Classification of E. coli and Shigella by SVM.

Bacterial phylogenetic analyses are commonly performed to explore the evolutionary relationships among various bacterial species and genera based on their 16S rRNA gene sequences; however, these results are limited by mosaicism, intragenomic heterogeneity, and difficulties in distinguishing between related species. In this study, we aimed to perform genome-wide comparisons of different bacterial species, namely Escherichia coli, Shigella, Yersinia, Klebsiella, and Neisseria spp., based on their K-mer profiles to construct phylogenetic trees. Pentanucleotide frequency analyses (512 patterns of 5 nucleotides each) were performed to distinguish between highly similar species. Moreover, Escherichia albertii strains were clearly distinguished from E. coli and Shigella, despite being closely related to enterohemorrhagic E. coli in the phylogenetic tree. In addition, our phylogenetic tree of Ipomoea species based on pentamer frequency in chloroplast genomes was correlated with previously reported morphological similarities. Furthermore, a support vector machine clearly classified E. coli and Shigella genomes based on their pentanucleotide profiles. These results suggest that phylogenetic analyses based on penta- or hexamer profiles are a useful methodology for microbial phylogenetic studies. In addition, we introduced an R application, Phy5, which generates a phylogenetic tree based on genome-wide comparisons of pentamer profiles. The online version of Phy5 can be accessed at https://phy5.shinyapps.io/Phy5R/ and its command line version Phy5cli can be downloaded at https://github.com/YoshioNakano2021/phy5.