Inter-C? atomic potentials derived from the statistics of average interresidue distances in proteins: Application to bovine pancreatic trypsin inhibitor

New effective potentials acting between pairs of residues in proteins are proposed based on statistics of average distances and standard deviations between Cα atoms of residues in protein tertiary structures. Gaussian functions are adopted as analytical forms of the potentials. A protein structure is modeled as a chain molecule with a fixed bond length connecting particles approximating the effects of amino acid residues. The potentials derived in this study are used for conformational sampling of trypsin inhibitor from bovine pancreas. Sampling is done with the Monte Carlo simulated annealing method. Sampled conformations can be classified into a few groups or structural classes, and one of these classes contains structures relatively close (with 7.8–8.7 Å root mean square [rms] deviation) to the X-ray structure. The native structure exhibits relatively low energy. These results denote a rather smooth landscape of the present potential energy surfaces. One class of classified structures contains nativelike structures, which suggests that the native structure can be predicted by further refinement of structures in this class. We discuss other properties and the effectiveness of the present potentials for description of protein structures.