Evaluating the Use of Local Search Strategies for a Memetic Algorithm for the Protein-Ligand Docking Problem

Molecular docking aims to predict the best conformation of a molecule when it is attached to another, to form a stable complex. This problem is considered to be of a high computational complexity because of the large number of factors that must be taken into account in the binding process. In this paper we present three different local search strategies that were implemented as part of a memetic algorithm for the molecular docking problem. The algorithms were tested with real complexes obtained from Protein Data Bank. The results were evaluated according to two criteria: minimum energy and RMSD respect to the crystallographic structure of the complex. The results generated by the algorithm are satisfactory, since it was possible to generate real-like conformations in 75% of cases. Finally the results showed that one of the techniques achieved the best free energy values in 90% of the tests. © 2017 IEEE.