CADD (Combined Annotation Dependent Depletion)

Integrates 60+ genomic features into a single deleteriousness score. Scores ALL possible SNVs and small indels in the human genome, not just missense.

What It Does

  • Scores all variant types: missense, synonymous, intronic, intergenic, UTR, splice
  • PHRED-scaled score: 10 = top 10%, 20 = top 1%, 30 = top 0.1%
  • Integrates: conservation, regulatory, protein impact, splice predictions
  • Pre-computed for all possible SNVs in GRCh37/GRCh38

How to Use

Web

  1. Go to https://cadd.gs.washington.edu/
  2. Score tab → paste variant(s) in VCF format
  3. Or lookup pre-computed: https://cadd.gs.washington.edu/snv

API

# Score single variant (GRCh38)
curl "https://cadd.gs.washington.edu/api/v1.0/GRCh38/15:43600551:A:C"

Pre-computed Download

# Whole genome (350GB!)
# Or per-chromosome
tabix whole_genome_SNVs.tsv.gz 15:43600551-43600551

Verified Status

VERIFIED — STRC E1659A CADD PHRED score: 27.5 (raw: 4.931). Top 0.18% most deleterious variants genome-wide. Obtained via Ensembl VEP (direct CADD API returns ‘Not found’ for this position). SIFT: Deleterious (0), PolyPhen: Probably damaging (0.991).

STRC Research Usage

  • STRC E1659A Conservation and Reclassification — additional computational evidence
  • PHRED 27.5 = strong deleteriousness signal (>20 is commonly used pathogenicity threshold)
  • Accessed via: curl 'https://rest.ensembl.org/vep/human/region/15:43600551-43600551:1/C?CADD=1' -H 'Content-Type:application/json'

Connections