0.000052725034],"about_ca_topic_score_codex":0.0027513534,"about_ca_topic_score_gemma":0.0016158181,"teacher_disagreement_score":0.0030589362,"about_ca_system_score_codex":0.0018484453,"about_ca_system_score_gemma":0.0020392877,"threshold_uncertainty_score":0.016177416},"labels":[],"label_agreement":null},{"id":"W2115024530","doi":"10.1142/s021972000400082x","title":"PAIRWISE PROTEIN STRUCTURE ALIGNMENT BASED ON AN ORIENTATION-INDEPENDENT BACKBONE REPRESENTATION","year":2004,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","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 Alberta","funders":"Fudan University; Purdue University","keywords":"Pairwise comparison; Representation (politics); Structural alignment; Algorithm; Computer science; Orientation (vector space); Multiple sequence alignment; Dynamic programming; Protein structure; Artificial intelligence; Mathematics; Sequence alignment; Geometry","score_opus":0.005996132849436699,"score_gpt":0.2523833834124025,"score_spread":0.24638725056296581,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2115024530","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.01151963,0.0001296208,0.98628104,0.00006154472,0.000017298009,0.0000462242,0.000058606878,0.0011550111,0.00073103077],"genre_scores_gemma":[0.10139999,0.00019777406,0.8968277,0.000044001983,0.000016888907,0.00008498226,0.000385137,0.00017363865,0.0008699188],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9992537,0.00016275297,0.00003861116,0.00023612313,0.00025610937,0.00005269534],"domain_scores_gemma":[0.9995004,0.00012903113,0.00009705797,0.000131449,0.00010194031,0.000040020845],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007290391,0.0006616486,0.0011203053,0.0014278799,0.0006076675,0.0008623493,0.0015362398,0.00072154216,0.0017630013],"category_scores_gemma":[0.002356738,0.00053405436,0.0006125272,0.0015292335,0.00057021214,0.001846242,0.00092751754,0.0012663817,0.0010142782],"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":[0.00028373676,0.00023874492,0.0012549565,0.00015498034,0.000117578325,0.00019154022,0.00015504909,0.12965736,0.12455946,0.05122953,0.003955488,0.6882015],"study_design_scores_gemma":[0.00006881522,0.0002460031,0.00087158056,0.000018526012,0.00004934549,0.00039268055,0.0000616711,0.91807526,0.037427735,0.03571985,0.0070269057,0.000041642645],"about_ca_topic_score_codex":0.001025128,"about_ca_topic_score_gemma":0.0013618207,"teacher_disagreement_score":0.0017630013,"about_ca_system_score_codex":0.0005512465,"about_ca_system_score_gemma":0.0010697233,"threshold_uncertainty_score":0.0058978796},"labels":[],"label_agreement":null},{"id":"W2119868771","doi":"10.1142/s021972001000504x","title":"ON COMPARING TWO STRUCTURED RNA MULTIPLE ALIGNMENTS","year":2010,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","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":"Western University","funders":"","keywords":"Multiple sequence alignment; Computer science; Pairwise comparison; Phylogenetic tree; Alignment-free sequence analysis; Structural alignment; Sequence alignment; Set (abstract data type); Dynamic programming; Block (permutation group theory); Tree (set theory); Web server; Data mining; Theoretical computer science; Algorithm; Computational biology; Biology; Artificial intelligence; Mathematics; Combinatorics; Genetics; The Internet; Peptide sequence","score_opus":0.009100982297949489,"score_gpt":0.25120980141436283,"score_spread":0.24210881911641335,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2119868771","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":[0.061475553,0.0010767616,0.92981213,0.00014615752,0.00023084902,0.00018862724,0.0006341302,0.002805926,0.0036298796],"genre_scores_gemma":[0.13688213,0.00046820546,0.85868305,0.00015800193,0.00008136159,0.0002132545,0.00197553,0.0006393144,0.00089922495],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99391246,0.0019839024,0.00042476054,0.0011460886,0.0023168037,0.00021593535],"domain_scores_gemma":[0.9855012,0.0090447,0.0017010483,0.001328004,0.0021083895,0.00031662267],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0057849777,0.0012230811,0.0014619366,0.007380808,0.0013338056,0.0026495024,0.0015999356,0.0015457835,0.0044388915],"category_scores_gemma":[0.02785137,0.00074729376,0.0009631651,0.005971102,0.0011746936,0.004046338,0.0021335902,0.0011640376,0.0020061638],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0023399687,0.00027044074,0.017269205,0.0017556737,0.0010588728,0.0011565968,0.0010777161,0.062180836,0.099740975,0.03861651,0.0058316346,0.76870155],"study_design_scores_gemma":[0.00032137547,0.0016856214,0.02062542,0.00048506906,0.0005705811,0.003834884,0.0015279674,0.6641399,0.08996728,0.16679294,0.049691584,0.0003573097],"about_ca_topic_score_codex":0.00062893383,"about_ca_topic_score_gemma":0.00096713204,"teacher_disagreement_score":0.007380808,"about_ca_system_score_codex":0.00064044364,"about_ca_system_score_gemma":0.00087966956,"threshold_uncertainty_score":0.03059423},"labels":[],"label_agreement":null},{"id":"W2120649366","doi":"10.1142/s0219720008003321","title":"DESIGN AND ANALYSIS OF QUANTITATIVE DIFFERENTIAL PROTEOMICS INVESTIGATIONS USING LC-MS TECHNOLOGY","year":2008,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","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 Toronto; University Health Network","funders":"","keywords":"Pipeline (software); Pooling; Computer science; Biomarker discovery; Proteomics; Quantitative proteomics; Data mining; Computational biology; Identification (biology); Deconvolution; Artificial intelligence; Biology; Algorithm","score_opus":0.037417142446872394,"score_gpt":0.30122854702450924,"score_spread":0.26381140457763685,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2120649366","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.027236747,0.00007087397,0.9670857,0.0000702448,0.000023986862,0.0015129538,0.00066693465,0.0024599514,0.00087257946],"genre_scores_gemma":[0.10626443,0.00012421662,0.8861123,0.000101325124,0.000013878597,0.005176685,0.0012197837,0.00022384444,0.0007635286],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9982595,0.0003129622,0.00011746249,0.00053686264,0.00057074084,0.00020250842],"domain_scores_gemma":[0.99910766,0.00030996322,0.00010006522,0.00012198948,0.00026617924,0.00009413139],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0032512024,0.0011679276,0.0009570892,0.00064447866,0.0005150494,0.001149692,0.0017050642,0.0005470411,0.0024081727],"category_scores_gemma":[0.0027157809,0.00081246806,0.0006991293,0.0006560958,0.00076798437,0.00059078843,0.0010688367,0.0010453974,0.0012658444],"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":[0.0017119996,0.0008408659,0.006175787,0.0007765585,0.00013929208,0.00036330704,0.00014559866,0.15736517,0.68286633,0.017625136,0.0022302922,0.12975967],"study_design_scores_gemma":[0.00024299977,0.0010929903,0.0028988493,0.000018290119,0.00006532134,0.00008770203,0.000036015776,0.7927226,0.18688418,0.008339106,0.0075506177,0.00006141296],"about_ca_topic_score_codex":0.0010186266,"about_ca_topic_score_gemma":0.001092903,"teacher_disagreement_score":0.0032512024,"about_ca_system_score_codex":0.0011157125,"about_ca_system_score_gemma":0.0024709206,"threshold_uncertainty_score":0.017194271},"labels":[],"label_agreement":null},{"id":"W2120691528","doi":"10.1142/s0219720006002284","title":"COMPARATIVE GENOMICS OF TRANSCRIPTIONAL REGULATION IN YEASTS AND ITS APPLICATION TO IDENTIFICATION OF A CANDIDATE ALPHA-ISOPROPYLMALATE TRANSPORTER","year":2006,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","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":"University of Toronto","funders":"Russian Academy of Sciences; Ludwig Institute for Cancer Research; Pew Charitable Trusts; Burroughs Wellcome Fund; Howard Hughes Medical Institute","keywords":"Biology; Start codon; Genetics; Genome; Conserved sequence; Regulator; Computational biology; Saccharomyces cerevisiae; Genomics; Gene; Peptide sequence; Messenger RNA","score_opus":0.007871627652280947,"score_gpt":0.24039515579791132,"score_spread":0.23252352814563038,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2120691528","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":[0.99243927,0.0004096092,0.0064330357,0.00001879621,0.0000019567972,0.0000074228706,0.00040344594,0.00005580442,0.00023062425],"genre_scores_gemma":[0.9904199,0.00028188666,0.0076620574,0.000018785071,0.000002787017,0.000023745195,0.0012968709,0.000017402732,0.0002765025],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99989593,0.000017855155,0.000005691773,0.000042546337,0.000023790091,0.000014152704],"domain_scores_gemma":[0.9998646,0.000047830432,0.000026153262,0.000015129881,0.000020772619,0.000025594194],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000178966,0.00013474622,0.00022963563,0.00042960592,0.00011885323,0.00021604684,0.00013539047,0.00011548563,0.00043018907],"category_scores_gemma":[0.00026478915,0.00012229406,0.00024968252,0.00042718204,0.00017711527,0.00009589187,0.00013894675,0.00020715693,0.00013933708],"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":[0.00013589145,0.000009084324,0.003415677,0.000036757847,0.000011642498,0.000019601548,0.000046680445,0.0002817606,0.9912636,0.000100273195,0.000015865056,0.004663044],"study_design_scores_gemma":[0.000030965486,0.000519253,0.3923104,0.00001669848,0.000091501824,0.00076484395,0.0002619728,0.016158612,0.58627504,0.0007195303,0.0028256606,0.000025491052],"about_ca_topic_score_codex":0.0010435604,"about_ca_topic_score_gemma":0.0013589612,"teacher_disagreement_score":0.0010435604,"about_ca_system_score_codex":0.00021211823,"about_ca_system_score_gemma":0.00014814243,"threshold_uncertainty_score":0.002074957},"labels":[],"label_agreement":null},{"id":"W2124808836","doi":"10.1142/s0219720008003485","title":"PP<i>i</i>Clust: EFFICIENT CLUSTERING OF 3D PROTEIN–PROTEIN INTERACTION INTERFACES","year":2008,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Protein Structure and 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":"Lunenfeld-Tanenbaum Research Institute; University of Toronto; Mount Sinai Hospital","funders":"","keywords":"Interface (matter); Cluster analysis; Cluster (spacecraft); Protein–protein interaction; Computer science; Computational biology; Scalability; Biological system; Data mining; Chemistry; Biology; Artificial intelligence; Database; Biochemistry; Parallel computing; Programming language","score_opus":0.00792344490157071,"score_gpt":0.24124251798026833,"score_spread":0.23331907307869762,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2124808836","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.015639592,0.0003465172,0.9498558,0.00025120386,0.00014379383,0.00024838,0.0036930689,0.028819093,0.0010024809],"genre_scores_gemma":[0.041006528,0.00034712258,0.9382896,0.00012653427,0.00005475186,0.000704872,0.013828905,0.003711686,0.0019300109],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9989202,0.00021285735,0.0000821144,0.00021129275,0.00049072463,0.000082727965],"domain_scores_gemma":[0.99885666,0.00025708362,0.00014449016,0.00031741807,0.0003080479,0.00011636992],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013219109,0.0020481932,0.0018316478,0.0032142194,0.0013941813,0.0019189452,0.0026872428,0.0013960646,0.003678318],"category_scores_gemma":[0.0039119646,0.001064314,0.0018724264,0.003968706,0.00087110425,0.0017556137,0.0022724213,0.0020776887,0.003836146],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013952942,0.0004009266,0.0062139737,0.0016944655,0.0010566646,0.00091006915,0.0007588399,0.07085714,0.14272112,0.01644251,0.14031924,0.61722976],"study_design_scores_gemma":[0.00027987032,0.00014195518,0.0038344667,0.000054676588,0.000119192184,0.0008118375,0.00013125355,0.8727344,0.065709,0.019565875,0.036464747,0.00015278594],"about_ca_topic_score_codex":0.0037200614,"about_ca_topic_score_gemma":0.004887376,"teacher_disagreement_score":0.0037200614,"about_ca_system_score_codex":0.00075996306,"about_ca_system_score_gemma":0.0015518042,"threshold_uncertainty_score":0.0123052},"labels":[],"label_agreement":null},{"id":"W2127251093","doi":"10.1142/s0219720012500217","title":"A