Zobrazit v EDS

Optimizing drug design by merging generative AI with a physics-based active learning framework

Uloženo v:
Podrobná bibliografie
Název: Optimizing drug design by merging generative AI with a physics-based active learning framework
Autoři: Isaac Filella-Merce, Alexis Molina, Lucía Díaz, Marek Orzechowski, Yamina A. Berchiche, Yang Ming Zhu, Júlia Vilalta-Mor, Laura Malo, Ajay S. Yekkirala, Soumya Ray, Victor Guallar
Zdroj: UPCommons. Portal del coneixement obert de la UPC
Universitat Politècnica de Catalunya (UPC)
Informace o vydavateli: Springer Science and Business Media LLC, 2025.
Rok vydání: 2025
Témata: Virtual screening, Generative models (GMs), Computational chemistry, Cheminformatics, Àrees temàtiques de la UPC::Informàtica::Aplicacions de la informàtica::Bioinformàtica, Machine learning, Àrees temàtiques de la UPC::Enginyeria química, Drug discovery and development, Drug design
Popis: Machine learning is transforming drug discovery, with generative models (GMs) gaining attention for their ability to design molecules with specific properties. However, GMs often struggle with target engagement, synthetic accessibility, or generalization. To address these, we developed a GM workflow integrating a variational autoencoder with two nested active learning cycles. These iteratively refine their predictions using chemoinformatics and molecular modeling predictors. We tested our workflow on two systems, CDK2 and KRAS, successfully generating diverse, drug-like molecules with high predicted affinity and synthesis accessibility. Notably, we generated novel scaffolds distinct from those known for each target. For CDK2, we synthetized 9 molecules yielding 8 with in vitro activity, including one with nanomolar potency. For KRAS, in silico methods validated by CDK2 assays identified 4 molecules with potential activity. These findings showcase our GM workflow’s ability to explore novel chemical spaces tailored for specific targets, thereby opening new avenues in drug discovery.
Druh dokumentu: Article
Popis souboru: application/pdf
Jazyk: English
ISSN: 2399-3669
DOI: 10.1038/s42004-025-01635-7
Přístupová URL adresa: https://hdl.handle.net/2117/440567
https://doi.org/10.1038/s42004-025-01635-7
Rights: CC BY NC ND
Přístupové číslo: edsair.doi.dedup.....6de1dc7a1ffcb1b0b2970a876d8996a8
Databáze: OpenAIRE
Popis
Abstrakt:Machine learning is transforming drug discovery, with generative models (GMs) gaining attention for their ability to design molecules with specific properties. However, GMs often struggle with target engagement, synthetic accessibility, or generalization. To address these, we developed a GM workflow integrating a variational autoencoder with two nested active learning cycles. These iteratively refine their predictions using chemoinformatics and molecular modeling predictors. We tested our workflow on two systems, CDK2 and KRAS, successfully generating diverse, drug-like molecules with high predicted affinity and synthesis accessibility. Notably, we generated novel scaffolds distinct from those known for each target. For CDK2, we synthetized 9 molecules yielding 8 with in vitro activity, including one with nanomolar potency. For KRAS, in silico methods validated by CDK2 assays identified 4 molecules with potential activity. These findings showcase our GM workflow’s ability to explore novel chemical spaces tailored for specific targets, thereby opening new avenues in drug discovery.
ISSN:23993669
DOI:10.1038/s42004-025-01635-7