{"id":"W4226251181","doi":"10.1093/bioadv/vbac030","title":"Expanding the Galaxy’s reference data","year":2022,"lang":"en","type":"article","venue":"Bioinformatics Advances","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":3,"is_retracted":false,"has_abstract":true,"ca_institutions":"Université de Sherbrooke; BC Centre for Disease Control","funders":"National Human Genome Research Institute; Biotechnology and Biological Sciences Research Council; Directorate for Biological Sciences; National Cancer Institute; National Institutes of Health; European Commission; Vlaamse regering; National Research Foundation; UK Research and Innovation; Cleveland Clinic","keywords":"Computer science; Galaxy; Task (project management); Interface (matter); Reference data; Information retrieval; Database; Astrophysics; Operating system; Physics; Engineering","routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":true,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008820063,0.002264489,0.002328244,0.008677315,0.002753458,0.007385784,0.008899019,0.003634826,0.05813423],"category_scores_gemma":[0.03517934,0.002299678,0.003103843,0.01191781,0.001228217,0.007303915,0.01136101,0.005709644,0.1071623],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.003094531,"about_ca_system_score_gemma":0.005915713,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.02502164,"about_ca_topic_score_gemma":0.02540175,"domain_scores_codex":[0.9915016,0.001330418,0.0007529081,0.002108167,0.00353454,0.0007723345],"domain_scores_gemma":[0.978265,0.002178986,0.0007865389,0.01140635,0.005809323,0.00155373],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.0002814845,0.00003393369,0.0034546,0.000814414,0.0001731543,0.0001931907,0.0004999128,0.001015072,0.003961211,0.0112516,0.9294353,0.04888617],"study_design_scores_gemma":[0.00007540942,0.00002302878,0.001914677,0.0001546534,0.00005742507,0.00009175904,0.00006971551,0.001023239,0.003701149,0.006491419,0.9863274,0.00007009719],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.007551462,0.002508709,0.08171578,0.003539497,0.001941751,0.00029846,0.571268,0.2756305,0.05554591],"genre_scores_gemma":[0.01397077,0.0006315576,0.0761793,0.002004811,0.0002346153,0.0005438075,0.846826,0.0508244,0.008784768],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.05813423,"threshold_uncertainty_score":0.1944784,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03741575304613087,"score_gpt":0.2864572823709869,"score_spread":0.2490415293248561,"validation_status":"score_only:v0-immature-baseline","note":"Baseline scores from an immature model (maturity gate not passed). Scores rank; they never assert a category."}}