{"id":"W4408347689","doi":"10.1021/acscentsci.5c00055","title":"Challenging Reaction Prediction Models to Generalize to Novel Chemistry","year":2025,"lang":"en","type":"article","venue":"ACS Central Science","topic":"Computational Drug Discovery Methods","field":"Computer Science","cited_by":16,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Division of Chemistry; Natural Sciences and Engineering Research Council of Canada; Novo Nordisk; Massachusetts Institute of Technology","keywords":"Computer science; Chemistry; Combinatorial chemistry","routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006461843,0.0001181211,0.0001026212,0.0001686733,0.0002866681,0.0002499332,0.001154258,0.00003362873,0.00000258483],"category_scores_gemma":[0.0001430136,0.0001226642,0.0000327716,0.002189753,0.00005147629,0.001270428,0.0005618854,0.00009627092,0.00001408838],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0004048395,"about_ca_system_score_gemma":0.0003396611,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00006503102,"about_ca_topic_score_gemma":0.000003131598,"domain_scores_codex":[0.9980427,0.00002033082,0.0001972432,0.0006669277,0.0005658311,0.0005069872],"domain_scores_gemma":[0.9990664,0.00006024046,0.00003697019,0.0004305436,0.000149669,0.0002561394],"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":[0.000004740613,0.0000503913,0.00003632831,0.00000726616,0.000003237453,8.716694e-7,0.0005581461,0.5760453,0.3393379,0.06428722,0.0002387155,0.0194299],"study_design_scores_gemma":[0.0001547591,0.00001582163,0.01144446,0.00004773625,0.000003070476,0.000006574283,0.00003515444,0.9086547,0.07132804,0.006567589,0.001580954,0.0001611293],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.2177576,0.00001438286,0.7756579,0.00277332,0.000739602,0.0001628948,0.000003663912,0.0001196419,0.002771072],"genre_scores_gemma":[0.8946461,0.000005389407,0.1040975,0.0008885816,0.00007674372,0.00002014883,0.000001406974,0.000003739529,0.0002603652],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.6768885,"threshold_uncertainty_score":0.5002101,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03330431404660639,"score_gpt":0.2996567613055541,"score_spread":0.2663524472589477,"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."}}