AI in Drug Development Seminar

For their Fall 2022 seminar series, the MBG industry initiative team set out to feature the profound applications of Artificial Intelligence (AI) in drug development, which includes therapeutic target identification, virtual drug screening, and personalized medicine through genomics. To showcase the wide-ranging uses of AI in biomedicine, the team hosted talks by Illumina Vice President Dr. Kyle Farh and Chief Data Officer of Relay Therapeutics, Dr. Pat Walters. These talks were complimentary with one other because they demonstrated how AI is driving innovation in two fields with divergent analytical needs: genomics and protein engineering. Dr. Farh kicked off his seminar by asking the question, “How do we predict the effect of a drug?” and outlined how Illumina’s AI Lab for Genome Interpretation, which uses human genomics and AI to generate actionable insights about biology and disease, is working to answer this question. Dr. Farh and his team are particularly interested in developing tools to predict mRNA splicing patterns from unprocessed mRNA sequences, which may aid the advancement of our understanding of autism and other diseases that are associated with altered splicing. To this end, they develop and maintain Illumina’s SpliceAI deep neural network, which annotates genetic variants with their predicted effect on splicing.

Dr. Walters from Relay Therapeutics began his talk with a similarly memorable quote: “AI is not going to solve drug development.” He elaborated on this by emphasizing that while AI is a powerful tool for predicting protein motion and protein-drug interactions, it relies on experimental validation, and we do not yet have enough experimentally solved protein structures and drug interactions to confidently predict every possible binding interaction. Essentially, AI can help drug developers make smarter decisions about which drugs they test and refine at the lab bench. Dr. Walters and his team use AI to predict protein motion, a historically intractable task because of its high computational demands, but one that may allow drug developers to target a new, more precise axis of protein biology. His team is especially interested in targeting proteins that have dysregulated activity in cancer development. In collaboration with D.E. Shaw Research, they have created the Dynamo platform, which is an integrative and iterative drug development workflow that centers AI-driven motion-based drug design. The MBG Industry Initiative is grateful to our speakers and the MIT community that came to these talks with compelling questions, we look forward to hosting more events that feature AI in Drug Development as the field advances!