Combining Water Percolation Analysis and Molecular Dynamics Simulations for Protein-Protein Binding Interface Prediction

Abstract

We consider ab initio prediction of protein-protein binding interfaces of weakly-associating proteins in aqueous environment. Systems of interest are important in cellular signalling pathways and manifest Kd values in the micromolar range (weak binding). We use molecular dynamics simulations of individual proteins at varying hydration levels to determine percolation thresholds for water networks. By mapping average local water density around all residues at the percolation threshold, we are able to determine clusters of residues with low water density which we consider as potential binding regions. We have performed analysis on a series of 20 proteins and compare to existing automated protein-protein binding interface prediction servers.