Protein-protein docking using three-dimensional reduced representations and based on a genetic algorithm

Abstract

An original scoring function dedicated to the docking of biological macromolecules is implemented in complementarity research within an automated algorithm. As these systems involve complicated atomic structures, we use for each partner reduced representations obtained by topological analysis of electron density maps at medium resolution, and develop specific terms for the characterization of the intermolecular interactions including a geometric fit based on the knowledge in a statistical survey, an electronic interaction potential using an expression of modified Coulomb type, and a penalty score based on detection of steric clashes. To validate the strategy, we performed automated docking runs, based on genetic algorithms (GA) for various protein-protein complexes including enzyme-inhibitor and antibody-antigen. For most complexes, the GA-proposed fit solutions have rmsd values below 3Å relative to the native structures. © 2008 Springer.