{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,19]],"date-time":"2026-10-19T20:16:49Z","timestamp":1792441009625,"version":"4.5.0"},"reference-count":24,"publisher":"American Chemical Society (ACS)","issue":"10","license":[{"start":{"date-parts":[[2026,6,29]],"date-time":"2026-06-29T00:00:00Z","timestamp":1782691200000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2026,10,19]]},"abstract":"<jats:title>Abstract<\/jats:title>\n                  <jats:p>Regulatory agencies require comprehensive genotoxicity assessments for novel small-molecule therapeutics prior to human trials. Developers often delay these studies until a candidate is nearing regulatory submission because they are expensive and secondary to bioactivity. This timing creates a bottleneck where late-stage failures can jeopardize &amp;gt;\\$10 million in capital and multiple years of developmental progress per candidate. The Ames assay detects a molecule\u2019s mutagenic potential. Regulators now explicitly support the use of in silico Ames mutagenicity models through enabling legislation, dedicated FDA AI toxicology programs, internationally harmonized guidelines, and benchmark challenges. However, current Ames models suffer from a sensitivity drop-off when they evaluate molecules outside their training domain. The out-of-domain (OOD) test set from the Second Ames\/QSAR International Challenge Project yielded a participant average sensitivity of 0.46. Sensitivity is the most important metric in Ames prediction because false negatives allow mutagenic compounds to advance undetected and trigger the most costly late-stage failures. Attempts to fix this sensitivity drop-off often reduce overall model performance, which can be represented by balanced accuracy (BA). For example, DeepAmes reports high levels of sensitivity (0.87) by sacrificing its BA (0.52). We introduce AmesNet, a task-conditioned Ames model that achieves class-leading performance in novel chemical spaces. AmesNet utilizes a dual branch architecture containing a molecular encoder and a dedicated channel to condition Ames assay context such as metabolic activation and bacterial strain type. In comparative benchmarks, AmesNet reached a sensitivity of 0.72 (95% CI: 0.68\u20130.76) and a simultaneous BA of 0.81 (95% CI: 0.78\u20130.83) on the OOD test data. This performance improves discrimination without the sensitivity trade-off limiting existing approaches. Structural analysis demonstrates that AmesNet recovers difficult-to-detect mutagenic compounds missed by existing models. This framework provides a high-confidence filtering mechanism that enables drug developers to turn a costly late-stage safety bottleneck into an earlier-stage mutagenicity triage.<\/jats:p>","DOI":"10.1021\/acs.chemrestox.6c00082","type":"journal-article","created":{"date-parts":[[2026,6,29]],"date-time":"2026-06-29T15:35:13Z","timestamp":1782747313000},"page":"1946-1956","source":"Crossref","is-referenced-by-count":4,"title":["AmesNet:\nA Task-Conditioned Deep Learning Model with\nEnhanced Sensitivity and Generalization in Ames Mutagenicity Prediction"],"prefix":"10.1021","volume":"39","author":[{"given":"Tyler J.","family":"Umansky","sequence":"first","affiliation":[{"name":"Model Medicines , , ,","place":["San Diego, California, United States, 92122"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Virgil A.","family":"Woods","sequence":"additional","affiliation":[{"name":"Model Medicines , , ,","place":["San Diego, California, United States, 92122"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Sean M.","family":"Russell","sequence":"additional","affiliation":[{"name":"Model Medicines , , ,","place":["San Diego, California, United States, 92122"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0000-5265-0126","authenticated-orcid":true,"given":"Daniel J.","family":"Haders","sequence":"additional","affiliation":[{"name":"Model Medicines , , ,","place":["San Diego, California, United States, 92122"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"316","published-online":{"date-parts":[[2026,6,29]]},"reference":[{"key":"2026101907024586400_cit1","doi-asserted-by":"publisher","first-page":"782","DOI":"10.1073\/pnas.70.3.782","article-title":"An improved bacterial test system\nfor the detection and classification of mutagens and carcinogens","volume":"70","author":"Ames","year":"1973","journal-title":"Proc. Natl. Acad. Sci. U.S.A."