A Parallel Algorithm for the Prediction of Protein Binding Sites

Abstract

The Pocket-Finder algorithm identifies the location of ligand binding sites in a protein and is a fundamental component for a range of applications including molecular docking, de novo drug design and structural identification and comparison of functional sites. In this paper, we propose a parallel version of the Pocket-Finder algorithm. The proposed parallel algorithm uses a geometrical approach to locate favorable binding sites and has been MPI-enabled for parallel execution. The proposed algorithm has been applied on a small test of 15 proteins and 2 proteins complexes. The algorithm gets very interesting results when applied to large proteins.