| Proteins bind to other proteins or small molecules to perform essential cellular functions. Protein side-chain flexibility is crucial for binding and molecular recognition. Hence modeling side-chain flexibility in protein-ligand docking algorithms to predict the optimal inter- and intra-molecular interactions is extremely desirable. However, modeling side-chain flexibility in docking and screening is computationally expensive due to the numerous side chains and their many degrees of freedom.; Our research indicates that direct, intra-protein hydrogen bonds and hydrophobic interactions are preserved to a significant extent upon ligand binding. This provides guidance to restrict the number of side-chain candidates for conformational sampling in docking. While these bond-preservation tendencies limit side-chain movements, large side-chain motions are also observed in the protein-ligand interface. Subsequently, the extent of these large side-chain motions from ligand-free to ligand-bound crystal-structure conformations are characterized, as is the suitability of using backbone-dependent rotamers for sampling these larger motions.; The ability to accurately identify which side chains move significantly upon ligand binding as well as their optimal conformations is crucial for docking and screening. A new scoring function, having good linear correlation with experimentally determined protein-ligand binding affinities, is presented for scoring dockings and side-chain interactions by SLIDE. Using the new scoring function as a cost measure, a mean-field based algorithm, exploiting rotamer-based side-chain flexibility modeling, is proposed and tested for optimizing interactions in protein-ligand interfaces. |