},{"key":"2026101907024586400_cit4","doi-asserted-by":"publisher","DOI":"10.1093\/bib\/bbaa160","article-title":"Current\ndevelopment of integrated\nweb servers for preclinical safety and pharmacokinetics assessments\nin drug development","volume":"22","author":"Hsiao","year":"2021","journal-title":"Briefings Bioinf."},{"key":"2026101907024586400_cit5","doi-asserted-by":"publisher","first-page":"99","DOI":"10.1016\/j.mrrev.2012.12.002","article-title":"Bacterial\nmutagenicity\nscreening in the pharmaceutical industry","volume":"752","author":"Escobar","year":"2013","journal-title":"Mutat.\nRes.\/ Rev. Mutat. Res."},{"key":"2026101907024586400_cit6","doi-asserted-by":"publisher","first-page":"19","DOI":"10.1016\/B978-0-12-821222-6.00002-6","article-title":"Preclinical testing\u2500Understanding\nthe basics first","volume-title":"Drug Delivery Aspects","author":"Shegokar","year":"2020"},{"key":"2026101907024586400_cit7","unstructured":"U.S. Congress FDA Modernization Act 3.0. S. 355, 119th Cong\n          2025\n          https:\/\/www.congress.gov\/bill\/119th-congress\/senate-bill\/355."},{"key":"2026101907024586400_cit8","unstructured":"FDA.\n          . U.S. Food and\nDrug Administration Artificial Intelligence (AI) Program for Toxicology\nat NCTR2024https:\/\/www.fda.gov\/about-fda\/nctr-research-focus-areas\/artificial-intelligence."},{"key":"2026101907024586400_cit9","unstructured":"FDA.\n          . U.S. Food and\nDrug Administration SafetAI Initiative2025https:\/\/www.fda.gov\/about-fda\/nctr-research-focus-areas\/safetai-initiative."},{"key":"2026101907024586400_cit10","doi-asserted-by":"publisher","DOI":"10.1016\/j.yrtph.2023.105486","article-title":"DeepAmes: A deep learning-powered\nAmes test predictive model with potential for regulatory application","volume":"144","author":"Li","year":"2023","journal-title":"Regul. Toxicol. Pharmacol."},{"key":"2026101907024586400_cit12","doi-asserted-by":"publisher","first-page":"3","DOI":"10.1093\/mutage\/gey031","article-title":"Improvement of quantitative\nstructure-activity relationship (QSAR) tools for predicting Ames mutagenicity:\noutcomes of the Ames\/QSAR International Challenge Project","volume":"34","author":"Honma","year":"2019","journal-title":"Mutagenesis"},{"key":"2026101907024586400_cit13","doi-asserted-by":"publisher","first-page":"983","DOI":"10.1080\/1062936X.2023.2284902","article-title":"Evaluation of QSAR models\nfor predicting mutagenicity: outcome of the Second Ames\/QSAR international\nchallenge project","volume":"34","author":"Furuhama","year":"2023","journal-title":"SAR QSAR Environ. Res."},{"key":"2026101907024586400_cit14","doi-asserted-by":"publisher","first-page":"29","DOI":"10.1016\/S0027-5107(00)00064-6","article-title":"The Ames Salmonella\/microsome\nmutagenicity assay","volume":"455","author":"Mortelmans","year":"2000","journal-title":"Mutat. Res., Fundam. Mol.\nMech. Mutagen."},{"key":"2026101907024586400_cit15","doi-asserted-by":"crossref","unstructured":"Umansky, T.; Woods, V.; Russell, S. M.; Smith, D.; Haders, D.\n          ChemPrint: An AI-Driven Framework\nfor Enhanced Drug Discovery\n          bioRxiv\n          2024\n          10.1016\/s0027-5107(00)00064-6.","DOI":"10.1101\/2024.03.22.586314"},{"key":"2026101907024586400_cit16","doi-asserted-by":"crossref","unstructured":"Umansky, T.; Woods, V.; Russell, S. M.; Garvey, D. S.; Smith, D.; Haders, D.\n          GALILEO Generatively Expands Chemical\nSpace and Achieves One-Shot Identification of a Library of Novel,\nSpecific, Next Generation Broad-Spectrum Antiviral Compounds at High\nHit Rates\n          bioRxiv\n          2025\n          10.1101\/2025.01.17.633620.","DOI":"10.1101\/2025.01.17.633620"},{"key":"2026101907024586400_cit17","unstructured":"Woods, V.; Umansky, T.; Russell, S. M.; Poole, I.; David, A.; Geerke, D. P.; Garvey, D. S.; Smith, D.; Haders, D.\n          Discovery\nof RdRp Thumb-1 as a novel\nbroad-spectrum antiviral family of targets and MDL-001 as a potent\nbroad-spectrum inhibitor thereof\n          bioRxiv\n          2025\n          10.1101\/2024.03.29.587401."