September 7, Tue 2010
1:00 pm, MRB 100 Conference Room
Dr. Xiaoqin Zou
Assistant Professor of Biophysics and Bioinformatics, University of Missouri-Columbia
A new scoring framework for selecting structural models
Reliable and efficient energy scoring functions are important for accurate structural model selection in protein structure prediction. Unfortunately, such scoring functions still remain at large. I will present our recent study on a statistical mechanics-based iterative method to extract effective pairwise interaction potentials from known protein structures, using the physical pair distribution function. The resulted scoring function has been validated using three diverse benchmarks. Significant improvement in performance has been achieved. The derived scoring function can be easily implemented as a ranking tool for structural selection in programs for protein structure prediction. I will also present our studies on developing computational methods for predicting protein-ligand interactions and application to rational drug design.