Exploring potential energy surfaces of biological molecules using a Multi-Niche Crowding genetic algorithm

Abstract

A research strategy based on the use of a genetic algorithm combined with high quantum chemical calculations has been developed in order to improve practical techniques for exploring the potential energy surfaces (PES) of molecules of biological interest. The genetic algorithm based on the Multi-Niche Crowding (MNC) technique was used initially to generate a set of optimized structures for a short peptide chain N-formyl-L-Tyrosinamide (HCO-L-Tyrosine-NH ${}_2$ ). These structures will be used hereafter as input conformers for calculations at DFT/B3LYP/6-31G(d) level of theory in order to locate most stable minima on its PES. The efficiency of the used strategy was tested by comparing the obtained results with those derived from an ordinary research one based on the use of hierarchical optimization procedure. This consists of a hierarchy of increasingly more accurate electronic structure calculations (HF/3-21G, HF/6-31G(d) and DFT/B3LYP/6-31G(d) geometry optimizations and single-point energies calculations respectively). The developed strategy was able to predict 23 conformations against 35 localized by the ordinary strategy and in which the 9 most stables ones are in the same stability order. Furthermore, comparison between the theoretical calculations and experimental X-ray data from the Brookhaven protein data bank (PDB) revealed that themost abundant conformers in the protein structuresshould have the lowest energy.