Particle Swarm Optimization with Reinforcement Learning for the Prediction of CpG Islands in the Human Genome

Abstract

BackgroundRegions with abundant GC nucleotides, a high CpG number, and a length greater than 200 bp in a genome are often referred to as CpG islands. These islands are usually located in the 5′ end of genes. Recently, several algorithms for the prediction of CpG islands have been proposed.Methodology/Principal FindingsWe propose here a new method called CPSORL to predict CpG islands, which consists of a complement particle swarm optimization algorithm combined with reinforcement learning to predict CpG islands more reliably. Several CpG island prediction tools equipped with the sliding window technique have been developed previously. However, the quality of the results seems to rely too much on the choices that are made for the window sizes, and thus these methods leave room for improvement.Conclusions/SignificanceExperimental results indicate that CPSORL provides results of a higher sensitivity and a higher correlation coefficient in all selected experimental contigs than the other methods it was compared to (CpGIS, CpGcluster, CpGProd and CpGPlot). A higher number of CpG islands were identified in chromosomes 21 and 22 of the human genome than with the other methods from the literature. CPSORL also achieved the highest coverage rate (3.4%). CPSORL is an application for identifying promoter and TSS regions associated with CpG islands in entire human genomic. When compared to CpGcluster, the islands predicted by CPSORL covered a larger region in the TSS (12.2%) and promoter (26.1%) region. If Alu sequences are considered, the islands predicted by CPSORL (Alu) covered a larger TSS (40.5%) and promoter (67.8%) region than CpGIS. Furthermore, CPSORL was used to verify that the average methylation density was 5.33% for CpG islands in the entire human genome.

DNA Methylation Biomarkers: Cancer and Beyond

“…The definition of a CpG island has been quite arbitrary and two algorithms have found widespread use throughout the scientific community to identify CpG-islands in genomic DNA sequences [ 12 , 13 ]. However, genome-wide studies have vastly increased our understanding of the human genome over the last few years, and more sophisticated algorithms for the identification of CpG-islands have been developed [ 14 , 15 , 16 ].…”

Ion Motion Optimization Algorithm Applied to CpG Island Prediction

“…In IMO algorithm, only two parameters need to be set: the number of iterations and the population size. In this study, the parameters for the population size is 300 (each of anions and cations are 150) [20], the iteration number is 100 [7].…”

“…Many methods have been proposed to predict CpG islands such as CpGIS [3], CpGProD [4], CpGcluster [5], CpGPlot [6] and CPSORL [7]. CpGPlot predicting CpG islands by sliding window approach is proposed by Rice, P. et al in 2000.…”

Inferring Gene Regulatory Networks by Singular Value Decomposition and Gravitation Field Algorithm

“…A suitable heuristic search algorithm should be used for searching the best solutions. Genetic algorithm (GA) [9], simulated annealing (SA) [10], particle swarm optimization (PSO) [11] and other algorithms do not comply with the reconstruction of GRNs algorithms. Because the connectivity of GRNs is equal to squared nodes, a novel algorithm called gravitation field algorithm (GFA) [12], which can resolve large-scale computational problems, should be used in our work.…”