Whole-genome prokaryotic clustering based on gene lengths Original Research Article

The fast-growing number of complete genome sequences prompts the development of new phylogenetic approaches. Until recently, understanding the phylogeny of prokaryotes was based on the comparison of highly conserved genes. Several novel whole-genome methods have been proposed during the last few years. Here, we present a novel method of taxonomic analysis, constructed on the basis of gene content and lengths of orthologous genes of 66 completely sequenced genomes of unicellular organisms using Clusters of Orthologous Groups (COGs). Our input data consist of average protein lengths related to ∼5000 COGs from 66 genomes. We clustered these data, using an application of the information bottleneck method for unsupervised clustering. This approach is not a regular distance-based method, distinguishing it from other recently published whole-genome based clustering techniques. Although our comprehensive genome clustering is independent of phylogenies based on the level of homology of individual genes, it correlates well with the standard “tree of life” based on sequence similarity of 16s rRNA.