Abstract: The classification of genes is quite important in the understanding of gene regulation. The genes are grouped into transcription units for the purpose of construction and regulation of gene expression and synthesis of proteins. This knowledge further contributes as essential information for the process of drug design and to determine the protein functions of newly sequenced genomes. It is possible to use the diverse biological information across multiple genomes as an input to the classification problem. The purpose of this work is to show that Particle Swarm Optimization may improve the results of classification as compared to other algorithms. To validate the approach E.Coli complete genome is taken as the benchmark genome.