Phage and virus biology · Software
CheckV
A command-line pipeline for assessing the completeness, contamination, and quality of viral genomes from metagenomes.
What CheckV is
CheckV is an automated pipeline designed to evaluate the quality of single-contig viral genomes recovered from metagenomic assemblies. Developed by Stephen Nayfach, Antonio Pedro Camargo, Frederik Schulz, and colleagues at the Lawrence Berkeley National Laboratory, it provides standard quality metrics for uncultivated virus genomes.
Metagenomic assembly often recovers viral sequences as incomplete fragments, chimeric sequences, or integrated proviruses flanked by host microbial genomic regions. CheckV addresses this challenge by identifying host contamination in integrated proviruses, estimating genome fragment completeness against reference databases, and detecting circular or closed complete genomes.
The tool categorises viral contigs into standardised quality tiers consistent with the Minimum Information about an Uncultivated Virus Genome (MIUViG) standards. This enables researchers to filter metagenome-assembled viruses, curate high-quality viral datasets, and report standardised quality metrics for comparative viral genomics.
How it works
CheckV executes four sequential modules. First, the contamination module annotates contig genes as viral or microbial using a custom database of hidden Markov models (HMMs). It slides adjacent gene windows from 5' to 3' to assess gene annotations and GC content, computing an intergenic score to predict host-virus integration breakpoints and delineate provirus boundaries. Second, the completeness module translates genes into proteins and matches them against the CheckV genome database using DIAMOND to measure average amino acid identity (AAI). Completeness is calculated as the ratio between contig length (or identified proviral length) and the matched reference length. For novel contigs lacking close AAI matches, CheckV provides an estimated completeness range representing a 90% confidence interval based on observed viral HMMs.
Third, the complete genome module detects direct terminal repeats (DTRs) of greater than 20 base pairs indicating circularity or concatamers, inverted terminal repeats (ITRs), or proviral regions flanked on both ends by host DNA, cross-referencing sequence length against estimated completeness to filter out assembly artefacts. Fourth, the pipeline aggregates these metrics into summary reports, assigning sequences into five quality tiers: Complete, High-quality (>90% completeness), Medium-quality (50–90% completeness), Low-quality (<50% completeness), or Undetermined quality.
Key facts
- Primary input format
- FASTA nucleotide sequences (.fna)
- Core dependencies
- DIAMOND (v2.2.0), HMMER (v3.3)
- Key output files
- quality_summary.tsv, completeness.tsv, contamination.tsv, complete_genomes.tsv
- Completeness methods
- AAI-based (point estimate) and HMM-based (90% confidence interval)
- Quality tiers
- Complete, High-quality, Medium-quality, Low-quality, Undetermined quality
What you can use it for
- Identification of host-provirus boundaries
- Detects host flanking regions on integrated proviral contigs and trims host contamination by finding genomic breakpoints based on gene annotations and GC content differences.
- Completeness estimation for uncultivated viral contigs
- Calculates the expected complete genome length and completeness percentage of viral contigs using average amino acid identity matches or profile HMMs.
- Detection of closed and circular viral genomes
- Identifies complete viral genomes by checking for direct terminal repeats, inverted terminal repeats, or dual host-flanked proviral integration sites verified by completeness estimates.
- Quality tiering for MIUViG compliance
- Assigns metagenome-assembled viral sequences to standardized quality tiers based on genome completeness to ensure reporting standards in viral metagenomics studies.
- Rapid pairwise ANI clustering
- Includes auxiliary scripts (anicalc and aniclust) to calculate pairwise average nucleotide identity via BLAST+ and cluster genomes at defined ANI and alignment fraction thresholds.
Run it here
Ask the research agent and it runs CheckV as a job on a dedicated machine, then reads the results back to you. Nothing to install, and it keeps running if you close the page.
Strengths and limits
Good at
- Provides automated trimming coordinates to accurately separate microbial host sequences from integrated proviruses.
- Combines average amino acid identity alignments and HMM profiling to estimate completeness across both well-characterised and novel viral lineages.
- Validates terminal repeats against estimated completeness lengths to prevent short metagenomic assembly artefacts from being misclassified as closed circular genomes.
- Allows users to extend or update the underlying reference database with newly discovered complete genomes using the checkv update_database command.
Watch out for
- Does not assess contamination on binned multi-contig viral metagenome-assembled genomes (MAGs) and requires contigs within a bin to be concatenated into a single sequence prior to completeness estimation.
- Cannot detect proviruses if host regions have already been excised by upstream tools such as VirSorter or VIBRANT.
- Assigns contigs with no reference AAI hits or viral HMM matches to 'Undetermined quality', which frequently occurs with short fragments or highly divergent novel lineages.
- Is not designed as a de novo virus classifier, requiring external tools such as geNomad for dedicated identification of viral contigs from whole metagenomes.
Questions
Can I run CheckV online without installing it?
Ask the research agent and it runs CheckV as a job on a dedicated machine, then reads the results back to you. Nothing to install, and it keeps running if you close the page.
What licence is CheckV under?
BSD-3-Clause-LBNL. The terms are published at https://bitbucket.org/berkeleylab/checkv.
Can CheckV be used to predict viruses from whole metagenomic assemblies?
CheckV evaluates viral quality and reports gene-level and genome-level viral signatures, but it is not intended for primary virus prediction; tools such as geNomad are recommended for de novo viral identification.
How do AAI-based and HMM-based completeness estimates differ?
AAI-based completeness calculates a single percentage by comparing the contig length to closely aligned reference genomes. HMM-based completeness provides a 90% confidence interval range based on viral gene markers when a contig lacks close database matches.
Why is a sequence labelled 'High-quality' when it contains substantial host contamination?
CheckV quality tiers are determined entirely by genome completeness rather than contamination, because host flanking sequence on proviruses can be readily trimmed using the predicted coordinates.
What causes a contig with a direct terminal repeat (DTR) to show less than 100% completeness?
This can occur when assembly artefacts produce terminal repeats on incomplete fragments, when completeness estimates fail on highly divergent circular sequences, or when analysing a complete segment of a multipartite genome.
What does the kmer_freq metric indicate in the quality summary?
The kmer_freq field indicates the number of times a viral genome is represented within the contig; values noticeably greater than 1.0 indicate assembly artefacts where the viral sequence has been concatenated into duplicate copies.
How to cite it
Nayfach, S., Camargo, A.P., Schulz, F. et al. CheckV assesses the quality and completeness of metagenome-assembled viral genomes. Nat Biotechnol 39, 578–585 (2021). https://doi.org/10.1038/s41587-020-00774-7
Written from CheckV's own README and project page. Statements citing a figure that does not appear there are removed before publication. Licence and running details come from our registry.