The problem of understanding the relationship between protein structure and sequence is central to the areas of protein design, homologous modeling, mutation analysis and the understanding and prediction of protein folding. We have developed a computer algorithm which is able to determine the conformation of short pieces of polypeptide chain in the larger protein environment. The algorithm systematically generates all possible conformations for a piece of structure and uses discriminatory functions based on electrostatics, including solvation, and exposed non- polar area to select a correct conformation. The method has been used to determine the conformation of a set of loops in known protein structures and to choose conformations most compatible with crystallographic data. A ligand docking procedure has been developed, utilizing the electrostatic discriminatory functions. A model based on the dominance of the burial of nonpolar area in stabilizing structure has allowed the identification of small independent folding units in proteins . Analyses of steric strain and electrostatics in proteins have been carried out to aid in the development of discriminatory functions. The scope of the algorithm will be enhanced, allowing the treatment of longer sequences. Discrimination based on packing, atom overlap and residue conformational preference will be added. A conditional probability formalism, measuring the information supporting the hypothesis that a structure is correct, will be developed to allow proper integration of the different discriminatory functions. Two additional conformational search techniques, torsion space Monte Carlo and genetic algorithms will introduced. Preliminary results show that the Monte Carlo method is able to determine the conformation of independent folding units, and that genetic algorithms should be able to greatly extent the effectiveness of these conformational search methods.
Showing the most recent 10 out of 19 publications
