Oral
in
Workshop: Generative and Experimental Perspectives for Biomolecular Design
Generative Active Learning for the Search of Small-molecule Protein Binders
Maksym Korablyov · Chenghao Liu · Moksh Jain · Almer van der Sloot · Eric Jolicoeur · Andrei Nica · Emmanuel Bengio · Kostiantyn Lapchevskyi · Daniel St-Cyr · Doris Schuetz · Victor Butoi · Jarrid Rector-Brooks · Simon Blackburn · Leo Feng · Hadi Nekoei · Vijaya Sai Krishna Gottipati · Priyesh Vijayan · Prateek Gupta · Ladislav Rampášek · Sasikanth Avancha · Pierre-Luc Bacon · William Hamilton · Brooks Paige · Sanchit Misra · Stanislaw Jastrzebski · Bharat Kaul · Doina Precup · José Miguel Hernández Lobato · Marwin Segler · Michael Bronstein · Edward Ruediger · Anne Marinier · Mike Tyers · Yoshua Bengio
Despite substantial progress in machine learning for scientific discovery in recent years, truly de novo design of small molecules which exhibit a property of interest remains a significant challenge. We introduce LambdaZero, a generative active learning approach to search for synthesizable molecules. Powered by deep reinforcement learning, LambdaZero learns to search over the vast space of molecules to discover candidates with a desired property. We apply LambdaZero with molecular docking to design novel small molecules that inhibit the enzyme soluble Epoxide Hydrolase 2 (sEH), while enforcing constraints on synthesizability and drug-likeliness. LambdaZero provides an exponential speedup in terms of the number of calls to the expensive molecular docking oracle, and LambdaZero de novo designed molecules reach docking scores that would otherwise require the virtual screening of a hundred billion molecules. Importantly, LambdaZero discovers novel scaffolds of synthesizable, drug-like inhibitors for sEH. In in vitro experimental validation, a series of ligands from a generated quinazoline-based scaffold were synthesized, and the lead inhibitor N-(4,6-di(pyrrolidin-1-yl)quinazolin-2-yl)-N-methylbenzamide (UM0152893) displayed sub-micromolar enzyme inhibition of sEH.