Unsupervised statistical identification of genomic islands using oligonucleotide distributions with application to Vibrio genomes

Abstract

Vibrio cholerae, Vibrio vulnificus, Vibrio parahaemolyticus and several other related Vibrio species show distinctly similar two-chromosomal genome organization. However, the modes of pathogenicity are very different among these species, and this is largely attributed to externally acquired genetic elements. We develop some statistical methods to determine these external genetic elements or genomic islands in genomes based on their differential oligonucleotide usage patterns compared to the rest of the genome. Genomic islands identified by these unsupervised statistical methods include integron and pathogenicity islands. After statistical determination of the genomic islands, we investigate their gene contents and their possible association with the pathogenic behaviour of the corresponding Vibrio species. These investigations lead to observations that are of evolutionary and biological significance.