Sample to run, via the ENA portal

Route id: sample->run:ena_filereport Provider: EBI ENA portal Direction: SRS/ERS/DRS → SRR/ERR/DRR

curl -s 'https://www.ebi.ac.uk/ena/portal/api/filereport?accession=SRS18229461&result=read_run&fields=run_accession&format=tsv&limit=0'

The step a reprocessing pipeline actually needs: given a sample, which runs carry its reads. One sample routinely spans several runs — sequenced across lanes, or topped up in a second run — and a pipeline that takes only the first produces a matrix from part of the library.

Measured

Corpus: sample:3000@reprocessed-srs, a seeded draw from the 103,440 INSDC samples in the reprocessed table. Surveyed: 2026-09-15.

Resolved	2,987
Empty	13
Failed	0
Unique runs	8,359

2.8 runs per sample. That is the number that makes this route matter: for most samples in this corpus, taking one run takes roughly a third of the data.

The 13 empties are the same class as the 23 missing runs — accessions that were in the reprocessed table and are no longer resolvable.

Secondary or primary?

SRS-style accessions are secondary sample accessions; SAMN/SAMEA/SAMD BioSamples are primary. ENA's filereport accepts both, and this project keeps them as separate entity types (sample and biosample) with separate routes, because the archives do not treat them as interchangeable: GEO records the BioSample and not the SRS, and ENA under-reports NCBI-registered BioSamples relative to ELink in one direction while finding 125,271 more in the other.