As before, the Acidaminococcaceae cluster together but within the Veillonellaceae. The position of Turicibacter sanguinis within the Bacilli group of Firmicutes is consistent Brefeldin A ARFs with the other two trees but contrasts with its taxonomic description at NCBI as a member of the Erysipelotrichia. Table 2 Universally conserved COGs Figure 3 Consensus tree based on the phylogenetic trees of 27 genes conserved in all 145 genomes. The collapsed branch of the Bacilli, marked (1), contains Turicibacter sanguinis. An uncollapsed tree is available as a supplemental figure. In conclusion, based on three independent phylogenetic analyses, the closest relatives to the Negativicutes seem to be the Clostridiaceae. The observed clustering of species within the Negativicutes is consistent with their assigned taxonomy.

Furthermore, these analyses show that Veillonella spp. form a distinct branch, most different from the other Negativicutes, while the recent change of status of the Acidaminococcaceae (they are no longer a separate family) is confirmed by these analyses. Apart from comparing proteins and genes, genomes can also be compared based on nucleotide composition irrespective of their coding capacity. For instance, the frequency of nucleotide combinations can reveal similarities between genomes that are independent of protein-coding information. We compared the frequency of tetranucleotides for all 145 genomes. The observed frequency of all 64 tetranucleotide combinations was extracted for each genome and these frequencies were divided by the theoretically calculated, expected frequencies (corrected for differences in base composition).

This ratio, which could be interpreted as a genomic signature, was expected to reflect taxonomic divisions [29]. However, although the analysis identified a high similarity in tetranucleotide frequency for all of the analyzed Veillonella genomes, most of the clustering observed was not in accordance with known taxonomic relationships. Not only were Negativicutes other than Veillonella separated from each other and strewn across the phyla, but also several other Firmicutes were distributed over various branches (data shown as supplementary material). In fact, for most of the analyzed genomes, members of identical phyla did not cluster together and even the Archaea were mixed with Bacteria, although some closely related species were indeed clustered.

This may explain why all Veillonella genomes grouped together. Several organisms with similar Dacomitinib tetranucleotide frequencies did not share a common ecological niche, in contrast to previously reported observations (reviewed in [30]). Neither was the obtained clustering dictated by GC-content. The conclusion from this analysis was that tetranucleotide analysis is only taxonomically informative for closely related genomes. We also compared whole-genome amino acid frequencies in each of the deduced proteomes.