Improving Protein Docking Using Sustainable Genetic Algorithms

Computational docking of ligands to protein structures is a key step in identifying potential drug candidates. The docking problem has been formulated into a ligand-protein binding energy optimization problem. Dozens of programs have been developed for molecular docking [1-9]. In any docking scheme, two requirements must be balanced: to get better precision with lower binding energy and to minimize the computational time. Recently, there is significant progress in computational protein docking [10]. The first type of new docking programs utilizes Fast Fourier Transformation (FFT) for efficient sampling of the conformation spaces [1;11-14]. The second type of programs exploits biochemical knowledge of the docking process to improve the docking performance. These include docking algorithms that use shape or physicochemical complementarily information [4;15;16] and predicted binding site information [17]. Another category of new docking algorithms are focused on improving the energy/potential function used for docking [18-22]. Finally, since the first successful application of genetic algorithms for protein docking, there has been significant progress in the global optimization field. New optimization algorithms such as spider-search [23], hybrid GA [24;25], differential evolution [26] have all been successfully applied to a variety of engineering problems. This paper followed this trend to apply new sustainable genetic algorithms to the protein docking problem.