BioProject to experiment, via the ENA portal¶
Route id: bioproject->experiment:ena_filereport
Provider: EBI ENA portal
Direction: PRJNA/PRJEB/PRJDB → SRX/ERX/DRX
The request¶
curl -s 'https://www.ebi.ac.uk/ena/portal/api/filereport?accession=PRJNA988806&result=read_run&fields=experiment_accession&format=tsv&limit=0'
limit=0 means no limit. Omitting it returns a truncated report and does not
say so.
Measured¶
Corpus: reprocessed-prj — all 12,755 BioProjects the reprocessing pipeline
has actually had to resolve. Surveyed: 2026-09-15.
| Resolved | 12,732 |
| Empty | 23 |
| Failed | 0 |
| Unique experiments | 481,315 |
Nothing failed. The 23 empties are projects ENA holds no runs for at all — the same population the BioProject to run census found empty, not a separate defect of this direction.
Why it is the route to beat¶
One request per project, returning every experiment in the project, from the
same endpoint that answers the run, study, sample and BioSample directions. The
result=read_run rows are runs, so an experiment returned here is an
experiment that carries data — the proves_data_exists property that separates
this route from the GEO-side ones, where
1.23% of the experiments SOFT names carry no runs at all.
Archive coverage¶
ENA indexes EBI- and DDBJ-registered projects as well as NCBI's, which is not symmetric: NCBI's own routes resolve 99.3% of PRJNA and 1.4% of PRJEB. See routing by archive.