Use of the genomic signature in bacterial classification and identification

In this study we investigated the correlation between dinucleotide relative abundance values (the genomic signature) obtained from bacterial whole-genome sequences and two parameters widely used for bacterial classification, 16S rDNA sequence similarity and DNA-DNA hybridisation values. Twenty-eight completely sequenced bacterial genomes were included in the study. The correlation between the genomic signature and DNA-DNA hybridisation values was high and taxa that showed less than 30% DNA-DNA binding will in general not have dinucleotide relative abundance dissimilarity (delta*) values below 40. On the other hand, taxa showing more than 50% DNA-DNA binding will not have delta* values higher than 17. Our data indicate that the overall correlation between genomic signature and 16S rDNA sequence similarity is low, except for closely related organisms (16S rDNA similarity >94%). Statistical analysis of delta* values between different subgroups of the Proteobacteria indicate that the beta- and gamma-Proteobacteria are more closely related to each other than to the other subgroups of the Proteobacteria and that the alpha- and epsilon-Proteobacteria form clearly separate subgroups. Using the genomic signature we have also predicted DNA-DNA binding values for fastidious or unculturable endosymbionts belonging to the genera Rickettsia, Wigglesworthia and Buchnera.