Genomics

Genomic sequencing produces data at successive stages (raw reads, aligned sequences, and processed variants or expression matrices), each with its own formats, repositories, and governance.

Standards

The core sequence and variant formats (SAM-BAM-CRAM, VCF, VRS, Phenopackets) are standardised by GA4GH, which also provides data access and interoperability APIs adopted by controlled-access repositories including EGA and dbGaP.

Data archives

Genomic data deposits follow the processing pipeline, and a few conventions cover most cases. Raw reads in FASTQ format are deposited in the INSDC (International Nucleotide Sequence Database Collaboration) partner archives: ENA (Europe), SRA (USA), and DDBJ (Japan), which synchronise their holdings daily. Human raw reads that cannot be openly released, because of re-identification risk, usually go to a controlled-access archive instead, EGA (Europe) or dbGaP (USA). Aligned reads in SAM-BAM-CRAM follow the same open or controlled-access split as the raw reads they derive from. Processed variants in VCF can often be shared as anonymised summary statistics, commonly through EVA (Europe) or dbSNP (USA). Large structural variants, usually those over 50 bp, are often archived separately in dbVar (USA), the structural-variation counterpart to dbSNP. Clinical significance classifications are curated in ClinVar, with expert panel assessments from ClinGen. Expression count matrices from bulk RNA-seq studies commonly go to NCBI GEO (open access) or, when human data needs controlled access, to dbGaP (USA) or EGA (Europe).

Depositing human data

The choice between open and controlled deposit arises only for human data, where the risk of re-identifying a participant governs access. Non-human and non-identifying data always takes the open path. For human genomic data, deposit then follows three axes: the stage of processing (raw reads, aligned reads, variants, expression), the access tier (open or controlled), and the region of origin.

graph LR
  start([Human genomic data])
  start --> eu([EU])
  start --> us([USA])

  eu -->|raw reads| eu_fastq([FASTQ])
  eu -->|aligned reads| eu_bam([SAM-BAM-CRAM])
  eu -->|variants| eu_vcf([VCF])
  eu -->|expression| eu_tsv([TSV])

  us -->|raw reads| us_fastq([FASTQ])
  us -->|aligned reads| us_bam([SAM-BAM-CRAM])
  us -->|variants| us_vcf([VCF])
  us -->|expression| us_tsv([TSV])

  ena([ENA])
  ega([EGA])
  eva([EVA])
  sra([SRA])
  dbgap([dbGaP])
  dbsnp([dbSNP])
  geo([NCBI GEO])

  eu_fastq -->|public| ena
  eu_fastq -->|controlled| ega
  eu_bam -->|public| ena
  eu_bam -->|controlled| ega
  eu_vcf -->|public| eva
  eu_vcf -->|controlled| ega
  eu_tsv -->|controlled| ega

  us_fastq -->|public| sra
  us_fastq -->|controlled| dbgap
  us_bam -->|public| sra
  us_bam -->|controlled| dbgap
  us_vcf -->|public| dbsnp
  us_vcf -->|controlled| dbgap
  us_tsv -->|public| geo
  us_tsv -->|controlled| dbgap
  geo -.->|brokers raw reads| sra
  dbsnp -.->|large variants| dbvar([dbVar])

Typical deposit routing for human genomic data by processing stage, region, and access tier, where the controlled-access path (EGA, dbGaP) holds data that could re-identify a participant.

Non-human and non-identifying data

Data from model organisms and other animal studies, together with human data that is aggregated or consented for open release, can be shared openly: anyone can download it from the INSDC archives (ENA in Europe, SRA in the USA, DDBJ in Japan) without applying for access. This is the common case in preclinical and animal neuroscience (see Preclinical and animal research). Only individual-level human data that could identify a participant goes to the controlled archives EGA and dbGaP, where each reuse request is approved by an access committee.

