Assoc. Prof. Daisuke Kihara
Biological Sciences
Purdue University
West Lafayette, IN

Monday, October 28, 2013
3:00 pm (CT)
3269 Beckman Institute

Abstract

Protein-protein binding events mediate many important biological functions in a cell. To provide molecular view of interacting proteins, our group has been developing a series of methods for predicting the three dimensional structure of protein docking complexes. In this presentation, first we will introduce a protein docking algorithm, LZerD. LZerD uses the 3D Zernike descriptor, a rotational invariant mathematical representation of protein surfaces, which is used to detect shape complementarity of interaction sites. Then, we will introduce BindML, which predicts docking interface regions in protein surfaces by detecting binding site specific mutation patterns. We have further developed a protein docking algorithm that uses imperfect predicted protein-protein docking interface prediction to guide docking conformation search (PI_LZerD). Finally, we present Multi-LZerD, which can predict the structure of multiple docking proteins by combining pairwise docking prediction computed by LZerD. Multi-LZerD was further extended to EM-LZerD that fits multiple component structures into a low-resolution electron microscopy