Extraction of Organism Groups from Whole Genome Comparisons

Abstract

The availability of a growing number of fully sequenced genomes makes it possible for us to conduct a large-scale comparative genomic research. In recent years, functional prediction and genome tree construction methods based on phylogenetic proles have been developed. The phylogenetic prole is dened as a bit pattern that encodes the presence or absence of orthologous genes in a set of organisms with fully sequenced genomes, so the phylogenetic prole can be regarded as the evolutionary pattern of gene acquisition and gene loss across organisms. The genome tree methods based on phylogenetic proles stem from the assumption that evolutionarily correlated organisms have similar gene inheritance patterns. The purpose of this paper is to conduct a whole genome scale phylogenetic analysis in order to extract major organism groups from the comparison of fully sequenced genomes using the independent component analysis, at the same, to identify genes that are characteristic to a certain organism group. In the experiment, we show that 24 major organism groups are extracted from phylogenetic proles composed of orthologous gene clusters, and our genome tree corresponds to the simplied classication of KEGG organism groups and suggests that two prokaryotic kingdoms, Archaea and Bacteria, are closer to each other than to Eukaryotes on the whole genome level. We also show a possibility for predicting biological functions of hypothetical genes by examining the correspondence between the extent of their aliation and specic properties of certain organism groups.