Conformational Sampling of Loop Structures Using Genetic Algorithms

Abstract

Abstract BLoop, a method for generating plausible loop conformations for protein models, is described. Conformations that are compatible with the local sequence are generated by using the search mechanisms of genetics algorithms to explore the possible combinations of tetrapeptide conformers. Because the amino acid propensities that are used to judge the quality of the structures are derived from a data set of known protein structures, BLoop reproduces many features of database loop searches. But unlike these strategies that are limited by their contents, BLoop is capable of generating novel conformations. The expected accuracy of the database and BLoop methods for loop generation are compared on a set of loops as a function of residue length.