Additional file 2 of Taxonomy-aware, sequence similarity ranking reliably predicts phage–host relationships

Additional file 2: Table S1. Distribution of Phirbo, BLAST and WIsH scores among interacting and non-interacting virus-prokaryote pairs obtained from Edwards et al. and Galiez et al. data sets. Score ranges were summarized separately for 16,757 interacting and non-interacting virus-host pairs from Edwards et al., and 26,024 interacting and non-interacting virus-host pairs from Galiez et al. Table S2. Number of virus-host pairs evaluated by Phirbo, BLAST, and WIsH in Edwards et al. and Galiez et al. data sets. Table S3. Phages assigned by BLAST to multiple, equally-scored host species. Phirbo differentiated between host species and provided the highest score to primary host species. Table S4. Host prediction accuracy of Phirbo, BLAST, and WIsH over virus contig length. Table S5. Phage sequence coverage of 485 phages correctly assigned by BLAST and Phirbo to their host species. Sequence coverage was calculated for each phage as the sum of the lengths of its non-overlapping high scoring pairs to the genome of the correct host species, divided by the size of the query phage genome. Prophages were assumed to have sequence coverage greater than or equal to 30%. Table S6. Summary of taxonomic affiliations of 236 phages that had sequence coverage < 30% with the host species genomes. Table S7. Protein families present in sequence regions of 236 phage genomes that were used by BLAST and/or Phirbo in host prediction. The table provides information on each protein family (prokaryotic Virus Orthologous Group (pVOG)) used by BLAST and Phirbo, including: (i) pVOG description and functional assignment (manually curated), (ii) pVOG count (number of times a given pVOG was present in the phage genome, as well as in sequences used by BLAST or Phirbo), (iii) pVOG percentage (pVOG count divided by pVOG count in the genome), and (iii) P-value of pVOG enrichment. Table S8. RNA families present in sequence regions of 236 phage genomes that were used by BLAST and Phirbo in host prediction. The table provides information on each Rfam family used by BLAST and Phirbo. Table S9. Comparison of Phirbo’s host prediction performance between BLAST-based and Mash-based rankings of prokaryotic species.