Public Bio Databases
Use the public-bio-api MCP rather than inventing URLs or scraping private portals.
Workflow
- Call
bio_api_catalogto select a source and read its evidence limit. - Translate the question into a precise identifier or search expression.
- Call
bio_api_querywith the smallest useful result limit. - Record the exact source URL, retrieval time, identifiers, release/header metadata when present, and empty/failed responses.
- Cross-check high-impact claims with a second independent source or primary literature.
Read references/api-catalog.md for source selection.
Interpretation rules
- Database annotation is evidence, not truth by authority; record evidence codes and record versions when available.
- Sequence similarity alone does not prove one-to-one orthology or conserved biological function.
- STRING edges are functional associations, not automatically physical binding.
- Predicted structures require coverage and confidence inspection and do not establish mechanism.
- Motif matches are candidate regulatory sites, not occupancy or regulation proof.
- Search results are discovery aids; read the paper before citing it as support.
Privacy boundary
Do not submit credentials, local paths, private identifiers, or excluded species-specific terms. Do not work around an MCP refusal. If a public source lacks a verified unauthenticated API, report that limit instead of scraping it.