A Novel Artificial Neural Network and an Improved Particle Swarm Optimization used in Splice Site Prediction

Abstract

The amount of DNA sequence data produced by several genomic projects had increased dramatically in recent years. One of the main goals of bioinformatics was to identify genes. A crucial part of the gene identification was to precisely detect the exon intron boundaries, i.e. the splice sites. This paper introduced a new type of artificial neural network (called NANN), which was designed specifically to solve the splice sites prediction problem. Moreover, the network connection weights of NANN were determined by an improved particle swarm optimization which was inspired by the wolves' activities circle. In addition, three types of encoding approaches were applied to generate the input for the NANN. Intensive experiments were presented in this paper, and the results showed that our algorithm was better than some current methods, that is, the NANN_IPSO was applicable to splice site prediction problem.