A Knowledge-Based Genetic Algorithm to Predict Three-Dimensional Structures of Polypeptides

Three-dimensional (3-D) protein structure determination has become an important area of research in structural bioinformatics. Proteins are responsible for the execution of different functions in the cell. Understanding the 3-D structure provides important information about the protein function. Many computational methodologies for the protein structure prediction were developed along the last 20 years, but the problem still challenges researchers because the complexity and high dimensionality of its large search space. In this article we present a strategy for reducing the search space explored by heuristic methods for solving the problem taken into consideration previous occurrences of amino acid residues in a well known protein database (PDB). We propose a genetic algorithm that takes advantages of this kind of information, reducing considerable the search space, allowing the algorithm to save time with less promising solutions. A simple Local Search operator helps the GA to intensify the search of the 3-D protein conformational space. We demonstrate the effectiveness of the strategy with a set of experimental results. © 2013 IEEE.