The project intends to improve and extend the accuracy and application of empirical potential energy functions for peptides and proteins. These functions form the basis of an extensive set of modeling techniques widely used to rationalize and predict the structure and conformational energetics of organic molecules and biopolymers. Based on simple principles from classical chemical physics, this type computer modeling software is widely used in a variety of settings ranging from undergraduate chemistry and biochemistry to research at the forefront of structural biology. While empirical potential energy functions often exhibit exquisite accuracy in modeling nonpolar organics, their application to typical biological systems presents several additional difficulties. The major goal of this project is to find solutions to the problems that most limit current application of this methodology in protein modeling: (1) achieving chemical accuracy for protein structure and energetics as governed by basic physics, and (2) describing widely varied chemical environments such as the solvent exposed surface of a protein and its hydrophobic interior. Since the treatment of electrostatic interactions is the largest source of error in these calculations, particular emphasis will be placed upon developing an atom-centered approach to electrostatics with inclusion of dipole polarization effects. A class of hybrid methods is proposed for further development, using a shell of explicit water coupled with a reaction field treatment of long range bulk solvent effects.
