vcontact

name	vcontact
description	vConTACT2 / vConTACT3 — gene-sharing network and ML-based hierarchical virus taxonomy for prokaryotic (and in v3, broader realm) viral genomes/contigs. Use after viral discovery (geNomad / VirSorter2) and quality (CheckV): (1) Genus-to-order style clustering, (2) Placing vMAGs vs RefSeq/ICTV, (3) Choosing v2 vs v3. Upstream: https://bitbucket.org/MAVERICLab/vcontact2 and https://bitbucket.org/MAVERICLab/vcontact3.
license	MIT
category	analysis-tools
tags	["virus","taxonomy","vConTACT","gene-sharing","ICTV"]
upstream	https://bitbucket.org/MAVERICLab/vcontact3
stage	mining

vConTACT2 / vConTACT3

Upstream: vConTACT2 · vConTACT3

Citation

Jang, H. B. et al. Taxonomic assignment of uncultivated prokaryotic virus genomes is enabled by gene-sharing networks. Nat. Biotechnol. 37, 632–639 (2019). https://doi.org/10.1038/s41587-019-0100-8

Bolduc, B. et al. Machine learning enables scalable and systematic hierarchical virus taxonomy. Nat. Biotechnol. (2025). https://doi.org/10.1038/s41587-025-02946-9

See also docs/references.md.

Analytical thinking

Viruses lack a universal marker gene. vConTACT builds gene-sharing networks (v2) and, in v3, ML hierarchical taxonomy with realm-aware cut-offs, scaling to large metagenomic viral sets and improving agreement with ICTV-style ranks.

Pipeline position:

contigs → genomad | virsorter2 → checkv (quality filter)
       → proteins (Prodigal/Bakta as required by vConTACT)
       → vcontact3 (prefer for new work) | vcontact2

Edition	Prefer when
vConTACT3	New analyses; hierarchical ranks; broader realms; scale
vConTACT2	Reproducing older Methods / published networks

Network clusters ≠ wet-lab host range. Report reference DB (Viral RefSeq / ICTV snapshot) and software version.

How to run

# follow Bitbucket/Apptainer docs for your edition — CLI differs by version
# typical inputs: viral genomes/contigs + predicted proteins + reference set

Decision tree

Viral sequences in hand?
├─ Not yet discovered → genomad / virsorter2
├─ Need quality → checkv
├─ Need taxonomy / clustering → vcontact (v3 preferred)
└─ Only presence of viral contigs → stop at genomad/checkv

Related skills

genomad · virsorter2 · checkv · microbial-mining · prodigal · tool-selection