# gnomAD

License: permissive (CC0-1.0); restricted part present; Mdexa hosting: link-only. main part permissive; separable restricted part(s) recorded in license_parts

- Commercial use: Yes · Derivatives: Yes · Redistribution: Yes · File on Mdexa: No
- License status: Permissive (CC0-1.0)
- Mdexa hosting: Link only — License permissive, judged by public rules and checked by a second-pass model. Link only.
- Last verified: 2026-11-13
- Card: https://mdexa.com/datasets/gnomad/

## Key facts

- Domain: Genomics and omics
- Content type: Databases
- Size: —
- Language: English
- Modality: Tabular
- Publisher: The Broad Institute, Inc.
- Year: —

## Commercial use of gnomAD

- Commercial use: Yes.
The overall verdict takes the strictest part; the parts carry different licenses, listed below.
- License status: Permissive, with a restricted part.
- Conditions: Notice kept, No re-identification
- License part: primary data from the gnomAD exomes and genomes — Permissive (CC0-1.0)
- License part: Screenshots of the website — Permissive (NOASSERTION)
- License part: SpliceAI annotations — Restricted (CC-BY-NC-4.0)
Verdicts record what sources state and are not legal advice.

## gnomAD license evidence

main part permissive; separable restricted part(s) recorded in license_parts

- **Terms of use page**: Statement found in snapshot s01 (terms-page) — "All data here are released openly and publicly for the benefit of the wider biomedical community." — https://gnomad.broadinstitute.org/terms (2026-11-13; Page opened)
- **Terms of use page**: Statement found in snapshot s01 (terms-page) — "There are absolutely no restrictions or embargoes on the publication of results derived from gnomAD data." — https://gnomad.broadinstitute.org/terms (2026-11-13; Page opened)
- **Terms of use page**: Statement found in snapshot s01 (terms-page) — "All users of gnomAD data agree to not attempt to reidentify participants." — https://gnomad.broadinstitute.org/terms (2026-11-13; Page opened)
- **Terms of use page** — CC0-1.0: Statement found in snapshot s01 (terms-page) — "The primary data from the gnomAD exomes and genomes are available free of restrictions under the Creative Commons Zero Public Domain Dedication." — https://gnomad.broadinstitute.org/terms (2026-11-13; Page opened)
- **Terms of use page**: Statement found in snapshot s01 (terms-page) — "Screenshots of the website may also be used without restriction." — https://gnomad.broadinstitute.org/terms (2026-11-13; Page opened)
- **Terms of use page** — CC-BY-NC-4.0: Statement found in snapshot s01 (terms-page) — "Some annotations may have restrictions on usage. For instance, SpliceAI annotations have been computed by Illumina and are provided with permission under a CC BY NC 4.0 license for academic and non-commercial use SpliceAI." — https://gnomad.broadinstitute.org/terms (2026-11-13; Page opened)
- **Terms of use page**: Statement found in snapshot s01 (terms-page) — "we ask that you refrain from incorporating "gnomAD" or "Genome Aggregation Database'' into the name of your tool and from using the gnomAD logo without permission." — https://gnomad.broadinstitute.org/terms (2026-11-13; Page opened)
- **Terms of use page**: Statement found in snapshot s01 (terms-page) — "gnomAD™ is a trademark owned by The Broad Institute, Inc." — https://gnomad.broadinstitute.org/terms (2026-11-13; Page opened)
- **Terms of use page**: Statement found in snapshot s01 (terms-page) — "It is the responsibility of users to abide by all relevant licensing requirements." — https://gnomad.broadinstitute.org/terms (2026-11-13; Page opened)
- **Terms of use page**: Statement found in snapshot s01 (terms-page) — "Some annotations may have restrictions on usage." — https://gnomad.broadinstitute.org/terms (2026-11-13; Page opened)
- **Terms of use page**: Statement found in snapshot s01 (terms-page) — "For instance, SpliceAI annotations have been computed by Illumina and are provided with permission under a CC BY NC 4.0 license for academic and non-commercial use SpliceAI." — https://gnomad.broadinstitute.org/terms (2026-11-13; Page opened)

## Source and lineage

—

## Known issues with gnomAD

- Variant calling and filtering from short-read sequencing data is an imperfect and probabilistic process, so many errors may remain.

