genomics-pipeline

Step 13: Variant Effect Predictor (VEP) Annotation

What This Does

Annotates every variant in the VCF with gene name, consequence type, predicted impact, pathogenicity scores (SIFT, PolyPhen), and population allele frequencies. This is the most comprehensive single annotation step in the pipeline.

Why

Raw VCF variants are just genomic coordinates and genotypes. VEP transforms them into biologically interpretable annotations — which gene is affected, what the functional consequence is, how rare the variant is in the population, and whether it is predicted damaging.

Tool

• Ensembl VEP release 112 (European Bioinformatics Institute)

Docker Image

ensemblorg/ensembl-vep:release_112.0

Prerequisites

• Offline VEP cache must be downloaded first (see step 00-reference-setup)
• Cache size: ~26 GB for GRCh38 homo_sapiens

Command

SAMPLE=your_sample
GENOME_DIR=/path/to/your/data

docker run --rm \
  -v ${GENOME_DIR}/${SAMPLE}/vcf:/data \
  -v ${GENOME_DIR}/vep_cache:/vep_cache \
  ensemblorg/ensembl-vep:release_112.0 \
  vep \
    --input_file /data/${SAMPLE}.vcf.gz \
    --output_file /data/${SAMPLE}_vep.vcf \
    --vcf \
    --offline \
    --cache \
    --dir_cache /vep_cache \
    --assembly GRCh38 \
    --fork 4 \
    --sift b \
    --polyphen b \
    --af \
    --af_gnomade \
    --canonical \
    --symbol \
    --force_overwrite

# Output: VCF with CSQ INFO field containing all annotations

Output Format

• Default: VCF with CSQ INFO field (pipe-delimited sub-fields)
• Alternative: add --tab instead of --vcf for tab-delimited output (easier to parse manually)
• Key CSQ sub-fields: SYMBOL, Consequence, IMPACT, SIFT, PolyPhen, gnomADe_AF, CANONICAL

Filtering for Clinical Relevance

After annotation, filter to actionable variants:

# Extract HIGH and MODERATE impact variants with AF <1%
docker run --rm \
  -v ${GENOME_DIR}/${SAMPLE}/vcf:/data \
  staphb/bcftools:1.21 \
  bcftools view -i 'INFO/CSQ[*] ~ "HIGH" || INFO/CSQ[*] ~ "MODERATE"' \
    /data/${SAMPLE}_vep.vcf \
    > /data/${SAMPLE}_vep_filtered.vcf

Important Notes

• Full WGS annotation takes 2-4 hours depending on CPU and variant count (~5M variants)
• --fork 4 enables parallelism — increase if more cores are available
• --canonical restricts to canonical transcripts, reducing redundant annotations per variant
• --sift b and --polyphen b output both prediction and score (e.g., deleterious(0.01))
• --af_gnomade adds gnomAD exomes frequency — use --af_gnomad (without ‘e’) for gnomAD genomes
• Variants with IMPACT=HIGH and gnomAD AF <0.01 (or absent) are the highest priority for clinical review
• VEP does NOT assess variant pathogenicity in ClinVar context — combine with ClinVar annotation for full picture

This site is open source. Improve this page.