gnomAD (Genome Aggregation Database)
Broad Institute
Allele counts from over 800,000 people, so you can tell a rare disease variant from harmless background noise.
gnomAD aggregates exome and genome sequencing and publishes allele frequencies by genetic ancestry group, never individual genotypes. Version 4 covers about 730,000 exomes and 76,000 genomes, and a long-read beta was announced in October 2026. Clinical geneticists use it to judge whether a variant is too common to cause a rare disease, and researchers use its gene-level constraint scores.
Access
- Cost
- FreeBrowser, downloads, and cloud copies (AWS, Google, Azure) at no charge.
- License
- Open licenseOpen Database License (ODbL)
- Sign-up
- No sign-up
- How you get it
- API and Bulk downloadVCF, Hail Table, TSV
What's in it
- One row is
- One row per variant, with counts and frequencies by ancestry group
- Covers
- Participants from studies worldwide, grouped into genetic ancestry groups
- Years
- 2016-present (ExAC predecessor from 2014)
- Updated
- Irregular
- Size
- About 807,000 individuals in v4; VCFs in the terabytes
Quality
The reference for population allele frequencies, with careful QC and versioned releases. Participants come from disease studies, with severe pediatric disease excluded, so it is not a random population sample.
- European-ancestry samples dominate
- Not a disease-free cohort
- No individual-level genotypes
Connects to
- 1000 Genomes Project and IGSR: Smaller open reference panel with individual genotypes
- GTEx Portal: Pair constraint scores with tissue expression for a gene
- AlphaFold Protein Structure Database: Map missense variants onto predicted protein structures
Questions it could answer
- Which genes tolerate almost no loss-of-function variants in healthy people?
- How does the frequency of a given variant differ across ancestry groups?