Study to run, via the ENA portal

Route id: study->run:ena_filereport Provider: EBI ENA portal Direction: SRP/ERP/DRP → SRR/ERR/DRR

curl -s 'https://www.ebi.ac.uk/ena/portal/api/filereport?accession=SRP446371&result=read_run&fields=run_accession&format=tsv&limit=0'

The shortest path from a study to its data. result=read_run rows are runs, so a run returned here is a run that exists — the proves_data_exists property that separates this from the GEO-side routes, where 1.23% of the experiments SOFT names carry no runs at all.

Measured

Corpus: a seeded 3,000-study draw from the 12,748 studies the BioProject to study census returned. Surveyed: 2026-09-15.

Resolved	3,000 (100%)
Empty	0
Failed	0
Unique runs	205,175

68.4 runs per study, and not one study that resolved to nothing. The 23 BioProjects that every ENA route reports empty never produced a study accession in the first place, so they cannot appear here — which is the right behaviour and also the reason this route's 100% is not a stronger claim than the BioProject route's 99.8%. They are measuring the same archive over populations that differ by exactly those 23.

Study or project?

Both directions from a BioProject are total over this corpus, and both are one request. Prefer the project where you have one: it is the accession GEO records and the one every cross-archive relation in this project is keyed on. The study route matters when the study is what you were handed — from an ArrayExpress IDF, or from a paper's data-availability statement, which quotes secondary study accessions far more often than project accessions.