{"id":"W7104034538","doi":"10.48620/92223","title":"Resolving data bias improves generalization in binding affinity prediction.","year":2025,"lang":"en","type":"article","venue":"Open Access CRIS of the University of Bern","topic":"Computational Drug Discovery Methods","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"Canadian Bulletin of Medical History","funders":"","keywords":"Benchmark (surveying); Generalization; Training set; Graph; Labeled data; Artificial neural network; Retraining; Test data","routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false,"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":[0.004871336,0.001315207,0.001244105,0.0007834922,0.0004483419,0.001016088,0.002501885,0.001874549,0.001931272],"category_scores_gemma":[0.01782612,0.0006346684,0.001401805,0.0007109579,0.001443984,0.002612367,0.002084934,0.002996241,0.001061799],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001250194,"about_ca_system_score_gemma":0.001379276,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00659434,"about_ca_topic_score_gemma":0.008034089,"domain_scores_codex":[0.9983864,0.00060307,0.0001030129,0.0005034957,0.0002850845,0.0001189936],"domain_scores_gemma":[0.9948114,0.003119657,0.0003562947,0.001104658,0.0004653713,0.0001426198],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0005025271,0.00034641,0.008818049,0.0002397421,0.0002668812,0.0002184927,0.0001607145,0.7757265,0.007042843,0.008127407,0.007004262,0.1915461],"study_design_scores_gemma":[0.00001869766,0.00009140744,0.0005557694,0.00001217307,0.00002434861,0.00004059389,0.000008485287,0.9878218,0.001879706,0.008816011,0.0007202909,0.0000108405],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1884737,0.00213684,0.7940817,0.002925562,0.0002825509,0.0002250525,0.001363087,0.005523377,0.004988028],"genre_scores_gemma":[0.898923,0.0006304674,0.0914811,0.001578746,0.0001409408,0.0002652172,0.002374268,0.0003383191,0.004268013],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.00659434,"threshold_uncertainty_score":0.02576238,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1290013584496259,"score_gpt":0.376518259563397,"score_spread":0.2475169011137711,"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."}}