FLEXIBLE NONPARAMETRIC APPROACH TO FIND CANDIDATE GENES ASSOCIATED WITH DISEASE IN MICROARRAY EXPERIMENTS","year":2012,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","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":"Hospital for Sick Children; McMaster University; Public Health Ontario; University of Toronto","funders":"Canadian Institutes of Health Research","keywords":"Candidate gene; Gene; Function (biology); Computational biology; Nonparametric statistics; Biology; Microarray; Genetics; Microarray analysis techniques; DNA microarray; Gene prediction; Gene expression; Mathematics; Statistics; Genome","score_opus":0.017192287097399873,"score_gpt":0.27662449809982526,"score_spread":0.25943221100242536,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2127251093","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.027955778,0.00011252227,0.97072923,0.00009976443,0.000021156084,0.00010093522,0.00008691855,0.0006523553,0.00024132509],"genre_scores_gemma":[0.42909575,0.0000667357,0.5688172,0.00013286402,0.00007075938,0.00068497367,0.0003725624,0.00008522181,0.000673961],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9911301,0.005866256,0.0003040152,0.0012372814,0.0012524379,0.00020998115],"domain_scores_gemma":[0.9754015,0.01873788,0.001658417,0.002764685,0.0011611242,0.00027635958],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.013816659,0.0007306572,0.0012932282,0.0025040014,0.0008386081,0.0010796023,0.0017921875,0.001605833,0.0010856113],"category_scores_gemma":[0.037185375,0.00040586098,0.0013520914,0.0021817675,0.0021147362,0.0009650522,0.0014711476,0.001665878,0.00034639574],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014091083,0.0006746615,0.038471717,0.00044412113,0.00073695625,0.00087915524,0.00046210946,0.29315495,0.052417196,0.02828565,0.003410247,0.57965416],"study_design_scores_gemma":[0.000076959856,0.0003415643,0.012156629,0.000016054586,0.000052680036,0.0003905069,0.000043070635,0.9637091,0.0044235126,0.017543223,0.0011787914,0.00006790356],"about_ca_topic_score_codex":0.0010656782,"about_ca_topic_score_gemma":0.0013789093,"teacher_disagreement_score":0.013816659,"about_ca_system_score_codex":0.0006570607,"about_ca_system_score_gemma":0.0011101755,"threshold_uncertainty_score":0.07307035},"labels":[],"label_agreement":null},{"id":"W2131989567","doi":"10.1142/s0219720004000661","title":"PATTERNHUNTER II: HIGHLY SENSITIVE AND FAST HOMOLOGY SEARCH","year":2004,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Topological and Geometric Data Analysis","field":"Computer Science","cited_by":231,"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; Bioinformatics Solutions (Canada); University of Waterloo","funders":"","keywords":"Homology (biology); Sensitivity (control systems); Smith–Waterman algorithm; Computer science; Mathematics; Computational biology; Biology; Genetics; Sequence alignment; Engineering; Gene","score_opus":0.011082962675431672,"score_gpt":0.24645442387469926,"score_spread":0.23537146119926758,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2131989567","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.013777559,0.00028938756,0.9273479,0.00019481477,0.00019225062,0.00037065335,0.0010743833,0.05463226,0.0021207745],"genre_scores_gemma":[0.04205138,0.00024391714,0.9419804,0.00025774533,0.000060055245,0.0007540928,0.0038572196,0.005696155,0.0050989767],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99765587,0.0004229766,0.00024930775,0.00065841665,0.00079451303,0.00021885126],"domain_scores_gemma":[0.99811006,0.0005311118,0.0001635919,0.0007357506,0.00027617236,0.00018333274],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002282726,0.0016072699,0.0018246198,0.0016342432,0.0008225278,0.0018760427,0.0027885071,0.0012182067,0.0099850455],"category_scores_gemma":[0.0063893106,0.0014380319,0.00085661496,0.0017703361,0.0006110144,0.0026020182,0.0025420357,0.0027363903,0.009640096],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0016433424,0.00038196886,0.0026265346,0.0008952788,0.00021345123,0.00049669767,0.00032246535,0.0026140949,0.46982017,0.011588046,0.049398333,0.45999968],"study_design_scores_gemma":[0.0005336692,0.0007214532,0.0035192373,0.000101277066,0.00013650379,0.002558076,0.00013134204,0.14124632,0.7127826,0.016178975,0.12181474,0.0002758991],"about_ca_topic_score_codex":0.0004803636,"about_ca_topic_score_gemma":0.0009404728,"teacher_disagreement_score":0.0099850455,"about_ca_system_score_codex":0.00048170827,"about_ca_system_score_gemma":0.0010911537,"threshold_uncertainty_score":0.033403337},"labels":[],"label_agreement":null},{"id":"W2133854884","doi":"10.1142/s0219720014500073","title":"Prediction of gene-based drug indications using compendia of public gene expression data and PubMed abstracts","year":2014,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","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":"University of Calgary","funders":"","keywords":"Microarray databases; Computer science; Drug discovery; Relevance (law); Data mining; Computational biology; Microarray analysis techniques; Disease; Raw data; Biological data; Drug development; Drug; Bioinformatics; Gene; Biology; Medicine; Gene expression; Genetics","score_opus":0.030645397612834073,"score_gpt":0.24865058207971955,"score_spread":0.21800518446688547,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2133854884","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":[0.25835463,0.011896551,0.1792768,0.0020156223,0.0005984392,0.00091095525,0.5116246,0.025177937,0.010144479],"genre_scores_gemma":[0.26566038,0.005014076,0.35008195,0.00056259055,0.00024066782,0.0009864868,0.37533972,0.00034328966,0.0017708379],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9992373,0.000108485605,0.00015408006,0.00021948433,0.00023220068,0.000048424572],"domain_scores_gemma":[0.99482536,0.0036457763,0.00053093076,0.00025436308,0.00058874785,0.0001548948],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013894294,0.0014756176,0.0012857089,0.0110864965,0.0003721359,0.001103497,0.00077465986,0.00060875725,0.0036616758],"category_scores_gemma":[0.005996827,0.00043438943,0.0018611018,0.005836069,0.00028318728,0.0008870605,0.00045010637,0.0008237601,0.0023210978],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0042702407,0.0010779477,0.10036076,0.010120551,0.0021121257,0.0071024355,0.00043138288,0.055945866,0.172084,0.006793343,0.074095495,0.5656058],"study_design_scores_gemma":[0.001072318,0.0022515948,0.19735079,0.00101994,0.0035891293,0.010268064,0.0006012784,0.4216313,0.1385591,0.018699963,0.2044799,0.0004764996],"about_ca_topic_score_codex":0.0043552285,"about_ca_topic_score_gemma":0.009624125,"teacher_disagreement_score":0.0110864965,"about_ca_system_score_codex":0.0007588334,"about_ca_system_score_gemma":0.002210976,"threshold_uncertainty_score":0.01224947},"labels":[],"label_agreement":null},{"id":"W2133903964","doi":"10.1142/s0219720004000326","title":"OPTIMAL SPACED SEEDS FOR HOMOLOGOUS CODING REGIONS","year":2004,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Genomics and Phylogenetic Studies","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":"University of Waterloo","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Genome; Coding (social sciences); Coding region; Homology (biology); Computational biology; Homologous chromosome; Computer science; Homologous recombination; Smith–Waterman algorithm; Gene; Biology; Genetics; Algorithm; Mathematics; Sequence alignment; Statistics","score_opus":0.013003031956219038,"score_gpt":0.25438642812638906,"score_spread":0.24138339617017002,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2133903964","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.12279987,0.00023919993,0.87334526,0.000102057464,0.00003631919,0.000077308505,0.00022615181,0.0022841883,0.0008896658],"genre_scores_gemma":[0.28108948,0.000074769894,0.7170447,0.000047988557,0.000014713045,0.0000950573,0.00059927633,0.0004430559,0.0005909331],"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99917114,0.00021230802,0.000066187065,0.00025589924,0.00022323782,0.00007122626],"domain_scores_gemma":[0.996618,0.00179083,0.0004560767,0.000500432,0.0004877512,0.00014694162],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012632651,0.0005552517,0.0008563069,0.0014433758,0.00072191464,0.00084209454,0.0008710564,0.0010677837,0.0021127297],"category_scores_gemma":[0.009675721,0.0006581612,0.0006713661,0.0012969935,0.0008749198,0.0013867755,0.0008702588,0.00078467344,0.0010311198],"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":[0.0017228344,0.00031314197,0.014547366,0.0006951745,0.000102540886,0.0015376052,0.0008296217,0.3232272,0.2657227,0.080567874,0.006760271,0.30397362],"study_design_scores_gemma":[0.000107676795,0.00020736609,0.0021544246,0.00004811069,0.000042901596,0.00062084146,0.00014722193,0.80437595,0.09976302,0.08753801,0.004937727,0.000056836354],"about_ca_topic_score_codex":0.0007512119,"about_ca_topic_score_gemma":0.0014541057,"teacher_disagreement_score":0.0021127297,"about_ca_system_score_codex":0.0007624261,"about_ca_system_score_gemma":0.001139249,"threshold_uncertainty_score":0.00706774},"labels":[],"label_agreement":null},{"id":"W2134922741","doi":"10.1142/s0219720012710023","title":"DETECTION AND DECOMPOSITION: TREATMENT-INDUCED CYCLIC GENE EXPRESSION DISRUPTION IN HIGH-THROUGHPUT TIME-SERIES DATASETS","year":2012,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Evolutionary Algorithms and Applications","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":"Lakehead University","funders":"","keywords":"Gene; Gene expression; Organism; Decomposition; Computational biology; Biology; Expression (computer science); Series (stratigraphy); Computer science; Genetics; Ecology","score_opus":0.012441089639826432,"score_gpt":0.27673103492857876,"score_spread":0.2642899452887523,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2134922741","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":[0.678147,0.0026543548,0.2895722,0.0009693151,0.0003315684,0.00065508665,0.020560702,0.0060885125,0.0010212722],"genre_scores_gemma":[0.66824627,0.000812136,0.26199293,0.00030929636,0.0001425903,0.0009932233,0.06646763,0.0002707969,0.0007651626],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9982666,0.0005008699,0.00021138822,0.00056308025,0.00031607214,0.00014201275],"domain_scores_gemma":[0.997279,0.0014294374,0.00039378117,0.000536928,0.00025450138,0.00010641886],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0034395321,0.00096966745,0.0009918247,0.0020768286,0.0004664146,0.00074293395,0.00094801636,0.0011820737,0.00043499132],"category_scores_gemma":[0.0071334783,0.0002662303,0.0015046829,0.0024069324,0.0005682706,0.0007693656,0.00075088406,0.0012592882,0.00038557829],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0031748135,0.0020322322,0.15843135,0.0028907198,0.00203909,0.0018650537,0.00055314787,0.119110994,0.18299966,0.003797561,0.031611018,0.4914944],"study_design_scores_gemma":[0.00019292391,0.00065842405,0.11764348,0.000073979456,0.00036161198,0.0013214911,0.00022373162,0.8155354,0.046074364,0.0068608127,0.010918612,0.0001350981],"about_ca_topic_score_codex":0.0019634268,"about_ca_topic_score_gemma":0.0027153238,"teacher_disagreement_score":0.0034395321,"about_ca_system_score_codex":0.00052783045,"about_ca_system_score_gemma":0.0008986224,"threshold_uncertainty_score":0.018190205},"labels":[],"label_agreement":null},{"id":"W2135899964","doi":"10.1142/s0219720014500255","title":"Evaluating