Variant Annotation

intermediate

Annotation Database Landscape

Key Annotation Databases by Category

📊Population FreqgnomAD v4.1, 1000G, TOPMed, UK Biobank

🏥ClinicalClinVar, OMIM, HGMD, ClinGen, DECIPHER

🎯Pathogenicity ScoresCADD, REVEL, SIFT, PolyPhen, AlphaMissense

🧬Gene FunctionrefGene, GENCODE, Ensembl, UniProt

✂️SplicingSpliceAI, dbscSNV, MaxEntScan

🔬CancerCOSMIC, CIViC, OncoKB, TCGA

What Annotation Adds

Gene/transcript: which gene, exon number, cDNA/protein change (e.g., BRCA1:c.5266dupC:p.Gln1756Profs*74)
Functional consequence: synonymous, missense, frameshift, stop-gained, splice-site
Population frequency: gnomAD allele frequency in NFE, AFR, EAS, SAS, AMR populations
Clinical significance: ClinVar classification (Pathogenic/VUS/Benign), OMIM disease
Pathogenicity scores: SIFT (<0.05=deleterious), PolyPhen-2 (>0.85=probably damaging), CADD (>20=top 1%), REVEL (>0.5=likely pathogenic)
Conservation: PhyloP, PhastCons (high score = evolutionarily conserved = likely important)
Splicing prediction: SpliceAI delta score >0.5 = likely splicing impact

ANNOVAR

code

# Direct VCF input (recommended)
table_annovar.pl sample.vcf humandb/ \
  --vcfinput \
  --buildver hg38 \
  --out sample_annotated \
  --remove \
  --protocol refGene,dbnsfp47a,clinvar_20240611,gnomad211_exome,avsnp150,cosmic70 \
  --operation gx,f,f,f,f,f \
  --nastring . \
  --thread 8 \
  --xreffile gene_fullxref.txt

# Protocol operations:
# g = gene-based (refGene, ensGene, knownGene)
# f = filter-based (frequency/pathogenicity databases)
# r = region-based (regulatory, conserved regions)

VEP (Ensembl)

code

# Offline mode (fastest)
vep \
  --input_file sample.vcf \
  --output_file annotated.vcf \
  --vcf \
  --offline \
  --assembly GRCh38 \
  --cache \
  --everything \
  --fork 8 \
  --stats_html vep_stats.html

# --everything enables:
# --sift b --polyphen b --ccds --uniprot --hgvs
# --symbol --numbers --domains --regulatory
# --canonical --protein --biotype --af --af_1kg
# --af_gnomad --max_af --pubmed --var_synonyms

Annotation Databases (hg38)

refGene / GENCODE / Ensembl - gene models; use GENCODE v45 for comprehensive transcripts
dbnsfp47a - 30+ pathogenicity scores in one file (SIFT, PolyPhen, CADD, REVEL, AlphaMissense)
gnomad211_exome - 125,748 exomes; allele frequencies in 8 populations
clinvar_20240611 - 2.6M variants with clinical interpretations
cosmic70 - 30M somatic mutations from 1.5M tumour samples
spliceai_filtered - pre-computed SpliceAI delta scores for all SNVs
intervar - automated ACMG/AMP classification for missense variants

Understanding VCF Files Population & Clinical Databases