Multi-omics

A multi-omics study measures the same samples in several ways, and the resulting datasets are deposited in different archives: sequence in ENA, variant calls in EVA, and expression counts in NCBI GEO. BioSamples is what holds these together. It is the EMBL-EBI database that gives each biological sample one record with a permanent identifier, known as an accession, and every sample submitted to ENA receives one automatically. Because each archive records that accession, the variant calls and expression counts for one sample can be matched up and analysed together even though they live in separate repositories.

graph LR
  sample([One sample<br/>SAMEA accession])
  sample -->|DNA-seq| var([Variant calls<br/>EVA])
  sample -->|RNA-seq| counts([Expression counts<br/>NCBI GEO])
  sample -->|other assay| other([Other omics])
  var -->|matched by SAMEA| join([Integrated<br/>multi-omics analysis])
  counts -->|matched by SAMEA| join
  other -->|matched by SAMEA| join

One sample measured by several assays lands in different archives, then rejoined for multi-omics analysis through its shared BioSamples (SAMEA) accession.

Federated search and analysis

The GA4GH Beacon specification defines a standardised query interface that allows institutions to expose whether their database contains a given genomic variant without sharing the underlying data. Participating institutions run a Beacon-compliant service locally and respond to queries from external researchers, enabling cross-institutional genomic discovery without moving sensitive data to a central location. BBMRI-ERIC operates a network of Beacon nodes across European biobanks, making population-scale variant frequencies queryable across the network.

The GDI (European Genomic Data Infrastructure) project implements this model at continental scale, deploying a federated infrastructure across 21 EU countries using GA4GH Beacon and DRS APIs, with EGA Federated national nodes as the controlled-access repository backbone. GDI implements the 1+MG Framework, the normative reference document produced by the 1+ Million Genomes (1+MG) initiative covering ELSI, data quality, technical standards, and healthcare integration for national genomics programmes. The Genome of Europe project (2024–2028) is building the primary reference dataset for GDI, generating whole-genome sequences from population cohorts across 20+ European countries coordinated through BBMRI-ERIC national nodes.

Single-cell data

Single-cell genomics differs from bulk sequencing in that each cell is profiled individually, producing a cells-by-features matrix as the primary data object rather than a sequence read file. The processing pipeline, tooling, and archives are therefore distinct from those above.

Single-cell data is analysed in two primary ecosystems: AnnData (Python/scverse) and Seurat (R), with the h5ad file format serving as the common exchange format between them. Cell type annotation uses Cell Ontology, and anatomical provenance of cell populations (brain region and tissue type) uses UBERON. Single-cell datasets are deposited in CELLxGENE, which provides interactive exploration and a programmatic Census API for large-scale cross-dataset analysis, or in NeMO Archive for BRAIN Initiative-funded data.

Notable open datasets

• BICAN is assembling a multi-resolution mammalian brain cell type atlas from single-cell transcriptomic data, deposited in NeMO Archive and CELLxGENE.
• UK Biobank includes whole-exome sequencing and array genotyping from 500,000 participants, linked to neuroimaging and phenotypic data, available under controlled access.
• All of Us Research Program has released whole genome sequences linked to electronic health records for hundreds of thousands of participants as of 2026, standardised to OMOP CDM and accessed through its Researcher Workbench.
• TOPMed adds whole genome sequencing to dozens of existing NHLBI cohort studies, releasing genotypes via dbGaP and aligned reads via SRA.
• GTEx catalogues genetic effects on gene expression across human tissues, including a dozen distinct brain regions, with individual-level data via dbGaP.
• PGC conducts genome-wide association meta-analyses of psychiatric disorders across disease-specific working groups, with data via dbGaP and EGA.
• ENIGMA Consortium coordinates genome-wide association studies integrated with neuroimaging data across hundreds of sites, with summary statistics openly shared.

For the phenotyping standards and variant curation infrastructure relevant to rare neurological disease, see Rare Disease and Phenotyping. For the health data access models and regulatory constraints governing sensitive genomic data, see Health.