Data sources
Six public datasets, and nothing else
Every fact in Biodatica came out of one of the six datasets below. All six are U.S. government works, free and public. Nothing here is scraped, licensed, purchased, or guessed at, and every row in the product links back to the filing it came from.
| Source | Endpoint | What it gives | Refresh | Access |
|---|---|---|---|---|
| NIH RePORTER | api.reporter.nih.gov/v2/projects/search | Funded projects, principal investigators, organizations, abstracts, award start and end dates | Weekly | Free · we hold to 1 request per second |
| ClinicalTrials.gov | clinicaltrials.gov/api/v2/studies | Registered trials, sponsor, phase, recruitment status and status changes | Daily | Free |
| SEC EDGAR | efts.sec.gov · data.sec.gov | Form D private raises: issuer, executives, amount, filing date | Daily | Free · 10 requests per second, with a descriptive User-Agent |
| USPTO | Open Data Portal — patents and assignments | Granted patents, assignee of record, titles, grant dates | Weekly | Free · API key |
| openFDA | api.fda.gov/drug · api.fda.gov/device | Approvals, 510(k) clearances, designations | Daily | Free · CC0 |
| SBIR.gov and NSF | api.www.sbir.gov · api.nsf.gov | Non-dilutive award history, award phase, firm records that link an institution to a company | Weekly | Free · public domain |
Refresh is the cadence we pull on, not the cadence the publisher updates on. The daily sources run at 06:00 UTC and the weekly ones early on Monday, so a Form D filed on a Tuesday afternoon reaches the graph that night, and a fresh NIH award can be up to a week old when it lands. The coverage page shows when each source last actually ran, including any that skipped or failed.
Copyright and redistribution
Works prepared by U.S. government officers as part of their official duties are not subject to copyright protection in the United States (17 U.S.C. § 105). That is what makes this whole product legally simple: the input layer is public domain, so there is no license to breach and no vendor who can switch us off.
What we add on top is ours: the entity resolution that decides these records describe one company, the momentum and runway scores, and the written briefs. The underlying filings belong to the filers and to the public.
Attribution
ClinicalTrials.gov
Trial data courtesy of the U.S. National Library of Medicine. The NLM does not endorse or review anything we do with it.
openFDA
Released under CC0. The FDA does not endorse Biodatica, and openFDA data is not for making clinical decisions.
NIH RePORTER, SBIR.gov, NSF
Award records are U.S. government works. Naming an agency as the source of an award is not a claim that the agency endorses this product.
SEC EDGAR and USPTO
Filings and patents are public records. Form D amounts are self-reported by the issuer and neither audited nor verified by the SEC, and Biodatica does not verify them either.
How we behave against these APIs
These are public services paid for by taxpayers, and a data business that hammers them is stealing from everyone else who needs them. Our commitments:
- NIH RePORTER is paced to 1 request per second, enforced by a single rate pacer rather than by hope.
- SEC EDGAR is held to 10 requests per second and identifies itself with a descriptive User-Agent carrying a contact address.
- Every pull is incremental against a stored watermark, so a run fetches what changed and not the archive.
- Raw payloads are stored once and replayed from our own storage rather than re-fetched during development.
- A source that rate-limits or errors backs off and skips rather than retrying in a loop. When it skips, the coverage page says so.
What we deliberately leave out
Two proprietary datasets sit embedded inside otherwise-open government data. They are not public domain, and they are excluded from everything Biodatica stores or shows.
D&B and DUNS fields
Present in USAspending records and licensed from Dun & Bradstreet. We do not ingest or resell them.
GMDN device nomenclature
Present in openFDA device records and licensed from the GMDN Agency. Excluded from our device handling.
We also do not ingest Crunchbase, PitchBook, or scraped LinkedIn. Their terms forbid redistributing what you would be paying us for, and the public spine stands on its own without them. If a fact appears in Biodatica, a government filing says it.
How these six sources become three numbers · Live coverage and source status · Disclosures