A novel approach to Caudoviricetes taxonomy utilising whole proteome structure-structure comparison
Smith, L.; Schmidt, T. S.; Tolstoy, I.; Skvortsov, T.; Hill, C.; Shkoporov, A. N.
Show abstract
Viral proteins often evolve so rapidly that sequence similarity is lost even among conserved functional genes, complicating phylogenetic analysis and classification. This challenge is acute for Caudoviricetes, a highly diverse class of dsDNA tailed bacteriophages recently redefined by the International Committee on Taxonomy of Viruses (ICTV). To address the limitations of sequence-based taxonomy, we present a structure-based classification framework using whole-proteome structural clustering. From 4,082 exemplar genomes, we predicted 445,098 protein structures with ESMFold and clustered them using Foldseek. Genomes were encoded as binary profiles of structural fold presence, and phylogenetic distances were inferred and normalized using Relative Evolutionary Divergence (RED). This yielded a revised structural taxonomy comprising 159 orders, 267 families, 502 subfamilies, and 1,189 genera. We also introduce PhagePleats, a Python tool for classifying novel phage genomes based on structural similarity. Our approach highlights the utility of protein structure for resolving distant viral relationships.
Matching journals
The top 8 journals account for 50% of the predicted probability mass.
Similar papers in this journal
- Genomic transfers help to decipher the ancient evolution of filoviruses and interactions with vertebrate hosts 93%
- Flexibility and modulation of translation initiation in enterovirus genomes 93%
- Detailed single-cell mapping of the transcriptional response to a virus infection driven by copy-back viral genomes 93%
Similar papers in this journal
- RNA genome expansion up to 64 kb in nidoviruses is host constrained and associated with new modes of replicase expression 95%
- Diversification of giant and large eukaryotic dsDNA viruses predated the origin of modern eukaryotes 94%
- Genomic determinants of pathogenicity in SARS-CoV-2 and other human coronaviruses 93%
"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.