Multi-objective ligand-protein docking with particle swarm optimizers

• Published in: Swarm and evolutionary computation Vol. 44; pp. 439 - 452
• Main Authors: García-Nieto, José, López-Camacho, Esteban, García-Godoy, María Jesús, Nebro, Antonio J., Aldana-Montes, José F.
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.02.2019
• Subjects: Algorithm comparison; Archiving strategies; Molecular docking; Multi-objective optimization; Particle swarm optimization; Archiving strategies; Algorithm comparison; Multi-objective optimization; Particle swarm optimization; Molecular docking
• ISSN: 2210-6502
• DOI: 10.1016/j.swevo.2018.05.007

In the last years, particle swarm optimizers have emerged as prominent search methods to solve the molecular docking problem. A new approach to address this problem consists in a multi-objective formulation, minimizing the intermolecular energy and the Root Mean Square Deviation (RMSD) between the atom coordinates of the co-crystallized and the predicted ligand conformations. In this paper, we analyze the performance of a set of multi-objective particle swarm optimization variants based on different archiving and leader selection strategies, in the scope of molecular docking. The conducted experiments involve a large set of 75 molecular instances from the Protein Data Bank database (PDB) characterized by different sizes of HIV-protease inhibitors. The main motivation is to provide molecular biologists with unbiased conclusions concerning which algorithmic variant should be used in drug discovery. Our study confirms that the multi-objective particle swarm algorithms SMPSOhv and MPSO/D show the best overall performance. An analysis of the resulting molecular ligand conformations, in terms of binding site and molecular interactions, is also performed to validate the solutions found, from a biological point of view.