#27 Connor Coley, Tailoring generative AI to small molecule design for early stage (drug) discovery

Описание: Title: Tailoring generative AI to small molecule design for early stage (drug) discovery


Abstract: Generative molecular design seeks to propose novel molecular structures that may surpass what is available in enumerated virtual libraries. While enumerated libraries are large, the enormity of chemical spaces means that there are ample opportunities for "creativity" in molecular design. In this talk, I will describe two recent projects that advance our ability to design molecular structures. First, I will introduce a 3D molecular generative model that facilitates interaction-aware chemical design by learning the joint distribution over 3D molecular structures and their shapes, electrostatics, and pharmacophores. Second, I will describe a generative framework that mitigates lack of synthesizable as a major failure mode of molecular design. Our model ensures that every generated molecule has a viable synthetic pathway, enabling the design of analogs and optimization of molecular properties while maintaining synthetic feasibility.



Bio: Connor W. Coley is the Class of 1957 Career Development Professor and an Associate Professor at MIT in the Department of Chemical Engineering and the Department of Electrical Engineering and Computer Science. His research group works at the interface of AI and chemistry to develop models that understand how molecules behave, interact, and react and use that knowledge to engineer new ones, with an emphasis on therapeutic discovery. He received his B.S. in Chemical Engineering from Caltech and his Ph.D. from MIT, followed by postdoctoral training at the Broad Institute. Outside of MIT, Connor is a scientific advisor to several companies pursuing AI-driven discovery. He serves on the advisory boards of ACS Central Science, Chemical Science, and Digital Discovery, and is an Associate Editor of the Journal of the American Chemical Society.