predictive performance of network biomarkers with network structures","year":2014,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","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":"University of Calgary","funders":"","keywords":"Social connectedness; Computer science; Data mining; Graph; Weighted network; Network performance; Network analysis; Support vector machine; Network structure; Gene regulatory network; Artificial intelligence; Machine learning; Complex network; Theoretical computer science; Biology; Gene; Gene expression","score_opus":0.008379428439424798,"score_gpt":0.25114489197362383,"score_spread":0.24276546353419903,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2135899964","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":[0.864374,0.0019574189,0.12714262,0.0009318907,0.00008871304,0.00014963368,0.0023087072,0.0010667725,0.0019802458],"genre_scores_gemma":[0.9763648,0.0003293641,0.020823933,0.000038941278,0.00005016049,0.0000668098,0.0019471262,0.00002909913,0.00034979536],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99854505,0.0006765284,0.00009045601,0.0003118366,0.00027941764,0.00009675109],"domain_scores_gemma":[0.984338,0.01255576,0.0012952914,0.0008560937,0.00073032704,0.00022452352],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005302463,0.0012551007,0.0007274218,0.003798261,0.00035996974,0.0010276742,0.00070644094,0.0010469726,0.0006667696],"category_scores_gemma":[0.021487916,0.00023908934,0.0006895478,0.0021604144,0.0005072387,0.0014415028,0.00084937515,0.0009020405,0.0001874055],"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":[0.0005701771,0.00018739807,0.10547045,0.00014700896,0.0004266167,0.00015023348,0.00007186029,0.81516343,0.0023434046,0.0013425826,0.0013147583,0.072812],"study_design_scores_gemma":[0.000012808493,0.0001368597,0.008758043,0.000015797263,0.000050729253,0.00005516574,0.000023553752,0.9875901,0.0012760687,0.0018652624,0.00020418795,0.000011390545],"about_ca_topic_score_codex":0.0048425547,"about_ca_topic_score_gemma":0.0042776926,"teacher_disagreement_score":0.005302463,"about_ca_system_score_codex":0.0008005324,"about_ca_system_score_gemma":0.0007282083,"threshold_uncertainty_score":0.028042436},"labels":[],"label_agreement":null},{"id":"W2137504100","doi":"10.1142/s0219720011005276","title":"ERROR TOLERANT NMR BACKBONE RESONANCE ASSIGNMENT AND AUTOMATED STRUCTURE GENERATION","year":2010,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Protein Structure and 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":"York University; University of Waterloo","funders":"","keywords":"Probabilistic logic; Computer science; Precision and recall; Protein secondary structure; Algorithm; Artificial intelligence; Biology","score_opus":0.006003271850188033,"score_gpt":0.2457398115828957,"score_spread":0.23973653973270767,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2137504100","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.020167196,0.00006194744,0.9659593,0.000101774305,0.000026979314,0.000078645986,0.00035035453,0.012540655,0.00071299507],"genre_scores_gemma":[0.10410949,0.00008642307,0.891758,0.0000760499,0.000019707708,0.00020149596,0.0014856155,0.0012366866,0.0010265661],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9976974,0.0007272478,0.00013077463,0.0005813958,0.000724139,0.00013886056],"domain_scores_gemma":[0.99665326,0.0012870028,0.0004242005,0.0007178654,0.0008406956,0.00007705615],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0028291068,0.0014074637,0.0011249937,0.0013018375,0.00076094316,0.0012945288,0.0022246852,0.0009838411,0.0045250757],"category_scores_gemma":[0.0074163266,0.0008802733,0.0008250315,0.0014093837,0.000647373,0.0017280311,0.001507343,0.001513153,0.0021375725],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010430795,0.0002546812,0.004019055,0.00056503434,0.00016524018,0.0005439487,0.00040317402,0.25532836,0.17429632,0.01655991,0.015399826,0.5314215],"study_design_scores_gemma":[0.00005179273,0.00011332159,0.0006999685,0.00001714613,0.000022681486,0.00016918863,0.00005830537,0.9040855,0.07948809,0.008686465,0.0065606,0.000047002555],"about_ca_topic_score_codex":0.0018276541,"about_ca_topic_score_gemma":0.0018534863,"teacher_disagreement_score":0.0045250757,"about_ca_system_score_codex":0.00074421614,"about_ca_system_score_gemma":0.0015463362,"threshold_uncertainty_score":0.015137851},"labels":[],"label_agreement":null},{"id":"W2140934347","doi":"10.1142/s0219720013500029","title":"WHOLE GENOME IDENTITY-BY-DESCENT DETERMINATION","year":2012,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Genetic Associations and Epidemiology","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 Alberta","funders":"Alberta Agricultural Research Institute","keywords":"Haplotype; Identity by descent; Haplotype estimation; Genetics; Genotyping; Genome; Biology; Linkage (software); Mendelian inheritance; Allele; Genotype; Computational biology; Single-nucleotide polymorphism; Gene","score_opus":0.010675163598570198,"score_gpt":0.2744912583305859,"score_spread":0.2638160947320157,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2140934347","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":[0.1269533,0.00015704539,0.8662523,0.00014936608,0.000032969274,0.00014529008,0.0018952735,0.0015239518,0.0028904884],"genre_scores_gemma":[0.22935246,0.00013330848,0.7621711,0.00008002839,0.000008560851,0.000161142,0.004006275,0.0002580919,0.0038290531],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9996947,0.000079873884,0.000017428772,0.00011388813,0.000054361724,0.000039843108],"domain_scores_gemma":[0.99944156,0.00022454814,0.000057126275,0.00013440991,0.00010991549,0.000032474847],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005746547,0.00044615183,0.0006773091,0.0007232219,0.00058318843,0.0005952485,0.0008008158,0.00045482352,0.0046893656],"category_scores_gemma":[0.0026044783,0.00032741693,0.00062009954,0.0007832768,0.00023381192,0.00042663427,0.0006536461,0.00046834594,0.0012075262],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006601605,0.00033050313,0.056180865,0.00031825883,0.00021200969,0.0006216722,0.0005342524,0.24748178,0.05868869,0.047155607,0.017875224,0.56994104],"study_design_scores_gemma":[0.00012139453,0.00011426844,0.012159416,0.000021341346,0.000056911722,0.00059335504,0.00013394313,0.9261061,0.01654269,0.028483955,0.015635276,0.00003141735],"about_ca_topic_score_codex":0.004019234,"about_ca_topic_score_gemma":0.0056732856,"teacher_disagreement_score":0.0046893656,"about_ca_system_score_codex":0.00041355335,"about_ca_system_score_gemma":0.0012324278,"threshold_uncertainty_score":0.015687466},"labels":[],"label_agreement":null},{"id":"W2141375003","doi":"10.1142/s0219720004000557","title":"ANCESTRAL MAXIMUM LIKELIHOOD OF EVOLUTIONARY TREES IS HARD","year":2004,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Evolution and Genetic Dynamics","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":"Memorial University of Newfoundland; McGill University","funders":"Israel Science Foundation; McGill University","keywords":"Maximum likelihood; Evolutionary biology; Biology; Phylogenetics; Tree rearrangement; Phylogenetic tree; Tree (set theory); Mathematics; Statistics; Combinatorics; Genetics; Gene","score_opus":0.007763901412861236,"score_gpt":0.24548396799682093,"score_spread":0.2377200665839597,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2141375003","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":[0.186122,0.0030108572,0.72561914,0.028736029,0.00033064312,0.00032806254,0.006058575,0.0019460573,0.04784864],"genre_scores_gemma":[0.80118376,0.0027319228,0.16804232,0.0040311236,0.0011707986,0.00054104114,0.0071717463,0.00081349583,0.014313726],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9966074,0.00088337844,0.00019190587,0.0011443815,0.00075193826,0.00042102087],"domain_scores_gemma":[0.9575303,0.03804238,0.0010695746,0.0018168164,0.0009650577,0.00057580875],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002490656,0.00080939604,0.0019616147,0.00083139667,0.0018857581,0.0047235894,0.0025411134,0.0032831703,0.008808886],"category_scores_gemma":[0.025954569,0.0012522641,0.0023688166,0.001823722,0.0026430113,0.009830246,0.0038168575,0.006352909,0.0015330235],"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":[0.0008122606,0.0003368313,0.006758002,0.0021877957,0.0003258822,0.0017708684,0.0018930832,0.23534115,0.0052353917,0.5654528,0.053481136,0.12640475],"study_design_scores_gemma":[0.000095941396,0.000023968474,0.00057350093,0.000066032626,0.000036396497,0.00062196975,0.0001977784,0.13631125,0.0015640599,0.8518083,0.008675903,0.000024886715],"about_ca_topic_score_codex":0.0011307467,"about_ca_topic_score_gemma":0.0011060523,"teacher_disagreement_score":0.008808886,"about_ca_system_score_codex":0.0017023489,"about_ca_system_score_gemma":0.0016215885,"threshold_uncertainty_score":0.029468656},"labels":[],"label_agreement":null},{"id":"W2148088562","doi":"10.1142/s0219720010005002","title":"A GRAPH-BASED ALGORITHM FOR MINING MULTI-LEVEL PATTERNS IN GENOMIC DATA","year":2010,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","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":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo; University of Guelph","funders":"","keywords":"Genome; Comparative genomics; Genomics; Computer science; Graph; Computational biology; Data mining; Gene; Theoretical computer science; Biology; Genetics","score_opus":0.039847663363638125,"score_gpt":0.2978313042449222,"score_spread":0.2579836408812841,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2148088562","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":[0.003522482,0.0003069291,0.9921463,0.00018484396,0.000040137435,0.0002517978,0.00046228006,0.0027459513,0.0003392161],"genre_scores_gemma":[0.018430255,0.00017157382,0.9792509,0.000100664416,0.000015838297,0.00026980307,0.0012267337,0.00010512541,0.0004290878],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99755186,0.00048643522,0.00031003854,0.0008476449,0.00066960027,0.00013442202],"domain_scores_gemma":[0.99692696,0.0018497859,0.0002961073,0.00033638353,0.0005060738,0.000084730564],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0022334526,0.002059942,0.0017173354,0.0082519185,0.0013409933,0.0019124983,0.0033885047,0.002210854,0.0026853508],"category_scores_gemma":[0.007927157,0.0009234639,0.002456253,0.009146586,0.00104108,0.0040412527,0.0017427142,0.002198603,0.0014285246],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00023061773,0.0003331239,0.0046789474,0.0006080999,0.00039081185,0.0003182447,0.0002899672,0.12579694,0.008672174,0.014312957,0.012251258,0.83211684],"study_design_scores_gemma":[0.000083684055,0.00012557399,0.00080253225,0.00006002182,0.000066256514,0.00044905275,0.00012205395,0.9536445,0.0039921,0.030760398,0.009850899,0.000043007447],"about_ca_topic_score_codex":0.0071218433,"about_ca_topic_score_gemma":0.009653553,"teacher_disagreement_score":0.0082519185,"about_ca_system_score_codex":0.001224144,"about_ca_system_score_gemma":0.0024224722,"threshold_uncertainty_score":0.014160752},"labels":[],"label_agreement":null},{"id":"W2160672912","doi":"10.1142/s0219720003000186","title":"RAPTOR: OPTIMAL PROTEIN THREADING