Michigan State University
Title: Opportunities and challenges for AI and Math in drug discovery
Date: Friday, January 31, 2020
Place and Time: Room 101, Love Building, 3:35-4:25 pm
Refreshments: Room 204, Love Building, 3:00 pm
Abstract. Drug discovery is one of the most challenging tasks in the biological sciences since it requires over 10 years and costs more than 2.6 billion to put an average novel medicine on the marketplace. The abundant availability of biological data along with the flourishing advanced AI algorithms opens a future with great hope for discovering new drugs faster and cheaper. Unfortunately, AI faces an enormous obstacle in drug discovery due to the intricate complexity of biomolecular structures and the high dimensionality of biological datasets. In our lab, these challenges have been tackled mathematically. We have introduced multiscale modeling, differential geometry, algebraic topology, and graph theory-based models to systematically represent the diverse biological datasets in the low-dimensional spaces. Combining these mathematical representations with cutting edge deep neural networks, we arrived at novel models not only perform well on virtual-screening targeting important drug properties but also have the ability to design new drugs at an unprecedented speed. Our team has emerged as a top winner in D3R Grand Challenges, a worldwide annual competition series in computer-aided drug design, in the past few years.