Rotamer-specific Statistical Potentials for Protein Structure Modeling.

Protein Structure Prediction;Statistical Potential Energy Function;Rotamer-specific Multibody Potential;Mechanical Engineering;Engineering;Mechanical Engineering

Park, JungkapVioli, Angela ;

University of Michigan

关键词: Protein Structure Prediction;    Statistical Potential Energy Function;    Rotamer-specific Multibody Potential;    Mechanical Engineering;    Engineering;    Mechanical Engineering;   

Others  :  https://deepblue.lib.umich.edu/bitstream/handle/2027.42/102342/jungkap_1.pdf?sequence=1&isAllowed=y

瑞士|英语

来源: The Illinois Digital Environment for Access to Learning and Scholarship

【 摘 要 】

Knowledge-based (or statistical) potentials are widely used as essential tools in protein structure modeling and quality assessment. They are derived from experimentally determined protein structures aiming to extract relevant structural features that characterize the tightly folded structures. Since the surrounding circumstances are inhomogeneous and anisotropic, multibody contributions are important for accurate account of cooperative effects of molecular interactions. On the other hand, protein residues have great flexibility. It is energetically favorable for residues to adopt only a limited number of staggered conformations, known as rotamers. Depending on the rotameric state, the residue conformation and intra-residue interaction vary significantly within protein structures, resulting in different solvent accessibility and different electric polarization effect as well as different steric effect on residue elements. The major goal of this thesis is the design and development of statistical potentials that take into account the rotamer-dependence of interactions. We hypothesized that the rotameric state of residues is related to the specificity of interactions within protein structures. We first investigated how amino acid residues in PDB structures show different interaction patterns with the environment depending on their rotameric states. Observed rotamer-specific environmental features were incorporated to a scoring function, ProtGrid for protein designs. Our tests demonstrated that the ProtGrid is superior to widely used Rosetta energy function in prediction of the native amino acid types and rotameric states.Next, we formulated a rotamer-specific atomic statistical potential, named ROTAS that extends an existing orientation-dependent atomic potential (GOAP) by including the influence of rotameric states of residues on the specificity of interactions. The results showed that ROTAS performs better than other competing potentials not only in native structure recognition, but also in best model selection and correlation coefficients between energy and model quality. Finally, we applied the ROTAS potential to the problem of side-chain prediction. Compared with the existing side-chain modeling programs, ROTAS achieved comparable or even better prediction accuracy. We expect that the effectiveness of our energy functions would provide insightful information for the development of many applications which require accurate side-chain modeling such as homology modeling, protein design, mutation analysis, protein-protein docking and flexible ligand docking.