BY LINEAR PROGRAMMING","year":2003,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Protein Structure and Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":294,"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":"Threading (protein sequence); Linear programming; Computer science; Pairwise comparison; Integer programming; Protein structure prediction; Server; Benchmark (surveying); Algorithm; Theoretical computer science; Mathematical optimization; Protein structure; Artificial intelligence; Mathematics; Biology; Computer network","score_opus":0.0053852942881868725,"score_gpt":0.24188663847241743,"score_spread":0.23650134418423055,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2160672912","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":[0.0060342485,0.00025874464,0.9881334,0.00023323324,0.0000452459,0.00005423913,0.00007130314,0.0021950423,0.002974596],"genre_scores_gemma":[0.1117465,0.0003258982,0.8816873,0.00027467628,0.00008781006,0.00037814662,0.00037719758,0.00087649305,0.004246027],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99887305,0.00042836595,0.000039376024,0.00019609385,0.000326645,0.00013643445],"domain_scores_gemma":[0.9986939,0.00089309696,0.00010775677,0.000103485276,0.00012893329,0.000072892064],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0017620978,0.001222624,0.0015036666,0.00072760927,0.00062732166,0.0011314852,0.0018082596,0.0011237641,0.005343844],"category_scores_gemma":[0.003394071,0.00089033577,0.0013451017,0.00084998115,0.00092196144,0.0015342569,0.0015551146,0.002358018,0.0017946203],"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":[0.000182637,0.0001565231,0.00045115323,0.00019074047,0.00006534307,0.00009346699,0.00010148406,0.81918424,0.0034362571,0.038606055,0.010205889,0.12732609],"study_design_scores_gemma":[0.000019958836,0.000029072218,0.00001746421,0.0000047641615,0.0000037987697,0.000013251893,0.000005302898,0.9911737,0.00045540615,0.0070184823,0.0012539965,0.000004783802],"about_ca_topic_score_codex":0.0029718957,"about_ca_topic_score_gemma":0.0030392755,"teacher_disagreement_score":0.005343844,"about_ca_system_score_codex":0.0009440032,"about_ca_system_score_gemma":0.0018988422,"threshold_uncertainty_score":0.017876983},"labels":[],"label_agreement":null},{"id":"W2167445572","doi":"10.1142/s0219720014500279","title":"IncMD: Incremental trie-based structural motif discovery algorithm","year":2014,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","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 Ottawa","funders":"","keywords":"Structural motif; Motif (music); Scalability; Computer science; Nucleic acid secondary structure; Trie; Sequence motif; Computational biology; A priori and a posteriori; Algorithm; Data structure; Theoretical computer science; RNA; Biology; Genetics","score_opus":0.006031222069649855,"score_gpt":0.23177202764777677,"score_spread":0.22574080557812692,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2167445572","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":[0.027216945,0.0006240863,0.9613871,0.00034826715,0.00007781451,0.0003200918,0.0011531094,0.0073456513,0.0015269838],"genre_scores_gemma":[0.08264842,0.00015933509,0.91101944,0.00023032406,0.000042771317,0.00042766245,0.0035165874,0.00020951593,0.0017458964],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9989963,0.00014400313,0.00010818868,0.00026282266,0.00038140317,0.00010733509],"domain_scores_gemma":[0.9980927,0.000983466,0.00016660061,0.00020814517,0.00044452006,0.000104533785],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00094683707,0.00087029696,0.0015156816,0.0038230505,0.00079098303,0.0011064059,0.003343614,0.001484986,0.0045914743],"category_scores_gemma":[0.0044023045,0.00076731545,0.0013666989,0.0028459257,0.0005357946,0.002061566,0.0017238682,0.0016867627,0.0017536668],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000931176,0.00058945594,0.0063066036,0.0004384042,0.00027544875,0.00053230824,0.00027704475,0.12033771,0.0171432,0.012898777,0.022238743,0.81803113],"study_design_scores_gemma":[0.00014894364,0.00016372572,0.0005107958,0.00001967053,0.00004119372,0.0004744067,0.00004565325,0.97937095,0.0049615814,0.008899606,0.0053376323,0.000025771074],"about_ca_topic_score_codex":0.0028195423,"about_ca_topic_score_gemma":0.004713781,"teacher_disagreement_score":0.0045914743,"about_ca_system_score_codex":0.0008719907,"about_ca_system_score_gemma":0.002201334,"threshold_uncertainty_score":0.015359998},"labels":[],"label_agreement":null},{"id":"W2168695474","doi":"10.1142/s0219720008003254","title":"THE DISTRIBUTION OF DISTANCES BETWEEN RANDOMLY CONSTRUCTED GENOMES: GENERATING FUNCTION, EXPECTATION, VARIANCE AND LIMITS","year":2008,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","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":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Ottawa","funders":"","keywords":"Genome; Combinatorics; Limit (mathematics); Biology; Graph; Breakpoint; Mathematics; Variance (accounting); Function (biology); Genetics; Chromosome; Gene; Evolutionary biology","score_opus":0.010180695695394211,"score_gpt":0.22572594971298096,"score_spread":0.21554525401758676,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2168695474","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":[0.19610606,0.0013487274,0.7981347,0.001071987,0.00004647085,0.000111594454,0.00032533743,0.00045686663,0.002398292],"genre_scores_gemma":[0.87994313,0.0017414533,0.112891495,0.00026760853,0.00019292736,0.0005309671,0.0011852778,0.0002842992,0.0029628389],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9965048,0.00176147,0.00014057336,0.00074231625,0.000636314,0.00021453897],"domain_scores_gemma":[0.9122414,0.07871269,0.0031038036,0.0031860045,0.0020215877,0.00073453225],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.012845529,0.0007937851,0.0014649391,0.0030604894,0.0007371482,0.002429989,0.0030260044,0.0019756067,0.0017807956],"category_scores_gemma":[0.07058723,0.0008354655,0.001007308,0.0018071515,0.004589389,0.005369888,0.0020625247,0.002357539,0.00042900947],"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":[0.00021642889,0.00013083631,0.019183973,0.00024585726,0.00012425834,0.0005158975,0.0006090146,0.40777323,0.0034664704,0.5311594,0.0018159376,0.03475868],"study_design_scores_gemma":[0.000027008096,0.000047986006,0.0023576345,0.000050307226,0.000018886054,0.00040335095,0.00007464025,0.85386527,0.0013599991,0.14108631,0.00065325724,0.00005532381],"about_ca_topic_score_codex":0.0014343269,"about_ca_topic_score_gemma":0.0010029682,"teacher_disagreement_score":0.012845529,"about_ca_system_score_codex":0.002306019,"about_ca_system_score_gemma":0.00091928744,"threshold_uncertainty_score":0.06793445},"labels":[],"label_agreement":null},{"id":"W2170694711","doi":"10.1142/s021972000900445x","title":"A PROBABILISTIC FRAMEWORK TO IMPROVE MICRORNA TARGET PREDICTION BY INCORPORATING PROTEOMICS DATA","year":2009,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"MicroRNA in disease regulation","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 Toronto","funders":"Canadian Institutes of Health Research","keywords":"Complementarity (molecular biology); microRNA; Computational biology; Messenger RNA; Biology; Probabilistic logic; Proteomics; Genetics; Computer science; Gene; Artificial intelligence","score_opus":0.008801695579537777,"score_gpt":0.255764492977427,"score_spread":0.2469627973978892,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2170694711","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":[0.008326817,0.00019760309,0.9903509,0.00014355037,0.000025334715,0.000024340563,0.00008526291,0.00055228,0.00029387846],"genre_scores_gemma":[0.40362176,0.0005577061,0.59208775,0.00027955853,0.00018421441,0.00038402018,0.0006751827,0.00022045978,0.0019893334],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99915373,0.0002943402,0.00006010879,0.00019549944,0.00022040268,0.0000758886],"domain_scores_gemma":[0.9978259,0.0015265598,0.0001718425,0.00010886638,0.00029859887,0.0000682295],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0029334857,0.001103334,0.0013512309,0.001110534,0.000571292,0.0011546231,0.001997592,0.0012583662,0.0012263112],"category_scores_gemma":[0.005814417,0.0011009948,0.0014607322,0.0009487234,0.00060593156,0.0013217485,0.0011152013,0.0011749336,0.0005026773],"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":[0.00008947671,0.00008235838,0.0008077025,0.0000636766,0.000109436536,0.00007821226,0.000024869545,0.9558827,0.0023606473,0.0059043122,0.00060458307,0.033992056],"study_design_scores_gemma":[0.0000049905793,0.000010597712,0.000046978537,0.0000014262173,0.0000074155673,0.0000084269495,6.937239e-7,0.9977792,0.00020518854,0.0018143374,0.00011541503,0.0000053419767],"about_ca_topic_score_codex":0.008214711,"about_ca_topic_score_gemma":0.008668643,"teacher_disagreement_score":0.008214711,"about_ca_system_score_codex":0.00087763055,"about_ca_system_score_gemma":0.0015661683,"threshold_uncertainty_score":0.016333818},"labels":[],"label_agreement":null},{"id":"W2212557680","doi":"10.1142/s0219720016500049","title":"Predicting GHS toxicity using RTCA and discrete-time Fourier transform","year":2015,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Computational Drug Discovery Methods","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":"Alberta Health; University of Calgary; University of Alberta","funders":"","keywords":"Cytotoxicity; Toxicity; In vivo; Acute toxicity; Computer science; Biological system; Pharmacology; Toxicology; Biology; Chemistry; In vitro; Biochemistry; Biotechnology","score_opus":0.02996827221963222,"score_gpt":0.3055107492716636,"score_spread":0.2755424770520314,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2212557680","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.12292479,0.0008650052,0.87079877,0.00025133695,0.0000916291,0.00009482758,0.00037345992,0.0028861319,0.0017140305],"genre_scores_gemma":[0.7010769,0.0012512757,0.29342672,0.0001008141,0.00009609122,0.00016244037,0.0007589301,0.00020005788,0.0029267191],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99982965,0.00003707319,0.00001328462,0.0000397575,0.00006191952,0.00001830369],"domain_scores_gemma":[0.99944144,0.00032404694,0.0000684489,0.00003340756,0.0001115903,0.000020932172],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006067658,0.000755293,0.0006626601,0.0014377164,0.00023681491,0.0006858566,0.00038452237,0.00064573257,0.001619167],"category_scores_gemma":[0.0017147878,0.00019781003,0.00097357953,0.0007549292,0.00032072363,0.0004440054,0.00028007184,0.0007406376,0.0005368891],"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":[0.00027467057,0.00013723932,0.0052201864,0.00016506847,0.000102482205,0.00023937174,0.000045324792,0.7576135,0.029121276,0.0016534711,0.0016285422,0.20379879],"study_design_scores_gemma":[0.0000040780487,0.000025867139,0.00058134587,0.00000338797,0.000009847964,0.00003548015,0.0000070804826,0.9963039,0.0023223474,0.00041914903,0.0002812452,0.000006250417],"about_ca_topic_score_codex":0.008608427,"about_ca_topic_score_gemma":0.005431694,"teacher_disagreement_score":0.008608427,"about_ca_system_score_codex":0.00044599173,"about_ca_system_score_gemma":0.0009338404,"threshold_uncertainty_score":0.017116666},"labels":[],"label_agreement":null},{"id":"W2294194070","doi":"10.1142/s0219720016410055","title":"Sequential construction of a model for modular gene expression control, applied to spatial patterning of the<i>Drosophila</i>gene<i>hunchback</i>","year":2016,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Gene Regulatory Network