Operon Prediction Using Particle Swarm Optimization and Reinforcement Learning

Abstract

An operon is a fundamental unit of transcription contains a specific function of genes for the construction and regulation of networks at the whole genome level. The operon prediction is critical for the understanding of gene regulation and functions in newly sequenced genomes. Various methods for operon prediction have been proposed in the literature shows that the experimental methods for operon detection are tend to be non-trivial and time-consuming. In this study, a binary particle swarm optimization (BPSO) and reinforcement learning (RL) are used for operon prediction in bacterial genomes. The intergenic distance, participation in the same metabolic pathway and the gene length ratio property of the Escherichia coli genome are used to design a fitness function based on the conception of RL. Then the three genomes are used to test the prediction performance of BPSO with RL. Experimental results show that the prediction accuracy of this method reached to 92.8%, 94.3% and 95.9% on Bacillus subtilis, Pseudomonas aeruginosa PA01 and Staphylococcus aureus genomes respectively. The proposed method for the predicted operons with the highest accuracy contains the three test genomes.