VICTOR: Genome-based Phylogeny and Classification of Prokaryotic Viruses

Abstract

Bacterial and archaeal viruses (\"phages\") play an enormous role in global life cycles and have recently regained importance as therapeutic agents to fight serious infections by multi-resistant bacterial strains. Nevertheless, taxonomic classification of phages is up to now only insufficiently informed by genome sequencing. Despite thousands of publicly available phage genomes, it still needs to be investigated how this wealth of information can be used for the fast, universal and accurate classification of phages. The Genome BLAST Distance Phylogeny (GBDP) approach is a truly whole-genome method currently used for in silico DNA: DNA hybridization and phylogenetic inference from prokaryotic genomes. Based on the principles of phylogenetic systematics, we here established GBDP for phage phylogeny and classification, using the common subset of genome-sequenced and officially classified phages. Trees inferred with the best GBDP variants showed only few deviations from the official phage classification, which were uniformly due to incorrectly annotated GenBank entries. Except for low resolution at the family level, the majority of taxa was well supported as monophyletic. Clustering genome sequences with distance thresholds optimized for the agreement with the classification turned out to be phylogenetically reasonable. Accordingly modifying genera and species is taxonomically optional but would yield more uniform sequence divergence as well as stronger branch support. Analysing an expanded data set containing > 4000 phage genomes from public databases allowed for extrapolating regarding the number, composition and host specificity of future phage taxa. The selected methods are implemented in an easy-to-use web service \"VICTOR\" freely available at http://ggdc.dsmz.de/victor.php.