Analysis","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 Victoria; British Columbia Institute of Technology","funders":"National Institute of General Medical Sciences; National Institutes of Health; Russian Foundation for Basic Research","keywords":"Gene regulatory network; Gene; Biology; Computational biology; Cis-regulatory module; Regulation of gene expression; Genetics; Gene expression; Regulator gene; Regulatory sequence; Transcription factor; Translation (biology); Regulator; Gap gene; Transcription (linguistics); Messenger RNA; Promoter","score_opus":0.008403906487245838,"score_gpt":0.22243375229412654,"score_spread":0.2140298458068807,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2294194070","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":[0.07696465,0.00016125143,0.91413087,0.00041773298,0.0000509875,0.00009680665,0.000274066,0.00023764692,0.0076659834],"genre_scores_gemma":[0.8096321,0.00035407953,0.17920022,0.00010792132,0.0000328198,0.0007630559,0.0003838427,0.00016629054,0.009359585],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99982566,0.00007770714,0.0000067896117,0.00003856097,0.00003315384,0.00001814601],"domain_scores_gemma":[0.9994186,0.00039771164,0.000061969484,0.00002070218,0.000060211827,0.000040889667],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006538716,0.000606516,0.0005991868,0.00037915513,0.00058372563,0.00058053946,0.0008109713,0.0011286056,0.0027593775],"category_scores_gemma":[0.0015151434,0.00044847376,0.00079225854,0.00048694393,0.0010400653,0.00045575792,0.00076063874,0.0007929036,0.00025642905],"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":[0.000010941535,0.0000056146505,0.00011020922,0.000009519861,0.0000034358875,0.00002324276,0.0000088187435,0.9928774,0.00049668987,0.005650946,0.000071005394,0.0007322826],"study_design_scores_gemma":[0.0000029098326,0.000003399137,0.000016700977,7.257805e-7,0.0000011445942,0.0000019439867,0.0000012463937,0.99884355,0.000072395764,0.00096894207,0.000086269654,8.421305e-7],"about_ca_topic_score_codex":0.012160452,"about_ca_topic_score_gemma":0.008412197,"teacher_disagreement_score":0.012160452,"about_ca_system_score_codex":0.0013578391,"about_ca_system_score_gemma":0.0015047917,"threshold_uncertainty_score":0.02417934},"labels":[],"label_agreement":null},{"id":"W2301989113","doi":"10.1142/s0219720016500153","title":"Gene-set activity toolbox (GAT): A platform for microarray-based cancer diagnosis using an integrative gene-set analysis approach","year":2016,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Gene expression and cancer classification","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":"King Mongkut's University of Technology Thonburi; Mahidol University; Prince of Songkla University; Thailand Research Fund; University of Toronto; King Mongkut's Institute of Technology Ladkrabang; University of Notre Dame; National Center for Genetic Engineering and Biotechnology; Higher Education Research Promotion","keywords":"Toolbox; Computational biology; Microarray analysis techniques; Computer science; Gene expression profiling; Data mining; Data set; Microarray databases; Set (abstract data type); Gene; Bioinformatics; Gene expression; Biology; Artificial intelligence; Genetics","score_opus":0.042005073215708566,"score_gpt":0.32428068830936896,"score_spread":0.2822756150936604,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2301989113","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.013134066,0.00051829143,0.54787046,0.00022254404,0.00015294409,0.00034825772,0.055849064,0.37888324,0.003021163],"genre_scores_gemma":[0.09969455,0.0008774274,0.7642211,0.0004886342,0.000080823345,0.0034633232,0.106053054,0.020706046,0.004415047],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99938965,0.00012488569,0.00006415384,0.00015600651,0.00021041695,0.00005499088],"domain_scores_gemma":[0.99890506,0.00059281127,0.00013449453,0.00016799012,0.00013055687,0.00006912292],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014101282,0.0020020667,0.001169034,0.0028121395,0.000365784,0.0010012229,0.0017100386,0.00068658555,0.015514919],"category_scores_gemma":[0.0026512444,0.0006747259,0.0016514733,0.0019746218,0.00034678497,0.00068080577,0.0012157914,0.0012958371,0.0072582546],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.003108725,0.0006581318,0.022903409,0.0042961203,0.0013120882,0.0014244227,0.0008376201,0.05786624,0.17246874,0.012172049,0.3217926,0.40115994],"study_design_scores_gemma":[0.00089344825,0.0008129185,0.031806394,0.00037753256,0.0004531616,0.002261896,0.0002707734,0.5441735,0.16345766,0.036540877,0.21853374,0.00041807283],"about_ca_topic_score_codex":0.0012182054,"about_ca_topic_score_gemma":0.00160961,"teacher_disagreement_score":0.015514919,"about_ca_system_score_codex":0.00053068047,"about_ca_system_score_gemma":0.00093607145,"threshold_uncertainty_score":0.051902533},"labels":[],"label_agreement":null},{"id":"W2344448545","doi":"10.1142/s0219720016500190","title":"A novel Arg H52/Tyr H33 conservative motif in antibodies: A correlation between sequence of antibodies and antigen binding","year":2016,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Monoclonal and Polyclonal Antibodies Research","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":"McGill University","keywords":"Sequence motif; Antibody; Antigen; Sequence (biology); Motif (music); Molecular biology; Computational biology; Biology; Chemistry; Genetics; Physics; Gene","score_opus":0.05648765464284376,"score_gpt":0.3410180651211903,"score_spread":0.28453041047834654,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2344448545","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":[0.9709853,0.00032231378,0.027273472,0.00006985857,0.000015911295,0.000017207067,0.000082510545,0.00011062918,0.0011229137],"genre_scores_gemma":[0.9915643,0.000049791306,0.008073955,0.00002782214,0.000003352793,0.0000048994475,0.00006462755,0.000007925808,0.00020337776],"study_design_codex":"bench_or_experimental","study_design_gemma":"observational","domain_scores_codex":[0.9998919,0.000020807865,0.0000071891545,0.000032202905,0.000030915966,0.000017085033],"domain_scores_gemma":[0.9998286,0.000047354108,0.00006090457,0.00001276074,0.000023021395,0.000027367798],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00017930471,0.00014781211,0.00019975465,0.00021473842,0.00016640635,0.00022112418,0.00019228963,0.00025463762,0.000736431],"category_scores_gemma":[0.00044102268,0.0000943204,0.00020244364,0.000291752,0.000186038,0.00022653218,0.00015292545,0.00024856615,0.00013587844],"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":[0.0007231686,0.00021000957,0.04829989,0.00026138782,0.000090366906,0.0025167635,0.00013419108,0.030551784,0.8775593,0.007037788,0.00072274177,0.031892553],"study_design_scores_gemma":[0.00016213462,0.0012869261,0.11505062,0.00004578139,0.00024370878,0.005380659,0.0001587193,0.5309429,0.33070466,0.009249486,0.00667347,0.000100900055],"about_ca_topic_score_codex":0.0006692749,"about_ca_topic_score_gemma":0.0009133792,"teacher_disagreement_score":0.000736431,"about_ca_system_score_codex":0.00014147515,"about_ca_system_score_gemma":0.00022542775,"threshold_uncertainty_score":0.0024636388},"labels":[],"label_agreement":null},{"id":"W2745681143","doi":"10.1142/s0219720017500196","title":"Systematic investigations of gene effects on both topologies and supports: An <i>Echinococcus</i> illustration","year":2017,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","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":false,"ca_institutions":"Université de Sherbrooke","funders":"","keywords":"Phylogenetics; Network topology; Phylogenetic tree; Biology; Computational biology; Nuclear gene; Coalescent theory; Tree (set theory); Evolutionary biology; Genome; Echinococcus; Computer science; Gene; Mathematics; Genetics; Zoology","score_opus":0.013017846059582075,"score_gpt":0.2589808945093313,"score_spread":0.24596304844974926,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2745681143","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":[0.9184895,0.0009069653,0.07795477,0.00023274918,0.000022110737,0.000020035293,0.00095777266,0.00034826485,0.0010678212],"genre_scores_gemma":[0.9628373,0.00031561663,0.03550463,0.000072886156,0.00001320817,0.000021274853,0.0009095521,0.000116280404,0.00020937542],"study_design_codex":"bench_or_experimental","study_design_gemma":"observational","domain_scores_codex":[0.9994672,0.00027091795,0.000028983915,0.00013434992,0.00006057248,0.000038029448],"domain_scores_gemma":[0.9932587,0.0052653905,0.0004592056,0.00054933014,0.0002885354,0.00017888325],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012557153,0.00049950124,0.00049146294,0.0011327923,0.0003549796,0.00048242512,0.00033171286,0.0002756702,0.0016007136],"category_scores_gemma":[0.004225279,0.00011581659,0.00053882494,0.0011442907,0.00055977365,0.00044506224,0.000499506,0.0006228387,0.0001770285],"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":[0.0016952098,0.00033097193,0.21493234,0.000870464,0.0012103559,0.003755386,0.0009333864,0.15260993,0.4213714,0.014517911,0.0032221752,0.18455042],"study_design_scores_gemma":[0.000093909206,0.00076699123,0.31879964,0.00011960295,0.0009878868,0.0024914287,0.0010060159,0.47206435,0.15516008,0.037327513,0.01096472,0.00021781554],"about_ca_topic_score_codex":0.00078474596,"about_ca_topic_score_gemma":0.00080619,"teacher_disagreement_score":0.0016007136,"about_ca_system_score_codex":0.00016319123,"about_ca_system_score_gemma":0.00023994822,"threshold_uncertainty_score":0.0066409707},"labels":[],"label_agreement":null},{"id":"W2765700971","doi":"10.1142/s0219720017400078","title":"Reconstructing protein and gene phylogenies using reconciliation and soft-clustering","year":2017,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","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 Ottawa; Université de Sherbrooke","funders":"","keywords":"Supertree; Phylogenetic tree; Ensembl; Tree (set theory); Gene; Computational biology; Cluster analysis; Biology; Heuristic; Computer science; Phylogenetics; Genetics; Genome; Genomics; Artificial intelligence; Mathematics; Combinatorics","score_opus":0.021867856078774235,"score_gpt":0.2618437589356458,"score_spread":0.2399759028568716,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2765700971","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.076852396,0.00083605066,0.91647947,0.00055813254,0.00006862947,0.00015035039,0.00059444754,0.0030350245,0.0014255233],"genre_scores_gemma":[0.21922909,0.00040900378,0.77350134,0.00023384325,0.000064995525,0.00026472137,0.0039163674,0.001263348,0.0011172695],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9959072,0.0015433222,0.0002314255,0.0012280131,0.00079518044,0.00029485387],"domain_scores_gemma":[0.9918013,0.003715252,0.0009551303,0.0022788933,0.00080589,0.00044354374],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005508164,0.0023455368,0.0028141448,0.006960697,0.002778572,0.0036901645,0.0048662093,0.004000738,0.0025828837],"category_scores_gemma":[0.016999042,0.0019484413,0.0040816874,0.0052676066,0.0028561444,0.005799972,0.0048198057,0.003993619,0.0017500869],"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":[0.0006843802,0.0002699957,0.0073799207,0.0006342068,0.0005370374,0.00050039776,0.0008950346,0.7345141,0.011895872,0.054305322,0.005533819,0.18284996],"study_design_scores_gemma":[0.000041160518,0.000048319587,0.0010011636,0.000037288202,0.000041910538,0.00013916816,0.00016098989,0.91089433,0.0029364987,0.08248602,0.0021531389,0.000060049024],"about_ca_topic_score_codex":0.0039883344,"about_ca_topic_score_gemma":0.0047432277,"teacher_disagreement_score":0.006960697,"about_ca_system_score_codex":0.002583654,"about_ca_system_score_gemma":0.0023016338,"threshold_uncertainty_score":0.02913034},"labels":[],"label_agreement":null},{"id":"W2899162278","doi":"10.1142/s0219720018400267","title":"HisCoM-GGI: