On the Efficiency of Local Search Methods for the Molecular Docking Problem

Evolutionary approaches to molecular docking typically hybridize with local search methods, more specifically, the Solis-Wet method. However, some studies indicated that local search methods might not be very helpful in the context of molecular docking. An evolutionary algorithm with proper genetic operators can perform equally well or even outperform hybrid evolutionary approaches. We show that this is dependent on the type of local search method. We also propose an evolutionary algorithm which uses the L-BFGS method as local search. Results demonstrate that this hybrid evolutionary outperforms previous approaches and is better suited to serve as a basis for evolutionary docking methods.