2016 Poster Sessions : Statistical Learning and Docking Uncover the Reaction Coordinates of a GPCR

Student Name : Evan Feinberg
Advisor : Vijay Pande
Research Areas: Theory
Abstract:
G-Protein Coupled Receptors (GPCRs) comprise one-third of targets of all FDA-approved drugs. Molecular dynamics (MD) simulations of GPCRs can contain over 60,000 atoms, counting for over 180,000 degrees of freedom. The technique described here reduces the dimensionality of GPCR MD simulations through a combination of unsupervised and supervised learning. In particular, time-structure Independent Component Analysis (tICA) and molecular docking are used complementarily. This dual use of unsupervised and supervised learning approaches determines the reaction coordinates relevant to agonist binding and receptor activation.

Bio:
Evan N. Feinberg is a Ph.D. student in Biophysics in the Pande Lab at Stanford University. He applies machine learning methods to both understand the dynamics of G-Protein Coupled Receptors (GPCRs), as well as to discover how drugs modulate their activity. He is the co-lead developer of deepchem (deepchem.io), an open source software package that enables scientists to adapt modern machine learning methods for drug discovery. Before joining the Pande Lab, Evan studied Applied Physics at Yale, where he graduated Magna Cum Laude, Phi Beta Kappa in 2013.