Hierarchical structural component analysis of gene–gene interactions","year":2018,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","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":"McGill University","funders":"","keywords":"Missing heritability problem; Single-nucleotide polymorphism; Biology; Genome-wide association study; Genetics; Computational biology; Heritability; Genetic association; Gene; SNP; Population; Pairwise comparison; Genotype; Computer science; Medicine","score_opus":0.01236623746561531,"score_gpt":0.2945866825847948,"score_spread":0.2822204451191795,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2899162278","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.039528277,0.000895518,0.9541423,0.00040778625,0.00012469353,0.00029592536,0.0012659498,0.0020576082,0.0012819608],"genre_scores_gemma":[0.46161184,0.001231006,0.52496743,0.00064410665,0.00023276363,0.0010909606,0.0061356323,0.000711814,0.003374522],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9981029,0.0008362269,0.000094673276,0.00043642236,0.00034733803,0.00018235113],"domain_scores_gemma":[0.9974016,0.0015343202,0.00021517239,0.00032880847,0.000394964,0.00012516565],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0025469058,0.0022174728,0.0013789106,0.002355643,0.00093049067,0.0011501188,0.001534229,0.00083850225,0.002507534],"category_scores_gemma":[0.008545179,0.00048429132,0.0036615126,0.0025731823,0.0008023814,0.00084302464,0.0017388941,0.0020732149,0.00065770943],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007258406,0.00041195564,0.074027725,0.00081191125,0.0033040112,0.0007355693,0.00069800747,0.2769955,0.014328387,0.076955356,0.02289723,0.5281085],"study_design_scores_gemma":[0.00009101422,0.00015027303,0.013941679,0.000044568544,0.00041585084,0.00020567095,0.000081449885,0.9347179,0.0026626424,0.042083956,0.005494833,0.00011014968],"about_ca_topic_score_codex":0.017291568,"about_ca_topic_score_gemma":0.020307897,"teacher_disagreement_score":0.017291568,"about_ca_system_score_codex":0.00088134705,"about_ca_system_score_gemma":0.002436401,"threshold_uncertainty_score":0.034381866},"labels":[],"label_agreement":null},{"id":"W2945564624","doi":"10.1142/s0219720019500173","title":"DSRIG: Incorporating graphical structure in the regularized modeling of SNP data","year":2019,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Genetic and phenotypic traits in livestock","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 Guelph","funders":"Agricultural Adaptation Council; Natural Sciences and Engineering Research Council of Canada; Ontario Ministry of Agriculture, Food and Rural Affairs","keywords":"Single-nucleotide polymorphism; Boar taint; Selection (genetic algorithm); Lasso (programming language); Candidate gene; Androstenone; SNP; Trait; Regression; Computer science; Biology; Statistics; Mathematics; Computational biology; Genetics; Artificial intelligence; Gene; Genotype","score_opus":0.019015620116786917,"score_gpt":0.265093397625932,"score_spread":0.24607777750914506,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2945564624","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":[0.00707543,0.000083277824,0.99088705,0.00014226811,0.000021521635,0.00003434107,0.0003199867,0.0012865717,0.00014956665],"genre_scores_gemma":[0.24105613,0.00028866334,0.7524115,0.00036403374,0.000088940935,0.00037634635,0.0031315289,0.0005604101,0.0017223895],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9979121,0.0014528004,0.00008131223,0.00025227954,0.00021608204,0.000085397936],"domain_scores_gemma":[0.9926388,0.0053086155,0.00050681253,0.0007894967,0.00057469186,0.00018148296],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004598724,0.0011053876,0.0011564085,0.00086293166,0.0003797864,0.0009795295,0.0018558556,0.0010891856,0.0016828489],"category_scores_gemma":[0.013350355,0.0007674829,0.0020184442,0.0007821703,0.0007324196,0.0010921224,0.0015117935,0.0020832007,0.00060727843],"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":[0.00016812458,0.0000876123,0.003985744,0.00012141163,0.00027090765,0.00014881311,0.00008669455,0.9200836,0.0026825538,0.013094656,0.0030471927,0.05622287],"study_design_scores_gemma":[0.000009544065,0.000016675764,0.00015192645,0.0000033797771,0.0000075607854,0.000013724174,0.0000028923575,0.994822,0.00021423858,0.004279241,0.0004726627,0.000006114053],"about_ca_topic_score_codex":0.007526151,"about_ca_topic_score_gemma":0.011192096,"teacher_disagreement_score":0.007526151,"about_ca_system_score_codex":0.00051964703,"about_ca_system_score_gemma":0.0015047293,"threshold_uncertainty_score":0.024320662},"labels":[],"label_agreement":null},{"id":"W2964428942","doi":"10.1142/s0219720019400109","title":"Measuring consistency among gene set analysis methods: A systematic study","year":2019,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Rheumatoid Arthritis Research and Therapies","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":"Royal University Hospital; University of Saskatchewan","funders":"Canadian Institutes of Health Research","keywords":"Gene; Computational biology; Consistency (knowledge bases); Set (abstract data type); Biology; Microarray analysis techniques; Data mining; Gene expression; Genetics; Bioinformatics; Computer science; Artificial intelligence","score_opus":0.0342293175648849,"score_gpt":0.3299604751487715,"score_spread":0.29573115758388663,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2964428942","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"methods","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"methods","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.5212131,0.24447785,0.20830707,0.0028520161,0.0010148134,0.004636773,0.011233264,0.0017853726,0.004479696],"genre_scores_gemma":[0.79319793,0.017373763,0.17651863,0.001218791,0.00021090444,0.003525081,0.0065366626,0.0010073344,0.0004110241],"study_design_codex":"observational","study_design_gemma":"systematic_review","domain_scores_codex":[0.7941171,0.10703488,0.040357463,0.023945183,0.033348594,0.0011967771],"domain_scores_gemma":[0.24403518,0.61557966,0.037460167,0.046713177,0.05463017,0.0015817037],"candidate_categories":["metaresearch"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.2135721,0.0016388875,0.0034953293,0.016902452,0.0023744917,0.004386506,0.0025249429,0.0016660533,0.0006161861],"category_scores_gemma":[0.4668919,0.0012032217,0.005455046,0.015057608,0.0045127156,0.003195867,0.004385844,0.0029025106,0.0002736726],"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":[0.0026639234,0.0004655528,0.48396125,0.042103827,0.043161705,0.00091716234,0.008136794,0.005659488,0.017176343,0.0049049766,0.0058956207,0.38495338],"study_design_scores_gemma":[0.0010573806,0.005077605,0.64700127,0.045355365,0.056201983,0.0072269593,0.008139678,0.04755019,0.062751606,0.027422763,0.090796575,0.0014186243],"about_ca_topic_score_codex":0.00227806,"about_ca_topic_score_gemma":0.0055636056,"teacher_disagreement_score":0.7864279,"about_ca_system_score_codex":0.0028157588,"about_ca_system_score_gemma":0.0066530113,"threshold_uncertainty_score":0.9698056},"labels":[],"label_agreement":null},{"id":"W3008341390","doi":"10.1142/s0219720020500122","title":"Reanalysis of Alzheimer’s brain sequencing data reveals absence of purported HHV6A and HHV7","year":2020,"lang":"en","type":"letter","venue":"Journal of Bioinformatics and Computational Biology","topic":"HIV Research and Treatment","field":"Immunology and Microbiology","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":"University of British Columbia","funders":"","keywords":"Shotgun sequencing; Variola virus; Smallpox; Biology; Disease; Shotgun; Computational biology; DNA sequencing; Virology; Genetics; Medicine; Gene; Pathology; Vaccination","score_opus":0.06476122469075482,"score_gpt":0.3092507797054867,"score_spread":0.24448955501473188,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3008341390","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":[0.1030161,0.0071118595,0.024510205,0.79379666,0.040069595,0.00025146286,0.0049988073,0.0024605002,0.023784833],"genre_scores_gemma":[0.35546407,0.0061468068,0.043746237,0.519954,0.027153982,0.00041473933,0.0054535153,0.001418953,0.040247593],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99779093,0.0003686677,0.00032155646,0.00034335876,0.0009943814,0.0001811175],"domain_scores_gemma":[0.98218,0.007543677,0.0005706984,0.0015477394,0.0071458784,0.0010119394],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0033226712,0.0003906668,0.0004434418,0.0012729221,0.0013740002,0.0015082347,0.0011227053,0.0039099306,0.0024173316],"category_scores_gemma":[0.02300734,0.00034908747,0.00045509805,0.0008051681,0.0008760702,0.0006538467,0.00039140153,0.004466696,0.004916719],"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":[0.000551539,0.00012497604,0.033428147,0.00029001074,0.00015355377,0.0057382523,0.0012384144,0.0003684245,0.016468273,0.002295405,0.7634142,0.1759288],"study_design_scores_gemma":[0.00011902694,0.00034085583,0.044423603,0.0005182414,0.00019553068,0.02945429,0.0018674876,0.0091232145,0.04227696,0.016256671,0.8552416,0.00018260782],"about_ca_topic_score_codex":0.003641223,"about_ca_topic_score_gemma":0.008813075,"teacher_disagreement_score":0.0039099306,"about_ca_system_score_codex":0.00091989717,"about_ca_system_score_gemma":0.0009228187,"threshold_uncertainty_score":0.017572224},"labels":[],"label_agreement":null},{"id":"W3044308231","doi":"10.1142/s0219720020400077","title":"BENIN: Biologically enhanced network inference","year":2020,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Gene expression and cancer classification","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":"Concordia University","funders":"","keywords":"Inference; Computer science; Data mining; Resampling; Gene regulatory network; Machine learning; Artificial intelligence; Biological data; Feature selection; Biological network; Computational biology; Bioinformatics; Biology; Gene; Gene expression","score_opus":0.0187800215957247,"score_gpt":0.2738799799279997,"score_spread":0.25509995833227495,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3044308231","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.003708521,0.00035559654,0.9867906,0.0005558881,0.00009701639,0.000069202106,0.0011950143,0.004825199,0.0024029987],"genre_scores_gemma":[0.120121814,0.00042979198,0.86369026,0.00067520724,0.00013194782,0.0004485342,0.0049767415,0.0015906006,0.007935109],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99860805,0.0007261117,0.000028212999,0.00025032368,0.00031739264,0.000069987116],"domain_scores_gemma":[0.99592894,0.002856636,0.00020280415,0.0005833398,0.00028783217,0.00014050982],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003682889,0.0012095545,0.0013033542,0.0020883847,0.0010070523,0.0016837134,0.0030919942,0.0020700602,0.009400335],"category_scores_gemma":[0.010750745,0.0012041029,0.0016183378,0.0014203632,0.0008665204,0.002175229,0.0027957496,0.0035932031,0.002387558],"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":[0.00038689747,0.0001821769,0.00338522,0.00043298054,0.00040830622,0.0002648604,0.00018434315,0.68024254,0.003358666,0.10953782,0.04025158,0.16136463],"study_design_scores_gemma":[0.000035609326,0.000010462167,0.00020021472,0.000018616336,0.00001555097,0.000038151204,0.000011037242,0.9403568,0.0006827836,0.051790062,0.0068308287,0.000010011012],"about_ca_topic_score_codex":0.007249135,"about_ca_topic_score_gemma":0.01385792,"teacher_disagreement_score":0.009400335,"about_ca_system_score_codex":0.0012967545,"about_ca_system_score_gemma":0.0017120545,"threshold_uncertainty_score":0.03144729},"labels":[],"label_agreement":null},{"id":"W3166605709","doi":"10.1142/s0219720021400126","title":"Comparing