Identifying drug binding sites on proteins from surface conservation and fragment docking

Tuesday, November 14th, 2017

1:00 p.m. Room 202 MRB

Professor Ho-Leung Ng

Associate Professor, Department of Biochemistry and Molecular Biophysics, Kansas State University, Manhattan, Kansas

Identifying drug binding sites on proteins from surface conservation and fragment docking

We describe two new computational approaches for predicting drug binding sites on proteins, ConDock and FragSite. ConDock is the first hybrid scoring function to use information from protein surface conservation and ligand docking to predict ligand binding sites. ConDock uses a simple product function of sequence conservation and binding energy scores. We describe the application of ConDock to predict ligand binding sites to two GPCRs with crystal structures, the b2 adrenergic receptor and A2A adenosine receptor, as well as an uncrystallized GPCR, G-protein coupled estrogen receptor (GPER). We compare the sites predicted by ConDock and traditional methods analyzing surface geometry, surface conservation, and ligand chemical interactions. Incorporating sequence conservation information in ConDock avoids errors resulting from physics-based scoring functions and modeling. We also describe FragSite, which predicts drug binding sites by docking a virtual library of fragments. FragSite uses a docking score optimized for low molecular weight fragments as opposed to drug-like molecules. We describe how FragSite identifies known and new drug binding sites in K-Ras and Stat3. The chemical properties of the bound fragments provide clues on how to expand to larger hit molecules.