HelixbankELIXIR node

Controlled-access repository

Population-scale genomics, responsibly shared

1.4 million whole genomes and 3.1 million exomes from 41 contributing cohorts, available to approved researchers through a data access committee and an audited enclave.

 1.4 M genomes managed access analysis in place
1.4M
whole genomes
3.1M
exomes
84PB
archive size
41
contributing cohorts

How the archive works

Managed access, not open access

Every dataset has a data access committee. Applications are reviewed against the consent under which the samples were collected, and decisions are published in aggregate.

Analysis in place

The enclave puts compute next to the data. Most projects never export a single CRAM — they export a results table.

The enclave

Egress is audited

Where data does leave, it leaves through a reviewed export with a record of what, when, by whom and under which approval.

Standard formats

CRAM 3.1 against GRCh38 and T2T-CHM13, gVCF per sample, and joint-called VCF per cohort. No bespoke containers.

Beacon and cohort browser

Query allele presence without access, then apply for the cohorts that matter. Beacon v2 with the filtering terms extension.

Provenance and versioning

Every callset records the pipeline version, reference build and the exact sample manifest. Old callsets are never deleted.

Volume

Why we would rather you did not download it

A single 30× whole genome is a 40 GB CRAM. A modest study — five thousand cases and five thousand controls — is 400 TB before you have written a line of analysis.

  • One 30× WGS CRAM: 38–44 GB
  • One cohort of 10,000: ~410 TB
  • Joint-called VCF for the same cohort: 2.1 TB
  • The same analysis run in the enclave: 4 GB of results
CRAM alignments: 68%gVCF: 18%Raw signal and QC: 9%Joint callsets: 5%
CRAM alignments68
gVCF18
Raw signal and QC9
Joint callsets5

Frequently asked

Researchers at a recognised institution with an ethics approval covering the proposed use, and a signed data access agreement from the institution's legal representative — not from the researcher personally.