Performance Analysis of Swarm Intelligence Algorithms for the 3D-AB off-lattice Protein Folding Problem
Rafael Stubs Parpinelli, Cesar M.V. Benitiez, Jelson Cordeiro and Heitor Silverio Lopes
This paper compares the performance of four swarm intelligence algorithms for the optimization of a hard bioinformatic problem: the protein structure prediction problem (PSP). The PSP involved the protein folding that is the process by which polypeptide chains are transformed into compact structures that perform biological functions. In this work, we tested the standard versions of the following algorithms: Particle Swarm Optimization (PSO), Artificial Bee Colony (ABC), Gravitational Search Algorithm (GSA), and the Bat Algorithm (BA). The algorithms were evaluated using two criteria: quality of solutions and the processing time. The results show that the PSO algorithm presented the overall best balance between these two criteria. Also, both PSO and GSA displayed potential to evolve even better solutions, if more iterations were given.
Keywords: Swarm intelligence; 3D-AB model; protein folding problem; particle swarm optimization; artificial bee colony; gravitational search algorithm; bat algorithm