vep

TAFFISH wrapper for Ensembl Variant Effect Predictor (VEP) 116.0, the Ensembl runtime for predicting functional consequences of genomic variants and running related VEP helper tools.

Package identity:

• name: vep
• command: taf-vep
• version: 116.0-r1
• kind: tool
• image: ghcr.io/taffish/vep:116.0-r1
• upstream: Ensembl VEP tag release/116.0
• runtime version: ensembl-vep 116.0; cache version should be 116
• app license: Apache-2.0
• upstream license: Apache-2.0

What Is Packaged

This app uses the official Ensembl VEP Docker runtime ensemblorg/ensembl-vep:release_116.0, pinned by OCI manifest digest. The runtime is built by Ensembl from the Ensembl/ensembl-vep source tree and includes VEP, Ensembl API modules, maintained VEP plugins, htslib tools, Bio::DB::HTS, Bio::DB::BigFile support, Haplosaurus dependencies, and GeneSplicer plugin support.

The container exposes:

• vep: default Variant Effect Predictor command
• filter_vep: filter VEP tabular or VCF/CSQ output
• haplo: Haplosaurus phased haplotype consequence tool
• variant_recoder: translate variant identifiers and HGVS-like encodings
• INSTALL.pl: Ensembl VEP installer for caches, FASTA, API, and plugins
• bgzip, tabix, htsfile, perl, curl, python, python2

Usage

Show TAFFISH wrapper help:

taf-vep --help

Show upstream VEP help and runtime version banner:

taf-vep -- --help
taf-vep vep --help

vep --version is not an upstream VEP option. The upstream version is reported at the top of vep --help as ensembl-vep : 116.0.

Annotate with an already prepared cache:

taf-vep -- --cache --offline --dir_cache "$PWD/vep-data" \
  --species homo_sapiens --assembly GRCh38 \
  --format vcf --vcf --force_overwrite \
  -i variants.vcf -o variants.vep.vcf

Use a custom GFF3/GTF and FASTA instead of an Ensembl cache:

taf-vep -- --gff annotation.gff3.gz --fasta genome.fa.gz \
  --format vcf --vcf --force_overwrite \
  -i variants.vcf -o variants.vep.vcf

Run helper commands in the same app environment:

taf-vep filter_vep --format vcf --filter "Consequence is missense_variant" \
  -i variants.vep.vcf -o missense.vcf --force_overwrite

taf-vep haplo --help
taf-vep variant_recoder --help
taf-vep INSTALL.pl --help

Cache and FASTA Data

This app intentionally does not bundle Ensembl VEP caches, FASTA files, dbNSFP, ClinVar, gnomAD, CADD, AlphaMissense, SpliceAI, or other large annotation data. Those resources are species-, assembly-, release-, and license-dependent.

For VEP 116.0, use Ensembl VEP cache version 116. Prepare a persistent cache directory in your project or another host-visible location:

mkdir -p vep-data
taf-vep INSTALL.pl -a cf -s homo_sapiens -y GRCh38 -c "$PWD/vep-data"

Then run VEP against that directory:

taf-vep -- --cache --offline --dir_cache "$PWD/vep-data" \
  --species homo_sapiens --assembly GRCh38 \
  -i variants.vcf -o variants.vep.txt --force_overwrite

The installer downloads data from Ensembl and therefore requires network access. Flow-oriented or reproducible production runs should pre-populate the cache directory and run VEP with --offline. If you use external plugin data or custom annotations, keep those files in explicit project/reference directories and pass the corresponding VEP options.

Command Mode

Command mode is enabled. Option-leading arguments go to the default vep command through taf-vep -- ...; explicit commands run inside the same container:

taf-vep -- --help
taf-vep vep --help
taf-vep filter_vep --help
taf-vep perl -MBio::EnsEMBL::VEP::Constants -e 'print "ok\n"'

Platform

The image is built for native linux/amd64 and linux/arm64. The packaged runtime comes from the official Ensembl multi-architecture Docker image.

Boundaries

This app packages the VEP runtime and helper tools. It does not package production annotation caches, reference FASTA bundles, Ensembl MySQL mirrors, Nextflow, Docker/Singularity itself, or third-party databases required by specific plugins. Commands such as INSTALL.pl -a c, plugin data downloads, remote database mode, and variant_recoder database lookups may need network access unless all required data is already local.

Smoke tests validate the VEP 116.0 runtime banner, helper command help, Perl module availability, htslib utilities, filter_vep on a synthetic VEP-CSQ VCF, and a real offline VEP annotation path using tiny synthetic GFF3, FASTA, and VCF files. Smoke does not download production caches or validate biological annotation correctness on real genomes.

Upstream

• homepage: https://www.ensembl.org/info/docs/tools/vep/index.html
• repository: https://github.com/Ensembl/ensembl-vep
• tag: https://github.com/Ensembl/ensembl-vep/tree/release/116.0
• Docker image: https://hub.docker.com/r/ensemblorg/ensembl-vep
• command-line documentation: https://www.ensembl.org/info/docs/tools/vep/script/index.html
• cache documentation: https://www.ensembl.org/info/docs/tools/vep/script/vep_cache.html
• upstream license: Apache-2.0

If you use Ensembl VEP in academic work, cite:

McLaren W, Gil L, Hunt SE, Riat HS, Ritchie GRS, Thormann A, Flicek P, Cunningham F. The Ensembl Variant Effect Predictor. Genome Biology. 2016;17:122. doi: 10.1186/s13059-016-0974-4, PMID: 27268795.