Home Department: Computer Science
Supported by: Bio-X
Mentor: Ron Dror, Computer Science

An important question to address in drug development is how to accurately predict the three-dimensional conformation that proteins will adopt when forming a complex with each other. A protein’s importance is conferred by its ability to interact with other molecules (especially other proteins), and understanding the exact interaction conformation is a prerequisite to rationally designing drugs. Rishi will leverage recent machine learning advances in deep neural networks to “learn” the features of likely protein interactions from a large publicly available dataset, the Protein Data Bank.

Poster presented at the Stanford Bio-X Interdisciplinary Initiatives Symposium on August 24, 2017:

Deep Learning-Driven Protein-Protein Docking

Rishi Bedi¹, Raphael Townshend¹, João Rodrigues², Ron Dror¹
[Departments of Computer Science¹ and Structural Biology², Stanford University]