Among the 848 non-redundant virophage genomes, several categories of sequences were considered as likely representing complete and near-complete genomes: (i) reference genomes from isolates (n = 4), (ii) sequences identified as integrated with upstream and downstream host regions ≥ 2 kb (n = 7), (iii) sequences with direct or inverted terminal repeats (n = 118 and n = 8, respectively), (iv) sequences predicted to be ≥90% complete based on CheckV (AAI-based prediction, n = 59), and (v) linear contigs ≥ 25 kb (n = 61). This latter category was based on the median length of predicted complete and near-complete genomes from all other categories (25,168 bp). Overall, 257 sequences were considered complete or near-complete virophage genomes.
These complete and near-complete genomes were used as input for phylogenetic trees and genome-wide clustering to establish groups and potential taxa within the virophages. For phylogenetic trees, the sequences of the four morphogenesis genes detected in the 257 complete and near-complete genomes using the new HMM profiles (see above) were used after excluding all sequences that covered <60% of the HMM profile to remove partial gene predictions. Multiple alignments were then built for each gene using an iterative clustering-alignment-phylogeny procedure specifically adapted for aligning highly diverging sequences [54 (link)]. The alignments were then automatically trimmed using clipkit v1.3.0 [55 (link)] using the kpi-smart-gap mode to remove uninformative positions, and the trimmed alignments were used as input for tree building with IQ-Tree v2.2.0.3 [56 (link)] with automatic detection of the most appropriate substitution matrix, and 1000 replicates of ultra-fast bootstraps. The best-fit model was Q.pfam+F+R7 for PRO, Q.yeast+F+R8 for ATPase, and Q.pfam+F+R8 for both MCP and penton. For the larger MCP phylogeny, including both complete and partial virophage genomes (Figure S6 ), multiple alignments were computed with MAFFT v7.490 based on the curated multiple alignment including MCP from complete and near-complete genomes only (options “–add” and “–keeplength”) [51 (link)], and the phylogeny was built with tree IQ-Tree v2.2.0.3 [56 (link)] with similar parameters as described above.
Genome-wide amino acid identity (AAI) clustering was performed as in [57 (link)]. Briefly, predicted protein sequences from the 257 complete and near-complete virophages were compared all-vs-all using diamond v0.9.24.125 [58 (link)] and the following options: “--evalue 1e-5 --max-target-seqs 10,000–query-cover 50–subject-cover 50”. The resulting file was used as input for the script “amino_acid_identity.py” to calculate the average AAI for all pairs of genomes. The script “filter_aai.py” was then used to select only pairs of genomes with a minimum normalized cumulative bit score of 0.05. Finally, these selected pairwise AAI values were used as input for an MCL clustering using MCL 14-137 (inflation parameter = 1.1) [50 ].
These complete and near-complete genomes were used as input for phylogenetic trees and genome-wide clustering to establish groups and potential taxa within the virophages. For phylogenetic trees, the sequences of the four morphogenesis genes detected in the 257 complete and near-complete genomes using the new HMM profiles (see above) were used after excluding all sequences that covered <60% of the HMM profile to remove partial gene predictions. Multiple alignments were then built for each gene using an iterative clustering-alignment-phylogeny procedure specifically adapted for aligning highly diverging sequences [54 (link)]. The alignments were then automatically trimmed using clipkit v1.3.0 [55 (link)] using the kpi-smart-gap mode to remove uninformative positions, and the trimmed alignments were used as input for tree building with IQ-Tree v2.2.0.3 [56 (link)] with automatic detection of the most appropriate substitution matrix, and 1000 replicates of ultra-fast bootstraps. The best-fit model was Q.pfam+F+R7 for PRO, Q.yeast+F+R8 for ATPase, and Q.pfam+F+R8 for both MCP and penton. For the larger MCP phylogeny, including both complete and partial virophage genomes (
Genome-wide amino acid identity (AAI) clustering was performed as in [57 (link)]. Briefly, predicted protein sequences from the 257 complete and near-complete virophages were compared all-vs-all using diamond v0.9.24.125 [58 (link)] and the following options: “--evalue 1e-5 --max-target-seqs 10,000–query-cover 50–subject-cover 50”. The resulting file was used as input for the script “amino_acid_identity.py” to calculate the average AAI for all pairs of genomes. The script “filter_aai.py” was then used to select only pairs of genomes with a minimum normalized cumulative bit score of 0.05. Finally, these selected pairwise AAI values were used as input for an MCL clustering using MCL 14-137 (inflation parameter = 1.1) [50 ].
Free full text: Click here
