{"id":"W4306754767","doi":"10.1038/s42004-022-00733-0","title":"Generative and reinforcement learning approaches for the automated de novo design of bioactive compounds","year":2022,"lang":"en","type":"article","venue":"Communications Chemistry","topic":"Computational Drug Discovery Methods","field":"Computer Science","cited_by":104,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Eshelman Institute for Innovation, University of North Carolina at Chapel Hill; Division of Chemistry; Office of Naval Research; Ontario Ministry of Economic Development and Innovation; Merck KGaA; Fundação de Amparo à Pesquisa do Estado de São Paulo; Genome Canada; U.S. Department of Health and Human Services; National Institutes of Health; National Science Foundation; Ministero dello Sviluppo Economico; U.S. Department of Defense; Novartis Pharma; Nvidia; Pfizer","keywords":"Generative grammar; Reinforcement learning; Computer science; Artificial intelligence; Machine learning; Artificial neural network; Generative model; Generative Design; Deep learning; Engineering","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":[0.0007954016,0.00007398933,0.00009174811,0.00002274246,0.0007508787,0.00004916602,0.001309813,0.00001777628,0.000004064734],"category_scores_gemma":[0.0001248176,0.00006867461,0.00003415572,0.0002366293,0.0001630314,0.00008741459,0.001315945,0.0001866108,1.37811e-7],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001104726,"about_ca_system_score_gemma":0.0001815756,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000007195673,"about_ca_topic_score_gemma":2.082988e-7,"domain_scores_codex":[0.9991695,0.0002906093,0.0001684959,0.0001414146,0.000126587,0.0001033921],"domain_scores_gemma":[0.9969809,0.002009286,0.0001471586,0.0007652566,0.00007293112,0.00002449027],"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":[0.00001222819,0.00004978139,0.00003274911,0.00001440189,0.00005526657,9.886323e-8,0.002030515,0.9769903,0.007140868,0.01059896,0.0001048556,0.002969998],"study_design_scores_gemma":[0.0001845516,0.0000294248,0.0001400982,0.000003890674,0.00001171843,0.00001598577,0.0007791918,0.9867876,0.009420938,0.001745786,0.0008084833,0.00007229162],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.003355381,0.0005644425,0.9936063,0.001448113,0.00001445891,0.0002434203,0.000005725944,0.00008415656,0.0006780602],"genre_scores_gemma":[0.7404139,0.00002670604,0.2591113,0.00003884429,0.000005656231,0.0003030644,0.00002605574,0.000004728108,0.00006980096],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.7370585,"threshold_uncertainty_score":0.5775228,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1211153519488273,"score_gpt":0.3308011182719686,"score_spread":0.2096857663231412,"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."}}