Comparative study of random-PSO and Linear-PSO algorithms

This paper presents a Linear-Particle Swarm Optimization (PSO) algorithm for discovering motifs in DNA sequences, and the strengths and weaknesses of using the Linear-PSO for discovering motifs of DNA sequences will be discussed. For the experiment, ten DNA sequences from an online database were used as an input for the algorithms. Linear-PSO uses a reference set for population initialization and new position calculation. Existing PSO algorithms generate random numbers to initialize population and velocity calculation. Random-PSO reduces the accuracy of the algorithm when applied to the detection of motifs for specific species. The results show that Linear-PSO consistently discovered the same motif in each execution with higher fitness values. Even though Linear-PSO detects motifs accurately, the speed of motif discovery is slow and this requires further study.