},{"key":"2026101907024586400_cit18","unstructured":"Woods, V.; Umansky, T.; Russell, S. M.; McGovern, B. L.; Rosales, R.; Rodriguez, M. L.; van Bakel, H.; Sordillo, E. M.; Simon, V.; Garci\u0301a-Sastre, A.; White, K. M.; Brubaker, W. F.; Smith, D.; Haders, D.\n          MDL-001: An Oral,\nSafe, and Well-Tolerated Broad-Spectrum Inhibitor of Viral Polymerases\n          bioRxiv\n          2025\n          10.1101\/2025.01.13.632836."},{"key":"2026101907024586400_cit19","doi-asserted-by":"publisher","first-page":"1248","DOI":"10.1021\/acs.chemrestox.2c00385","article-title":"Mechanistic\nTask Groupings Enhance\nMultitask Deep Learning of Strain-Specific Ames Mutagenicity","volume":"36","author":"Lui","year":"2023","journal-title":"Chem. Res. Toxicol."},{"key":"2026101907024586400_cit20","doi-asserted-by":"publisher","first-page":"1871","DOI":"10.1021\/acs.chemrestox.5c00198","article-title":"Comparative\nChemical Space Analysis of Pesticides and\nSubstances with Genotoxicity Data","volume":"38","author":"Foil","year":"2025","journal-title":"Chem. Res.\nToxicol."},{"key":"2026101907024586400_cit21","doi-asserted-by":"publisher","DOI":"10.3390\/ph18121853","article-title":"SAR Modeling to Predict\nAmes Mutagenicity Across Different Salmonella typhimurium Strains","volume":"18","author":"Dmitriev","year":"2025","journal-title":"Pharmaceuticals"},{"key":"2026101907024586400_cit22","doi-asserted-by":"publisher","first-page":"28","DOI":"10.1021\/acs.jcim.5c02332","article-title":"Chemprop v2: An Efficient,\nModular Machine Learning Package for Chemical Property Prediction","volume":"66","author":"Graff","year":"2026","journal-title":"J. Chem. Inf. Model."},{"key":"2026101907024586400_cit23","unstructured":"Rong, Y.; Bian, Y.; Xu, T.; et al. In  Self-Supervised Graph\nTransformer\non Large-Scale Molecular Data, Advances in Neural Information\nProcessing Systems; NeurIPS, 2020."},{"key":"2026101907024586400_cit24","doi-asserted-by":"publisher","first-page":"1227","DOI":"10.1021\/acs.chemrestox.2c00372","article-title":"Multiple Instance Learning Improves\nAmes Mutagenicity Prediction for Problematic Molecular Species","volume":"36","author":"Feeney","year":"2023","journal-title":"Chem. Res. Toxicol."},{"key":"2026101907024586400_cit25","doi-asserted-by":"publisher","first-page":"843","DOI":"10.1021\/tx2000398","article-title":"Comparative Evaluation of in Silico Systems for Ames\nTest Mutagenicity Prediction: Scope and Limitations","volume":"24","author":"Hillebrecht","year":"2011","journal-title":"Chem. Res. Toxicol."},{"key":"2026101907024586400_cit26","doi-asserted-by":"publisher","first-page":"14","DOI":"10.1016\/j.mrfmmm.2007.03.014","article-title":"Genotoxicity of non-covalent interactions: DNA intercalators","volume":"623","author":"Ferguson","year":"2007","journal-title":"Mutat. Res., Fundam. Mol. Mech. Mutagen."},{"key":"2026101907024586400_cit27","doi-asserted-by":"publisher","first-page":"986","DOI":"10.1016\/j.tiv.2009.06.025","article-title":"Mutagenicity\nof aromatic amines and\namides with chemical models for cytochrome P450 in Ames assay","volume":"23","author":"Inami","year":"2009","journal-title":"Toxicol. In Vitro"}],"container-title":["Chemical Research in Toxicology"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/pubs.acs.org\/doi\/pdf\/10.1021\/acs.chemrestox.6c00082","content-type":"application\/pdf","content-version":"vor","intended-application":"unspecified"},{"URL":"https:\/\/pubs.acs.org\/crtoec\/article-pdf\/39\/10\/1946\/65279162\/acs.chemrestox.6c00082.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"https:\/\/pubs.acs.org\/crtoec\/article-pdf\/39\/10\/1946\/65279162\/acs.chemrestox.6c00082.pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,10,19]],"date-time":"2026-10-19T11:12:01Z","timestamp":1792408321000},"score":1,"resource":{"primary":{"URL":"https:\/\/pubs.acs.org\/crtoec\/article\/39\/10\/1946\/5170705\/AmesNet-A-Task-Conditioned-Deep-Learning-Model"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,6,29]]},"references-count":24,"journal-issue":{"issue":"10","published-online":{"date-parts":[[2026,6,29]]},"published-print":{"date-parts":[[2026,10,19]]}},"URL":"https:\/\/doi.org\/10.1021\/acs.chemrestox.6c00082","relation":{},"ISSN":["0893-228X","1520-5010"],"issn-type":[{"value":"0893-228X","type":"print"},{"value":"1520-5010","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,6,29]]}}}