the topology of phylogenetic network generators","year":2021,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","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":"Simon Fraser University","funders":"Nederlandse Organisatie voor Wetenschappelijk Onderzoek; Government of Canada; Australian Government","keywords":"Phylogenetic network; Phylogenetic tree; Phylogenetics; Computational phylogenetics; Prior probability; Reticulate; Reticulate evolution; Bayesian network; Horizontal gene transfer","score_opus":0.013106485682164168,"score_gpt":0.23954018837405458,"score_spread":0.22643370269189042,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3166605709","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":[0.8839181,0.0006673979,0.10683351,0.00047941582,0.00005411197,0.0000937996,0.0030020296,0.001089663,0.0038619884],"genre_scores_gemma":[0.9578701,0.0003633988,0.032405917,0.00005968391,0.000029868108,0.0001102388,0.008384263,0.00023408503,0.0005423115],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9985656,0.0006541165,0.00008937673,0.00030884784,0.0002805561,0.00010142421],"domain_scores_gemma":[0.97246176,0.019787021,0.0023724244,0.002288214,0.0022330852,0.00085742783],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0036224583,0.00045057194,0.0005496513,0.0041000927,0.0005872367,0.0018460583,0.0009310982,0.0008873668,0.002155813],"category_scores_gemma":[0.026886,0.00039340963,0.0006091261,0.001913722,0.0008775147,0.0036737218,0.0008822305,0.0008078407,0.00039949542],"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":[0.001747128,0.0002875075,0.15241387,0.0010346916,0.0007110968,0.0007165899,0.0018687797,0.5675875,0.015177701,0.09590492,0.009418077,0.15313222],"study_design_scores_gemma":[0.0000694862,0.00027971555,0.0415548,0.0001029705,0.00019581073,0.0006322397,0.000477851,0.82753664,0.0071831513,0.11612403,0.005741495,0.00010185534],"about_ca_topic_score_codex":0.00068907056,"about_ca_topic_score_gemma":0.0008971026,"teacher_disagreement_score":0.0041000927,"about_ca_system_score_codex":0.00088659354,"about_ca_system_score_gemma":0.0005739163,"threshold_uncertainty_score":0.019157588},"labels":[],"label_agreement":null},{"id":"W3189850282","doi":"10.1142/s0219720021400096","title":"Small parsimony for natural genomes in the DCJ-indel model","year":2021,"lang":"en","type":"preprint","venue":"Journal of Bioinformatics and Computational Biology","topic":"Genome Rearrangement Algorithms","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":"Simon Fraser University","funders":"","keywords":"Indel; Genome; Phylogenetic tree; Tree (set theory); Maximum parsimony; Biology; Gene duplication; Computational biology; Gene; Evolutionary biology; Genetics; Computer science; Algorithm; Combinatorics; Mathematics","score_opus":0.026352074951493613,"score_gpt":0.26751098967713766,"score_spread":0.24115891472564405,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3189850282","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.13589908,0.00041093823,0.8583644,0.0010431284,0.000021547616,0.00006394749,0.0005597171,0.0005438357,0.0030934666],"genre_scores_gemma":[0.54773265,0.00052900106,0.44544172,0.00033403168,0.00006247891,0.00026301772,0.0016495896,0.0004090184,0.0035784922],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99954396,0.00021870073,0.000015072184,0.00012430012,0.00005826702,0.0000397753],"domain_scores_gemma":[0.99649066,0.0030249874,0.0001552642,0.00014830894,0.00008707922,0.000093701645],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015287742,0.00048706212,0.0009414128,0.000977114,0.0007282562,0.0011146665,0.0015411525,0.001365134,0.0023490947],"category_scores_gemma":[0.006348275,0.0004965845,0.0007731425,0.0010986171,0.0010118672,0.0017731576,0.0009385008,0.0016270455,0.00044749904],"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":[0.00012226772,0.000055615656,0.0013183434,0.000083604835,0.000026457206,0.00018556858,0.000096594515,0.9359782,0.001444591,0.040089022,0.0014937841,0.019106017],"study_design_scores_gemma":[0.000013968106,0.000008994609,0.00012399138,0.0000031123782,0.00000358974,0.00003982385,0.000014394977,0.96831393,0.00018329771,0.030859668,0.0004312538,0.00000396209],"about_ca_topic_score_codex":0.003910066,"about_ca_topic_score_gemma":0.004959491,"teacher_disagreement_score":0.003910066,"about_ca_system_score_codex":0.0011201095,"about_ca_system_score_gemma":0.0010870991,"threshold_uncertainty_score":0.008126974},"labels":[],"label_agreement":null},{"id":"W3201936911","doi":"10.1142/s0219720021500268","title":"Compression for population genetic data through finite-state entropy","year":2021,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","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":"Simon Fraser University","funders":"Canadian Network for Research and Innovation in Machining Technology, Natural Sciences and Engineering Research Council of Canada","keywords":"Computer science; Data compression; Computation; Population; Entropy (arrow of time); Data mining; Theoretical computer science; Artificial intelligence; Algorithm","score_opus":0.03354631748808762,"score_gpt":0.3029123453269183,"score_spread":0.2693660278388307,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3201936911","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":[0.03767535,0.0003923114,0.9551224,0.0005302071,0.00018588899,0.000060678267,0.0006050137,0.0027968509,0.0026312382],"genre_scores_gemma":[0.48431468,0.00074692396,0.50662494,0.00036133802,0.00020746751,0.00026024997,0.0028159223,0.00050903036,0.0041594],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9994404,0.00011335071,0.0000585886,0.00008147684,0.00026296658,0.00004318384],"domain_scores_gemma":[0.9970559,0.001567329,0.0001515755,0.0007572786,0.0004239138,0.00004396152],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00076551986,0.00044121192,0.00045299332,0.0010367241,0.00037775747,0.001201786,0.0008002998,0.0004306626,0.004180449],"category_scores_gemma":[0.0065906243,0.00015439665,0.00041452926,0.0016197097,0.0006860328,0.0018496228,0.0011667757,0.00094194984,0.0011695921],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00053878967,0.00013469144,0.0036397676,0.00025334492,0.00006623643,0.00057794625,0.00044790015,0.17062452,0.034169976,0.13829254,0.012138605,0.63911575],"study_design_scores_gemma":[0.00004485581,0.000095944044,0.00078694744,0.00006331344,0.000022599872,0.000392933,0.00007782177,0.8273822,0.06540252,0.09632767,0.009368272,0.000034876415],"about_ca_topic_score_codex":0.0009667999,"about_ca_topic_score_gemma":0.0010420801,"teacher_disagreement_score":0.004180449,"about_ca_system_score_codex":0.00052084064,"about_ca_system_score_gemma":0.00073681463,"threshold_uncertainty_score":0.013985038},"labels":[],"label_agreement":null},{"id":"W3204767928","doi":"10.1142/s0219720021400102","title":"Colorful orthology clustering in bounded-degree similarity graphs","year":2021,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Bioinformatics and Genomic Networks","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":"Cluster analysis; Bounded function; Similarity (geometry); Single-linkage clustering; Graph; Complete-linkage clustering; Computer science; Cluster (spacecraft); Combinatorics; Mathematics; Theoretical computer science; Correlation clustering; Artificial intelligence; CURE data clustering algorithm","score_opus":0.015041494555418982,"score_gpt":0.2547225072202825,"score_spread":0.23968101266486352,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3204767928","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":[0.060190715,0.00017553335,0.93655133,0.00026210325,0.000022273784,0.00012267743,0.00017548681,0.0012541311,0.0012457096],"genre_scores_gemma":[0.38628304,0.00019971209,0.60971814,0.00017343064,0.000030057761,0.00014596277,0.0007464133,0.00034845324,0.0023548133],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9982101,0.0005900576,0.00006302262,0.00066004583,0.00031841555,0.00015829659],"domain_scores_gemma":[0.9944864,0.0028196045,0.0005193079,0.0011275965,0.000716839,0.0003303054],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0018240971,0.0010079772,0.0012515659,0.001603686,0.0017432957,0.0014290382,0.003323396,0.001551799,0.002065603],"category_scores_gemma":[0.007432861,0.00073755014,0.0013464264,0.0018670086,0.001522992,0.0031525416,0.0019795264,0.0013034559,0.0005535334],"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":[0.0002544409,0.00017818046,0.0022331518,0.0002778717,0.000093942246,0.00035880995,0.00046418206,0.77686745,0.020937292,0.077940546,0.0037606456,0.11663344],"study_design_scores_gemma":[0.00002232844,0.000021708232,0.00031256533,0.000007180673,0.000014937461,0.00008579471,0.000049625996,0.9346651,0.004882769,0.05878011,0.0011397294,0.000018076156],"about_ca_topic_score_codex":0.007253539,"about_ca_topic_score_gemma":0.008737203,"teacher_disagreement_score":0.007253539,"about_ca_system_score_codex":0.00244561,"about_ca_system_score_gemma":0.0017065443,"threshold_uncertainty_score":0.017744243},"labels":[],"label_agreement":null},{"id":"W3214208812","doi":"10.1142/s0219720021400072","title":"BOPAL 2.0 and a study of tRNA and rRNA gene evolution in <i>Clostridium</i>","year":2021,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","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 Manitoba","funders":"","keywords":"Gene; Transfer RNA; Ribosomal RNA; Genome; Phylogenetics; Genome evolution; Molecular evolution","score_opus":0.00720873067538824,"score_gpt":0.2322406395318946,"score_spread":0.22503190885650637,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3214208812","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":[0.5417663,0.0036867496,0.38698375,0.0010116984,0.0003075896,0.00035454938,0.0126983095,0.042827226,0.010363862],"genre_scores_gemma":[0.36020434,0.001023326,0.61031485,0.00038993277,0.00006947336,0.0002526752,0.022732122,0.0025086184,0.0025046645],"study_design_codex":"simulation_or_modeling","study_design_gemma":"observational","domain_scores_codex":[0.99954647,0.00012460806,0.000026067835,0.00012179713,0.000116126896,0.000064871856],"domain_scores_gemma":[0.9987355,0.00075827073,0.00009506861,0.00019106414,0.0001499794,0.00007002389],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014848772,0.0009306268,0.00057295547,0.0015448947,0.0006325657,0.0007957265,0.0016083316,0.0009051955,0.0024838664],"category_scores_gemma":[0.0042871204,0.0006650939,0.0006877925,0.0017128012,0.00030367525,0.0010845962,0.0010046139,0.0011096976,0.0011279007],"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":[0.0018984703,0.00054513704,0.03501371,0.0019307269,0.0005891628,0.00087794376,0.00046107476,0.5013614,0.049561996,0.009068392,0.030007765,0.36868414],"study_design_scores_gemma":[0.0001865496,0.00021846453,0.007444424,0.000039760464,0.000064266686,0.00030245993,0.00006399672,0.9638608,0.01021127,0.003229976,0.014320731,0.000057243386],"about_ca_topic_score_codex":0.008795074,"about_ca_topic_score_gemma":0.009376406,"teacher_disagreement_score":0.008795074,"about_ca_system_score_codex":0.0006019501,"about_ca_system_score_gemma":0.001129987,"threshold_uncertainty_score":0.017487764},"labels":[],"label_agreement":null},{"id":"W3216894728","doi":"10.1142/s0219720021400084","title":"The