{"id":"W4408690205","doi":"10.1039/d5sc01476j","title":"Directly optimizing for synthesizability in generative molecular design using retrosynthesis models","year":2025,"lang":"en","type":"article","venue":"Chemical Science","topic":"Computational Drug Discovery Methods","field":"Computer Science","cited_by":22,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Natural Sciences and Engineering Research Council of Canada; Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung","keywords":"Retrosynthetic analysis; Generative grammar; Computer science; Artificial intelligence; Machine learning; Engineering; Chemistry; Stereochemistry","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":[0.002269789,0.000140181,0.000212156,0.0001944237,0.0001534646,0.0002099999,0.001328872,0.00005233262,0.000001180949],"category_scores_gemma":[0.001891511,0.0001355844,0.00007210582,0.001837563,0.0003316041,0.0008449954,0.0004487447,0.0001005515,4.959851e-7],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0004990554,"about_ca_system_score_gemma":0.0006312423,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00001418504,"about_ca_topic_score_gemma":3.387914e-7,"domain_scores_codex":[0.9980063,0.0001546757,0.0002718788,0.0008033327,0.0003571245,0.0004066596],"domain_scores_gemma":[0.997773,0.001485909,0.00005941044,0.0004266878,0.0001685307,0.00008647272],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.000007967449,0.0000342632,0.00001039032,0.000008955109,0.000002398672,0.000001285518,0.00009704942,0.3878654,0.5683939,0.03636272,0.000003970372,0.007211675],"study_design_scores_gemma":[0.0000496051,0.000003457157,0.00001437261,0.00002438137,0.00000191054,6.255984e-7,0.000004053566,0.4883248,0.4304923,0.0810107,0.000001706531,0.00007210167],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.1395209,0.00006612034,0.8591824,0.000357764,0.00009928741,0.0002933326,0.000001437115,0.00004753884,0.000431143],"genre_scores_gemma":[0.4578143,7.040276e-7,0.5420301,0.0001167781,0.000005794064,0.00002671685,1.302394e-7,0.00000268148,0.000002794585],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.3182934,"threshold_uncertainty_score":0.552897,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.06296741142680679,"score_gpt":0.3462113161996178,"score_spread":0.283243904772811,"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."}}