| Protein-protein interactions play an essential role in all cellular activities and elucidation of structural details of protein-protein complexes will contribute to the understanding of their functions. Currently, only a small fraction of experimentally determined structures are of protein-protein complexes, but the gap can be bridged by protein-protein docking methods which aim at computationally generating structural models for protein complexes starting from the structures of the individual protein components. However, protein molecules are dynamic and conformational changes often occur when they associate to form a complex. These changes, which occur at side-chain or/and backbone levels, present a significant challenge for protein-protein docking. In this work, we first present a rapid and efficient method to improve sidechain conformational sampling in protein-protein docking. We go on to demonstrate that with the improved treatment of side-chain flexibility high-resolution docking models with atomic accuracy can be obtained unambiguously in the CAPRI blind docking experiment when conformational changes are restricted to side-chains. Finally, we describe our effort to tackle the much more challenging task of accounting for backbone flexibility in protein-protein docking. We show that the explicit incorporation of backbone flexibility significantly improves both sampling in the vicinity of the native docked conformation and the energetic discrimination between near native and incorrect models. |