## Preflight

Not run (data not ingested).

## Downloading gnomAD and API access

Mdexa does not serve files. Original source:
- https://gnomad.broadinstitute.org/
- https://gnomad.broadinstitute.org/policies

### Access recipe

- Latest: 4.1.1
- chr1 sites VCF (Google) · https://storage.googleapis.com/gcp-public-data--gnomad/release/4.1.1/vcf/exomes/gnomad.exomes.v4.1.1.sites.chr1.vcf.bgz
- chr1 sites VCF (Amazon) · https://gnomad-public-us-east-1.s3.amazonaws.com/release/4.1.1/vcf/exomes/gnomad.exomes.v4.1.1.sites.chr1.vcf.bgz
- chr1 sites VCF index (Google) · https://storage.googleapis.com/gcp-public-data--gnomad/release/4.1.1/vcf/exomes/gnomad.exomes.v4.1.1.sites.chr1.vcf.bgz.tbi
- Bulk download: `gsutil ls gs://gcp-public-data--gnomad/release/`
- Bulk download: `aws s3 ls s3://gnomad-public-us-east-1/release/`

```sh
curl -L --create-dirs -o "gnomad.exomes.v4.1.1.sites.chr1.vcf.bgz" "https://storage.googleapis.com/gcp-public-data--gnomad/release/4.1.1/vcf/exomes/gnomad.exomes.v4.1.1.sites.chr1.vcf.bgz"
curl -L --create-dirs -o "gnomad.exomes.v4.1.1.sites.chr1.vcf.bgz.tbi" "https://storage.googleapis.com/gcp-public-data--gnomad/release/4.1.1/vcf/exomes/gnomad.exomes.v4.1.1.sites.chr1.vcf.bgz.tbi"
```

## Citing gnomAD

No recommended citation text.

## Machine-readable

- JSON: https://mdexa.com/datasets/gnomad.json
- 2026-11-14 Known defects, version notes and access statements from cited sources (papers, release notes, official pages) were added; each shows who said it, when, and what kind of source it is.
- 2026-11-14 Practical fields (scope, identifiers, access methods, versions, known defects) were added by the automated pipeline from stored page snapshots; every statement carries a verbatim quote.
- 2026-11-14 Known defects, version notes and access statements from cited sources (papers, release notes, official pages) were added; each shows who said it, when, and what kind of source it is.
- 2026-11-13 Listed on 2026-11-13. Judged by public rules and checked a second time before listing. This part could not be retrieved: the later sections of the gnomAD terms page (privacy policy draft, ethics policy, data preservation, data generation, low-frequency allele count statement, citation in publications). The conclusion rests on the part that was retrieved, which holds the CC0 statement and the notes on separate parts.

## Practical information

### Content scope

- The v4 release includes data from 807,162 total individuals.
  > Today, we are delighted to announce the release of gnomAD v4, which includes data from 807,162 total individuals.
  https://gnomad.broadinstitute.org/news/2023-11-gnomad-v4-0/ (2026-11-13)
- The v4 data set contains data from 730,947 exomes and 76,215 whole genomes, all mapped to the GRCh38 reference sequence.
  > The gnomAD v4 data set contains data from 730,947 exomes and 76,215 whole genomes, all mapped to the GRCh38 reference sequence.
  https://gnomad.broadinstitute.org/downloads (2026-11-13)
- The combined filtering allele frequency integrates variant allele frequencies across 734,947 exomes and 76,215 genomes.
  > For the first time, in gnomAD v4.0, we released a combined filtering allele frequency (FAF), integrating variant allele frequencies across the 734,947 exomes and 76,215 genomes.
  https://gnomad.broadinstitute.org/news/2024-04-gnomad-v4-1/ (2026-11-13)
- The v4.1 callset includes 416,555 UK Biobank and 314,392 non-UK Biobank exomes.
  > The gnomAD v4.1 callset includes 416,555 UK Biobank and 314,392 non-UK Biobank exomes, requiring the use of two calling interval lists for processing and quality control pipelines.
  https://gnomad.broadinstitute.org/news/2024-04-gnomad-v4-1/ (2026-11-13)

