Research
VHSeek: biology foundation model-based virus representations revealing virus host associations
The Brief
- VHSeek is a tool that uses DNA and protein sequence representations from a biology foundation model to predict viral hosts.
- It is designed to prioritize candidate virus-host associations across large viral collections.
- The work was published in Nature Communications on 20 November 2026.
Why it matters
Predicting which hosts viruses infect from sequence alone can help virologists and microbiologists prioritize candidate associations for follow-up study across large viral datasets.
Who’s affected
- virologists and microbiologists
- computational biologists working with viral sequence data
- #virus host prediction
- #foundation models
- #viral genomics
- #bioinformatics
- #protein sequences
Nature Communications, Published online: 20 November 2026; doi:10.1038/s41467-026-79392-x The authors present VHSeek, a tool that uses DNA and protein sequence representations to predict viral hosts and prioritize candidate virus-host associations across large viral collections.