monoploid chromosome complement of reconstructed ancestral genomes in a phylogeny","year":2021,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Genome Rearrangement Algorithms","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 Saskatchewan; University of Ottawa","funders":"","keywords":"Genome; Extant taxon; Phylogenetics; Ancestor; Complement (music); Ploidy","score_opus":0.011550101814240892,"score_gpt":0.2524576239165685,"score_spread":0.2409075221023276,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3216894728","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":[0.92276365,0.00040673057,0.07091865,0.00013734902,0.00001062584,0.000020257303,0.0020704216,0.0006844854,0.0029878155],"genre_scores_gemma":[0.9369002,0.00022710604,0.056562096,0.000037949605,0.000012631527,0.000021940099,0.0050649955,0.00020742549,0.00096562185],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99979097,0.000044118748,0.000009721956,0.000079274054,0.000052507552,0.000023287],"domain_scores_gemma":[0.998691,0.00073923566,0.00013836863,0.00018577887,0.00016878378,0.00007685599],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00062122406,0.00024606506,0.000366707,0.002214702,0.00040496857,0.0010237844,0.00044187356,0.00047192277,0.0040053213],"category_scores_gemma":[0.003486574,0.0002153634,0.00032675275,0.0011573662,0.00039835356,0.0008130357,0.00058423635,0.00045817817,0.00056571496],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0021070677,0.00014780152,0.21938336,0.0005668722,0.00047426907,0.0014885012,0.0018803924,0.19051166,0.27620524,0.04224688,0.003898652,0.26108935],"study_design_scores_gemma":[0.00012658689,0.00025548512,0.13853417,0.0001633705,0.00023515032,0.0019511426,0.00059500325,0.78164405,0.033725403,0.028272562,0.014407683,0.00008930664],"about_ca_topic_score_codex":0.0018095579,"about_ca_topic_score_gemma":0.0036065215,"teacher_disagreement_score":0.0040053213,"about_ca_system_score_codex":0.00043140262,"about_ca_system_score_gemma":0.00047508362,"threshold_uncertainty_score":0.013399124},"labels":[],"label_agreement":null},{"id":"W4281385537","doi":"10.1142/s0219720022500123","title":"Invariant transformers of Robinson and Foulds distance matrices for Convolutional Neural Network","year":2022,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Evolution and Paleontology Studies","field":"Earth and Planetary Sciences","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":"Fonds de Recherche du Québec - Santé","keywords":"Phylogenetic tree; Tree rearrangement; Distance matrices in phylogeny; Distance matrix; Convolutional neural network; Tree (set theory); Artificial intelligence; Supertree; Phylogenetic network; Set (abstract data type); Invariant (physics); Phylogenetics; Inference; Computer science; Mathematics; Algorithm; Biology; Combinatorics; Gene","score_opus":0.014108761810670247,"score_gpt":0.23022461871044725,"score_spread":0.216115856899777,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4281385537","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":[0.02427782,0.00021532537,0.9718259,0.00010468791,0.00004955295,0.000057774476,0.00019036989,0.00039375873,0.0028848131],"genre_scores_gemma":[0.53430325,0.0005182717,0.45736727,0.00013184702,0.00008289794,0.00017204393,0.00080771407,0.00014371943,0.0064729685],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9994572,0.000090500835,0.000039911556,0.00015569228,0.00019651171,0.000060191473],"domain_scores_gemma":[0.9992065,0.00024164646,0.000119790566,0.000101957674,0.0002715891,0.000058484744],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00079620094,0.00056731945,0.00031994728,0.0011995342,0.00030299913,0.000787329,0.0010746226,0.0005936445,0.004072187],"category_scores_gemma":[0.0040641734,0.00020546248,0.0005611717,0.00074645533,0.0006290834,0.0014072747,0.00067310716,0.0009606168,0.0007737852],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00021585869,0.00010468811,0.0022973055,0.00017963762,0.00006129256,0.00020881,0.00015341557,0.28310192,0.023467325,0.21025795,0.0057925056,0.4741593],"study_design_scores_gemma":[0.00000566823,0.000037877737,0.00065252633,0.000010575953,0.0000059396143,0.00007607892,0.000015324887,0.9694113,0.0036218308,0.024612287,0.0015352558,0.000015342055],"about_ca_topic_score_codex":0.008877197,"about_ca_topic_score_gemma":0.008358426,"teacher_disagreement_score":0.008877197,"about_ca_system_score_codex":0.0015567117,"about_ca_system_score_gemma":0.00087304605,"threshold_uncertainty_score":0.017651021},"labels":[],"label_agreement":null},{"id":"W4397003240","doi":"10.1142/s0219720024500112","title":"Body plan evolvability: The role of variability in gene regulatory networks","year":2024,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Animal Genetics and Reproduction","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":"British Columbia Institute of Technology","funders":"","keywords":"Evolvability; Body plan; Gene regulatory network; Gene; Plan (archaeology); Biology; Computational biology; Regulator gene; Genetics; Evolutionary biology; Regulation of gene expression; Gene expression","score_opus":0.004502595958025534,"score_gpt":0.22243530743384907,"score_spread":0.21793271147582352,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4397003240","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":[0.93417555,0.00019500789,0.062250093,0.00041792853,0.000012708014,0.000009101864,0.00014153706,0.0002664245,0.0025316107],"genre_scores_gemma":[0.99667645,0.00004787923,0.0029305625,0.000020660682,0.0000034880165,0.000005859808,0.000030243567,0.000015425985,0.00026952173],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9997218,0.0000767382,0.00001344951,0.00010117202,0.000052756743,0.00003394421],"domain_scores_gemma":[0.99857855,0.0006732833,0.00030568006,0.00029509774,0.000070084905,0.00007731427],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006212593,0.00015621008,0.00017195757,0.0002822603,0.0002326602,0.0011862993,0.00041820927,0.0005462145,0.0013769954],"category_scores_gemma":[0.0031430973,0.00020115601,0.00026121075,0.00017538884,0.0011206159,0.0010433537,0.000662891,0.0006179405,0.00016128876],"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":[0.00034359784,0.00008788557,0.105744466,0.00012491488,0.00017885493,0.0006491636,0.00063786434,0.4613894,0.2795839,0.06981664,0.0008249054,0.08061843],"study_design_scores_gemma":[0.000030367253,0.00022068211,0.11781749,0.000019318384,0.000050900562,0.00059820316,0.00026901858,0.74812573,0.04015634,0.09025686,0.0023932473,0.00006184377],"about_ca_topic_score_codex":0.0008120207,"about_ca_topic_score_gemma":0.00060517754,"teacher_disagreement_score":0.0013769954,"about_ca_system_score_codex":0.0004510072,"about_ca_system_score_gemma":0.00017287493,"threshold_uncertainty_score":0.0046064854},"labels":[],"label_agreement":null},{"id":"W4410642778","doi":"10.1142/s0219720025500076","title":"An algorithm for peptide de novo sequencing from a group of SILAC labeled MS/MS spectra","year":2025,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Advanced Proteomics Techniques and Applications","field":"Chemistry","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":"Western University","funders":"","keywords":"Stable isotope labeling by amino acids in cell culture; Shotgun proteomics; Tandem mass spectrometry; Proteomics; Peptide; Mass spectrometry; Computational biology; Chemistry; Bottom-up proteomics; Peptide sequence; Shotgun; Biology; Chromatography; Protein mass spectrometry; Biochemistry","score_opus":0.01027624271425914,"score_gpt":0.28816092053916004,"score_spread":0.2778846778249009,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4410642778","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":[0.0021766159,0.00006535137,0.99533844,0.000050837243,0.000027416449,0.00015463482,0.00006880686,0.0018598826,0.00025800962],"genre_scores_gemma":[0.009314818,0.000039196188,0.9894755,0.000038869992,0.000012614365,0.00017823622,0.00032175513,0.00011709021,0.00050183706],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99859697,0.00018796792,0.00012993242,0.000536599,0.00044188098,0.00010673403],"domain_scores_gemma":[0.9983165,0.0005474531,0.00015995247,0.00019755903,0.0006866253,0.00009189171],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0022511124,0.0019458326,0.0016234393,0.0023227106,0.0016401291,0.0019398218,0.0026278608,0.0014646365,0.004891529],"category_scores_gemma":[0.00416274,0.00091490615,0.0013719696,0.002044342,0.0007832921,0.0014084015,0.0021103504,0.0021072398,0.002784941],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0004717082,0.00025815007,0.0020476102,0.00033124373,0.0002459126,0.00028355903,0.00020531645,0.10648656,0.031663,0.009481558,0.0076229484,0.8409025],"study_design_scores_gemma":[0.00014500388,0.00013111245,0.0007484664,0.000029637162,0.00006901435,0.00030047222,0.00008130312,0.96293676,0.014040935,0.013420953,0.0080491165,0.0000472735],"about_ca_topic_score_codex":0.0050137346,"about_ca_topic_score_gemma":0.0064100716,"teacher_disagreement_score":0.0050137346,"about_ca_system_score_codex":0.0011401909,"about_ca_system_score_gemma":0.0030847432,"threshold_uncertainty_score":0.01636374},"labels":[],"label_agreement":null},{"id":"W4412923866","doi":"10.1142/s0219720025500106","title":"Visual-SELEX: a technology ensemble for evaluating aptamer structural similarity via 3D visual spatial conformational analysis","year":2025,"lang":"en","type":"article","venue":"Journal of Bioinformatics and Computational Biology","topic":"Cell Image Analysis Techniques","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":"Discovery Centre","funders":"","keywords":"Aptamer; Systematic evolution of ligands by exponential enrichment; Artificial intelligence; Similarity (geometry); Visualization; Pattern recognition (psychology); Computer science; Structural similarity; Computational biology; Biology; Genetics; Image (mathematics)","score_opus":0.007414970158243245,"score_gpt":0.3394696976078277,"score_spread":0.3320547274495844,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4412923866","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.19267891,0.00041383188,0.7901994,0.00017824984,0.000046954134,0.00028951615,0.001048675,0.01212252,0.0030220079],"genre_scores_gemma":[0.4487043,0.0007436526,0.5428497,0.0001552354,0.00001804626,0.00087973266,0.0026175585,0.0011945734,0.0028371252],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9993524,0.00010851166,0.000039868413,0.00016932083,0.00027969957,0.000050131013],"domain_scores_gemma":[0.99943286,0.00022267726,0.00006967336,0.00009702856,0.00012352945,0.00005412518],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016759566,0.0010692267,0.0008618609,0.0014590202,0.0004396166,0.0010720913,0.0011248699,0.0006085079,0.0018029381],"category_scores_gemma":[0.001530968,0.0004907755,0.0008524298,0.0006952904,0.00039696269,0.001025845,0.0015130256,0.00068825576,0.00049637246],"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":[0.0005414203,0.00038517726,0.0080500245,0.0004616823,0.00037954972,0.00029830675,0.00037355526,0.10303692,0.6568664,0.007342756,0.0036119204,0.21865238],"study_design_scores_gemma":[0.000073475276,0.00037753527,0.002987517,0.000019562274,0.00009672708,0.0003403337,0.00009485394,0.6028037,0.3816128,0.0032765802,0.008191119,0.00012579618],"about_ca_topic_score_codex":0.0009698323,"about_ca_topic_score_gemma":0.0010613505,"teacher_disagreement_score":0.0018029381,"about_ca_system_score_codex":0.00066723087,"about_ca_system_score_gemma":0.0006646598,"threshold_uncertainty_score":0.008863449},"labels":[],"label_agreement":null}]}