### Identifiers

- The page shows a variant identifier written as 1-55039909-TC-T in gnomAD v4.1.
  > This variant maps to 1-55039909-TC-T in gnomAD v4.1, where it still has an allele count of 1 (still present in only a single exome sample).
  https://gnomad.broadinstitute.org/news/2024-04-gnomad-v4-1/ (2026-11-13)
- The page gives 1-55505582-TC-T as an example of a variant present in gnomAD v2.1.1 only in the v2.1.1 exomes.
  > For example, 1-55505582-TC-T in gnomAD v2.1.1 is present only in the v2.1.1 exomes, and therefore the total AN at this site only contains information from the exomes.
  https://gnomad.broadinstitute.org/news/2024-04-gnomad-v4-1/ (2026-11-13)

### Access

- Bulk download: https://gnomad.broadinstitute.org/downloads
- Database dump: https://gnomad.broadinstitute.org/downloads
- Web interface only: https://raw.githubusercontent.com/broadinstitute/gnomad-browser/main/browser/about/policies/terms.md

### Versions

- v4.1 (19 April 2024) fixed the v4.0 allele-number issue and added joint allele number across all callable sites plus a flag for discordant exome/genome frequencies.
  > This update fixes the allele number issue in gnomAD v4.0 previously described here and critically adds two new functionalities: joint allele number across all called sites in the exomes and genomes and a new flag indicating when the frequencies in the exomes and genomes are highly discordant.
  https://gnomad.broadinstitute.org/news/2024-04-gnomad-v4-1/ (2026-11-14)
- The v4 release follows a model of incremental releases, with v4.0 a minimum viable product release containing only the most critical components of the database.
  > In the v4 release, we have moved towards the model of incremental releases. This means that the v4.0 release is a minimum viable product (MVP) release, or a release that contains only the most critical components of the gnomAD database.
  https://gnomad.broadinstitute.org/news/2023-11-gnomad-v4-0/ (2026-11-13)
- The v4.0 release does not have all the features of the v2 or v3 releases; those features are to be released in subsequent minor releases.
  > This means that the v4.0 release will not have all of the features of the v2 or v3 releases, and these features will be released in subsequent minor releases.
  https://gnomad.broadinstitute.org/news/2023-11-gnomad-v4-0/ (2026-11-13)
- The gnomAD v3 release had five subsets, while gnomAD v4 has a single subset (non-UKBB).
  > The gnomAD v3 release had five subsets, whereas gnomAD v4 has a single subset (non-UKBB; see Dataset selection ).
  https://gnomad.broadinstitute.org/news/2023-11-gnomad-v4-0/ (2026-11-13)
- The v4.1 update fixes the allele number issue in gnomAD v4.0 previously described on the page.
  > This update fixes the allele number issue in gnomAD v4.0 previously described here
  https://gnomad.broadinstitute.org/news/2024-04-gnomad-v4-1/ (2026-11-13)
- v4 removed support for cohort subsets (such as non-disease subsets) from the exomes, partly because phenotype metadata are lacking for biobank samples.
  > We have removed support for subsets from the gnomAD v4 exomes for two major reasons: given the large size of gnomAD v4 and the small size of any single cohort, we do not feel that the prevalence of any disease is high enough to warrant subsets.
  https://gnomad.broadinstitute.org/news/2023-11-gnomad-v4-0/ (2026-11-14)
- Variants from the v3.1 genomes were brought into v4 and relabelled as v4 genomes.
  > Variants from the genomes in v3.1 were brought directly into v4 and are now labeled v4 genomes.
  https://raw.githubusercontent.com/broadinstitute/gnomad-browser/main/browser/help/topics/variant-qc.md (2026-11-14)

