Sequence Read Archive (SRA)

The Sequence Read Archive (SRA), previously known as the Short Read Archive, is NCBI’s repository of raw, high-throughput DNA and RNA sequencing data — the largest publicly available collection of its kind. SRA operates as the United States node of the International Nucleotide Sequence Database Collaboration (INSDC), alongside the European Bioinformatics Institute’s European Nucleotide Archive (ENA) in Europe and the DNA Data Bank of Japan (DDBJ). A dataset submitted to any one of the three partner archives is automatically shared and mirrored across all three, so researchers can search or download the same underlying data regardless of which regional archive they start from.

What SRA accepts — and what it doesn’t

SRA specifically holds raw or minimally processed sequencing reads, accepted in FASTQ, BAM, or CRAM format, from genomic, metagenomic, and transcriptomic experiments across all branches of life. It deliberately does not accept assembled genomes, contigs, or genome assemblies — those belong in GenBank or RefSeq — nor does it hold interpreted variant calls, which belong in a resource such as dbSNP or ClinVar. SRA is specifically the archive for the sequencing instrument’s own raw output, before or immediately after basic quality processing.

BioProject and BioSample: the required context around every submission

An SRA submission is never freestanding. Every set of sequencing reads must be linked to a BioProject record, which describes the overall research study or initiative the sequencing supports, and at least one BioSample record, which describes the actual biological source material — the organism, tissue, cell line, or environmental sample — that was sequenced. That three-tier structure (BioProject → BioSample → SRA reads) is what lets a downstream researcher understand not just what sequence was generated, but what it was generated from and why, without having to reconstruct that context from a paper’s methods section.

Examples

• A lab that has just completed a whole-genome sequencing run registers a BioProject describing the study, creates a BioSample record for the organism or tissue sequenced, then uploads the raw FASTQ files to SRA under that BioProject — satisfying a funder’s or journal’s data-availability requirement before publication.
• A bioinformatician re-analyzing a published RNA-seq study downloads the original raw reads from SRA, rather than relying on the paper’s processed count tables, to run a different alignment or quantification pipeline and check whether the published results are pipeline-dependent.

Counter-example

A finished, assembled bacterial genome submitted to GenBank is not an SRA submission — SRA specifically holds unassembled raw reads and alignments; the assembled sequence itself belongs in GenBank/RefSeq, with the raw reads that produced it optionally also deposited separately in SRA.

Related infrastructure

See the companion guide, Submitting Sequence Data to NCBI SRA, for the practical submission mechanics, and dbSNP for where SRA-derived variant calls are typically catalogued once analyzed.