A Genetic Algorithm Based on Restricted Tournament Selection for the 3D-PSP Problem

Three-dimensional Protein Structure Prediction is one of the most challenging problems in Structural Bioinformatics. The difficulty in predicting the native-like structure of a protein comes from the many plausible shapes that a protein molecule can assume. In this paper, we proposed a Genetic Algorithm based on the Restricted Tournament Selection approach that makes use of a priori information about known protein structures to avoid premature convergence along the search procedure. Our method combines a template fragment library, obtained from experimental-determined protein structures, with the conformational preferences of amino acid residues to reduce the protein conformational search space. The proposed method was tested with seven protein sequences with different folding patterns. Predicted structures are topologically compatible with the experimental correspondent showing the effectiveness of the proposed method.