{"id":"W4413337851","doi":"10.1038/s42256-025-01080-0","title":"The importance of negative training data for robust antibody binding prediction","year":2025,"lang":"en","type":"article","venue":"Nature Machine Intelligence","topic":"Monoclonal and Polyclonal Antibodies Research","field":"Medicine","cited_by":1,"is_retracted":false,"has_abstract":false,"ca_institutions":"McMaster University","funders":"","keywords":"Training (meteorology); Antibody; Training set; Computer science; Medicine; Immunology; Artificial intelligence; Geography; Meteorology","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":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007794794,0.0001386971,0.0002514385,0.0001141267,0.000260518,0.0000233775,0.000548707,0.0001518938,0.00003045417],"category_scores_gemma":[0.001839519,0.00008459544,0.00008729748,0.0004624639,0.0002091939,0.000085861,0.000215012,0.0007612684,0.00000243889],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00003331816,"about_ca_system_score_gemma":0.0001971082,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00005233368,"about_ca_topic_score_gemma":0.0001022908,"domain_scores_codex":[0.9986469,0.0000371303,0.0003756113,0.0003442404,0.0003212076,0.000274974],"domain_scores_gemma":[0.9978834,0.001140404,0.0001120959,0.0005556131,0.0002490941,0.00005942916],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.006669248,0.0004310793,0.2744626,0.001779968,0.001469241,0.00005422018,0.001333384,0.0004290497,0.02881979,0.1076978,0.0232482,0.5536054],"study_design_scores_gemma":[0.001927013,0.00152485,0.0802139,0.002657713,0.0005480102,0.0001656503,0.003404231,0.5296832,0.1194821,0.02116669,0.2385407,0.000685922],"study_design_candidate":"design_other","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.3905837,0.05781972,0.4202317,0.05923273,0.005006782,0.007131556,0.007302096,0.0003419975,0.0523497],"genre_scores_gemma":[0.9848515,0.001269679,0.0076253,0.0004661893,0.0002653995,0.00002508326,0.0007483395,0.00001668871,0.004731857],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.5942677,"threshold_uncertainty_score":0.3449702,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.06454479274959338,"score_gpt":0.4149155629044256,"score_spread":0.3503707701548322,"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."}}