ENA: Public Sequencing Data Archive

The ENA (European Nucleotide Archive) is a public sequencing-data repository operated by EMBL-EBI in Europe. Researchers around the world submit raw reads produced by sequencers, assembled sequences, and experimental descriptions called metadata. Raw reads are commonly provided as FASTQ files, and users can download public .fastq.gz files directly by URL.

This reference explains what to find in ENA and how to download it when practising analysis or obtaining reference data.

Web browser: www.ebi.ac.uk/ena/browser
Operated by EMBL-EBI (European Bioinformatics Institute, United Kingdom)

Why download from ENA rather than SRA?

The same data exists in all three archives, but download convenience differs.

Archive	Operator	How to obtain raw reads
ENA (Europe)	EMBL-EBI	Use a `.fastq.gz` URL directly with `curl` or `wget`, the simplest method
SRA (United States)	NCBI	Use dedicated tools (`prefetch` then `fasterq-dump`) to unpack `.sra` files
DDBJ (Japan)	National Institute of Genetics	DRA, also mirrored to ENA

Are all three mirrored? The INSDC agreement

ENA, SRA, and DDBJ, whose raw-read section is called DRA, are linked by the INSDC (International Nucleotide Sequence Database Collaboration), an international agreement that has existed for more than 30 years. Its principle is:

Public data submitted to any one archive is exchanged daily among all three institutions so that it can be searched and downloaded from each of them.

The practical answer to “Do they hold the same data?” is therefore yes. At two levels:

Level	Degree of mirroring
Metadata (accessions and study/sample information)	Fully synchronized. This is why data submitted to DDBJ can be searched and inspected unchanged in Europe’s ENA.
Raw-read data itself	Exchanged among the three institutions and available anywhere, although its storage and delivery format differs.

Same data, different packaging

The content, the read sequences, is the same, but the file form and access method differ by institution.

SRA (NCBI): stores its own .sra format, which must be converted to FASTQ with dedicated tools
ENA (EBI): prepares FASTQ in advance and serves it by direct URL, allowing immediate download with curl
DDBJ: mirrored through DRA

FASTQ files produced by ENA and converted by SRA may differ slightly in header notation, and newly submitted data may take several days to appear in all three locations.

Accession identifier structure

Hierarchical IDs organize the data. The third character of the prefix identifies the type, while the first identifies the archive: E for ENA, S for SRA, and D for DDBJ.

Level	Examples	Meaning
Study / Project	`PRJEB…` `PRJNA…` `PRJDB…`	One research project
Sample	`SAMEA…` `SAMN…` `SAMD…`	Biological sample
Experiment	`ERX…` `SRX…` `DRX…`	Library and sequencing conditions
Run	`ERR…` `SRR…` `DRR…`	Actual read file, the download unit

A Run is a collection of reads produced by one run of a sequencing instrument. It is the unit used to download raw reads. Running the same Experiment, with the same settings, several times can produce multiple Runs.

Finding data with the Portal API

You can insert an accession into the browser URL (.../ena/browser/view/<accession>), but the REST API is more convenient for conditional searches and automation.

Find runs matching conditions with `search`

Example: a list of human, RNA-seq, Illumina, paired-end runs.

curl -s -G 'https://www.ebi.ac.uk/ena/portal/api/search' \
  --data-urlencode 'result=read_run' \
  --data-urlencode 'query=tax_eq(9606) AND library_strategy="RNA-Seq" AND instrument_platform="ILLUMINA" AND library_layout="PAIRED"' \
  --data-urlencode 'fields=run_accession,fastq_bytes,read_count,fastq_ftp,fastq_md5' \
  --data-urlencode 'format=tsv'

tax_eq(9606) means human, using the NCBI taxonomy ID
Useful fields: read_count, fastq_bytes for size, fastq_ftp for URLs, and fastq_md5 for integrity
Complete field list: returnFields?result=read_run

File information for a specific accession with `filereport`

curl -s -G 'https://www.ebi.ac.uk/ena/portal/api/filereport' \
  --data-urlencode 'accession=DRR024878' \
  --data-urlencode 'result=read_run' \
  --data-urlencode 'fields=fastq_ftp,fastq_md5,fastq_bytes' \
  --data-urlencode 'format=tsv'

Build a curl command yourself

Use the GUI below to select options for both APIs and assemble a curl command. Switch between conditional search (search) and file information (filereport) with the tabs. Changing filters, fields, or format updates the command in real time. Use the Copy button to paste it directly into a terminal.

Downloading data

URL structure

An example value returned by fastq_ftp:

ftp.sra.ebi.ac.uk/vol1/fastq/DRR024/DRR024878/DRR024878_1.fastq.gz

Add https:// because the protocol prefix is absent.
Depending on the number of digits in the accession, the path may contain a zero-padded subdirectory such as 001/. Do not assemble the path yourself; use the fastq_ftp value unchanged.

Paired-end means two files

_1.fastq.gz, the forward read, and _2.fastq.gz, the reverse read, form a pair. Alignment requires both files from a paired library, so download them together.

curl -sS -O https://ftp.sra.ebi.ac.uk/vol1/fastq/DRR024/DRR024878/DRR024878_1.fastq.gz
curl -sS -O https://ftp.sra.ebi.ac.uk/vol1/fastq/DRR024/DRR024878/DRR024878_2.fastq.gz

-O saves the file under the name in the URL; -sS hides the progress bar but displays errors
Resume a download with curl -C - -O <url>
macOS includes curl, but not wget, by default

Use Aspera for large data

For tens or hundreds of gigabytes, ENA provides Aspera (fasp), which is faster than curl ([email protected]:...). It is unnecessary for small datasets.

Required: verify integrity with md5

Compare each downloaded file against the fastq_md5 value to make sure it is intact.

md5 DRR024878_1.fastq.gz     # macOS  (Linux: md5sum)

Restricted access: controlled data

All of this mirroring and public downloading applies only to public data. Human data with consent restrictions, such as patient genomes, goes through controlled-access systems such as EGA in Europe or dbGaP in the United States, not the public INSDC archives. Those controlled repositories do not mirror one another.

In a controlled-access study, you may therefore be unable to download raw reads directly and may have to use only de-identified summaries or expression matrices published separately by the researchers.

Summary cheat sheet

Find:      filereport API → obtain fastq_ftp and fastq_md5
Download:  curl -O <https://…fastq.gz>   (for paired data, both _1 and _2)
Verify:    compare md5

See the Data Formats Reference for related formats.