Please use this identifier to cite or link to this item:
Title: Improving Protein-Ligand Docking by Particle Swarm Optimization
Authors: NG, CHON KIT (吳俊杰)
Department: Department of Computer and Information Science
Faculty: Faculty of Science and Technology
Issue Date: 2014
Citation: NG, C. K. (2014). Improving Protein-Ligand Docking by Particle Swarm Optimization (Outstanding Academic Papers by Students (OAPS)). Retrieved from University of Macau, Outstanding Academic Papers by Students Repository.
Abstract: Molecular docking is an indispensable tool in computer-aided drug design. Given a target protein related to the disease of interest, molecular docking helps to decide if a small molecule called ligand is able to bind to the protein’s binding pocket with certain level of affinity. Protein-ligand docking consists of two main steps: A conformation (a possible configuration of the molecule structure) is generated by conformational search algorithm and then the conformation is evaluated by a fitness function. The conformation which has the best value returned by the fitness function will be determined as the predicted structure. To be able to search quickly and intelligently over the huge conformational space is necessary in the virtual screening step of drug design, where millions of ligands in a drug compound library will have to be screened through by docking. In recent years, swarm intelligence algorithms have emerged as a fast and reasonably accurate technique in solving complex search problems in computer science. But the applicability of this technique has not been fully explored in the molecular docking problem. Therefore, in this project one of the most popular swarm intelligence algorithms, particle swarm optimization (PSO) was studied for their applicability to the ligand conformational search problem in protein-ligand docking. The algorithm is, for the first time, implemented into the popular docking program AutoDock Vina, here called PSOxVina. Using a benchmark dataset of 201 experimental protein-ligand complexes, the prediction accuracy and time efficiency for docking pose prediction were rigorously tested. Remarkably, PSOxVina completes the docking process in only 54% of the time used in AutoDock Vina, but the prediction accuracy is increased by 7% measured by the averaged RMSD of the predicted structures from experimental structures. Our work demonstrates that PSO is superior to conventional search algorithms such as Monte Carlo in molecular docking.
Instructor: Dr. SIU, WENG IN
Programme: Bachelor of Science in Software Engineering
Appears in Collections:FST OAPS 2014

Files in This Item:
File Description SizeFormat 
OAPS_2014_FST_014.pdf3.01 MBAdobe PDFView/Open

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.