Comparison of differential evolution variants for the molecular ligand-receptor docking problem

One of the objectives of the docking methods is to correctly predict the conformation of a small ligand molecule inside the binding site of a biological macromolecule. In this work the performance of six variants of the Differential Evolution (DE) algorithm was evaluated for the flexible ligand docking problem. The algorithm variants were evaluated using five highly flexible HIV-1 protease ligands in the context of the DockThor receptor-ligand docking program. The docking results showed that the DE mutation strategies that uses randomly generated individuals seems to be a better choice for the molecular docking problem than the ones that uses the best individual. Moreover, the results presented in this work provided important guidelines for the development of the next generation of the DockThor optimization algorithm.