### Known defects and misuse

- The page states that variant calling and filtering from short-read sequencing data is an imperfect and probabilistic process, so many errors no doubt remain.
  > This data set has been subjected to extensive quality control, but variant calling and filtering from short-read sequencing data is an imperfect and probabilistic process, so many errors no doubt remain.
  https://gnomad.broadinstitute.org/policies (2026-11-13)
- The 76,156 genomes from v3.1.2 included in v4 did not have their quality control updated, although some frequencies within the HGDP/1KG subset were updated.
  > The 76,156 genomes from gnomAD v3.1.2 have been included in gnomAD v4. We have not updated the quality control (QC) of these samples, but we have updated some of the frequencies within the HGDP and 1000 Genomes (HGDP/1KG) subset .
  https://gnomad.broadinstitute.org/news/2023-11-gnomad-v4-0/ (2026-11-13)
- The page states that combining the datasets introduces challenges related to datasets that differ in sequencing and processing methods and exhibit variations in sample composition due to ascertainment biases.
  > However, it also introduces challenges related to integrating datasets that not only differ in their sequencing and processing methods, but also exhibit variations in sample composition due to ascertainment biases.
  https://gnomad.broadinstitute.org/news/2024-04-gnomad-v4-1/ (2026-11-13)
- The page states that indels generally tend to have less stable frequency estimates, likely because most indels in gnomAD tend to occur in low-complexity regions or in regions overlapping a segmental duplication.
  > In addition, indels generally tend to have less stable frequency estimates 1 , likely because most of the indels in gnomAD tend to occur in low-complexity regions or in regions overlapping a segmental duplication (see flags ).
  https://gnomad.broadinstitute.org/news/2024-04-gnomad-v4-1/ (2026-11-13)
- Gudmundsson et al.: genes with close paralogs, pseudogenes or segmental duplications may have short-read mapping problems that inflate observed variant counts.
  > However, genes with highly similar paralogs, pseudogenes, or segmental duplications may have mapping challenges with short read data that can inflate observed variant counts.
  https://pmc.ncbi.nlm.nih.gov/articles/PMC9160216/ (2026-11-14)
- An allele count of 0 or missing can mean no data (e.g. outside exome capture), no alternate calls, or no high-quality calls.
  > There are three reasons why a variant will have an allele count of 0 or an undefined (missing) allele count (indicated by `-`):
  https://raw.githubusercontent.com/broadinstitute/gnomad-browser/main/browser/help/topics/allele-count-zero.md (2026-11-14)

### Suited for

- The page states that users can freely download and search the data and encourages the use and publication of results generated from these data.
  > You can freely download and search the data, and we encourage the use and publication of results generated from these data.
  https://gnomad.broadinstitute.org/policies (2026-11-13)
- The page states that in use cases involving querying a relatively small number of records with low latency, using the API to request records as needed may be more suitable than working with downloaded gnomAD data.
  > In use cases involving querying a relatively small number of records with low latency, using the API to request records as needed may be more suitable than working with downloaded gnomAD data.
  https://gnomad.broadinstitute.org/downloads (2026-11-13)
- The filtering allele frequency (FAF) is intended for filtering variants against a disease-specific frequency threshold (e.g. ACMG/AMP BA1); it is not computed for bottlenecked groups (ami, asj, fin, rmi).
  > This annotation can be used for filtering variants by allele frequency against a disease-specific threshold that can be set for each disease (e.g. BA1 in the 2015 ACMG/AMP guidelines).
  https://raw.githubusercontent.com/broadinstitute/gnomad-browser/main/browser/help/topics/faf.md (2026-11-14)

### Not suited for

- The page states that, given the large size of gnomAD v4 and the small size of any single cohort, it does not consider the prevalence of any disease high enough to warrant subsets.
  > given the large size of gnomAD v4 and the small size of any single cohort, we do not feel that the prevalence of any disease is high enough to warrant subsets.
  https://gnomad.broadinstitute.org/news/2023-11-gnomad-v4-0/ (2026-11-13)
- The page states that for querying a relatively small number of records with low latency, working with downloaded gnomAD data may be less suitable than using the API.
  > In use cases involving querying a relatively small number of records with low latency, using the API to request records as needed may be more suitable than working with downloaded gnomAD data.
  https://gnomad.broadinstitute.org/downloads (2026-11-13)

### No sourced statement yet

Commercial licensing

## Continue in this database

This record is one of 109 processed and verified records (datasets, data resources and open models). To find others, call `https://mcp.mdexa.com/v1/datasets?q=<words>` or connect the MCP server at `https://mcp.mdexa.com/mcp`; no key is needed for 20 calls per minute. Rows are under 1 KB; an empty result means "not listed". For a list of items, one batch check (free account) returns a report. Guide: /agents.md. Similar records: AlphaFold Protein Structure Database (https://mdexa.com/datasets/alphafold-db/); GEO (https://mdexa.com/datasets/geo/); ClinPGx/PharmGKB (https://mdexa.com/datasets/pharmgkb/).