Try the tool
Making public biological data trustworthy

Is public metadata holding back your science?

Start with any public accession. OpenBioData traces it back to the source, finds related and reused studies, cross-checks the evidence, and fills in missing biological context — with a citation and confidence score for every field.

Cited. Current. Yours.

Fix your metadata →
81% field recovery vs 19% from NCBI alone Every field cited Confidence scored 0–100
Real example — public NCBI accession
Input: BioProject PRJNA976261 (12 BioSamples)
host—
clinical_status—
ancestry—

The NCBI record itself is thin on context. OpenBioData resolves it and follows the linked publication record to pull the full context around it — no need to track down a paywalled paper yourself.

↳ Source paper located: Favale N et al., Molecular Oral Microbiology, 2024 — PMID 37257865

Real example — public NCBI accession

It doesn't stop at the source paper — it searches for every other publication that cites or reuses the same accession.

2019Farina R et al., Archives of Oral Biology — PMID 31153098 — original recruitment and sequencing of this cohort
2024Favale N et al., Molecular Oral Microbiology — PMID 37257865 — functional reanalysis of the same 12 subjects

↳ 2 independent publications found describing the same 12-sample cohort, five years apart

Real example — public NCBI accession

Both papers define exactly how each sample was classified by health status — detail the NCBI record doesn't carry at all.

NCBI BioSample recordno clinical status field
Farina et al. 2019 (Methods)4 subgroups by periodontitis + T2D status
Favale et al. 2024 (Methods)same 12 subjects, same criteria

↳ Two independent papers, five years apart, agree on the same classification — a strong cross-check signal

Real example — public NCBI accession
hostHomo sapiens96
tissuesubgingival plaque93
ancestrynorth-Italian, Caucasian adults81
clinical_statusperiodontitis / T2D subgroup78

Fields NCBI never carried, filled in directly from what both papers state — each one citable back to its source, never a black box.

Fix your metadata

Two ways to get there.

OpenBioData

Free & open source — the engine, run it yourself
OpenBioData demo — tracing an accession back to its source paper
  • 10 samples without an account, 30 once you sign in
  • Pay-as-you-go per sample after that — $0.10/sample
  • Confidence score and source citation on every field
  • MIT-licensed — self-host it with your own API key, no cap, nothing sent to us

You run it, you own the workflow. Nothing to sign. Scaling past a few hundred samples? Talk to us about BioMetadataAudit.

Try the tool → Self-host on GitHub →

BioMetadataAudit

Managed metadata auditing — for labs & institutions

Self-hosting gives you the software. BioMetadataAudit gives you metadata you can trust — and a team responsible for keeping it that way.

  • Validate — continuously cross-check your internal records against public metadata and your own schema
  • Stay current — automatically re-check records as new evidence, corrections, and upstream changes emerge
  • Get it fixed — expert review and dedicated support turn flagged problems into corrected, recoverable metadata
  • Run it at scale — institution-ready infrastructure for large datasets, with your data handled on terms we agree in writing

A small number of institutions at a time, by design.

Talk to us about your data →

Still early. Already in use.

50+
users
4
continents — EU, NA, SA, Asia
84%
average recovery rate, self-reported

Researchers and teams currently testing it or in active pilots:

Oxford OUCRU Levi Waldron Lab · CUNY SPH FDA GenomeTrakr MassIVE · UCSD BacDive / DSMZ · Leibniz Institute
Get started with BioMetadataAudit

Ready for someone else to own this?

This isn't a one-off lookup — that's what the free tool is for. Tell us about your ongoing metadata needs and we'll scope what a managed, continuously-current audit looks like for your team.

Has missing or wrong public metadata ever cost you time on a project?

Roughly how many datasets or accessions does your team rely on regularly?

What do you actually need — beyond a one-time lookup?

Last thing — how should we reach you?

Prefer to skip the form? Book a 30-minute call →

OpenBioData gives you the engine.

BioMetadataAudit takes responsibility for the outcome.