FDA Substance Knowledge Graph

The FDA Substance Knowledge Graph publishes the U.S. Food and Drug Administration's substance registry as linked data. It is built from two public openFDA datasets: the Global Substance Registration System (GSRS, 167,385 substance records) and the Unique Ingredient Identifier list (UNII, 176,619 identifiers), 53.7 million triples. Every source field is kept, and every record carries its source dataset, record ID, source URL and download date.

Main class. fdasub:Substance, one node per substance. A substance with a UNII has the IRI https://bmedesign.org/fda-substance-kg/id/unii/<unii in lowercase>. A GSRS substance without a UNII uses its GSRS UUID: .../id/gsrs/<uuid>. Each substance has rdfs:label (the GSRS display name), fdasub:unii, fdasub:substanceClass (chemical, protein, nucleicAcid, polymer, mixture, structurallyDiverse, concept, specifiedSubstanceG1, mention) and the full GSRS record as child nodes.

Child nodes sit under their substance's IRI and are typed by GSRS section: fdasub:GsrsNames (all names, with type, language and preferred flags), fdasub:GsrsCodes (every external code with fdasub:codeSystem and fdasub:code), fdasub:GsrsStructure and fdasub:GsrsMoieties (SMILES, formula, molecular weight, stereochemistry, molfile), fdasub:GsrsProtein, fdasub:GsrsNucleicAcid, fdasub:GsrsPolymer, fdasub:GsrsMixture, fdasub:GsrsStructurallyDiverse (organism and part for natural products), fdasub:GsrsRelationships, fdasub:GsrsProperties, fdasub:GsrsNotes and fdasub:GsrsReferences (citations).

Links between substances. fdasub:GsrsRelationships nodes carry the relationship fdasub:type (for example ACTIVE MOIETY, SALT/SOLVATE->PARENT, METABOLITE->PARENT, ENANTIOMER->RACEMATE, TARGET->INHIBITOR) and point to the related substance with fdasub:relatedSubstanceSubstance. Natural products link to their parent organism substance with fdasub:parentSubstanceSubstance, mixture components with fdasub:substanceSubstance, polymer monomers with fdasub:monomerSubstanceSubstance, structural modifications with fdasub:molecularFragmentSubstance and fdasub:agentSubstanceSubstance, and relationship mediators with fdasub:mediatorSubstanceSubstance.

Shortcut identifiers for joins (PRIMARY codes only, directly on the substance): fdasub:casNumber, fdasub:pubchemCid, fdasub:dtxsid (EPA CompTox), fdasub:rxcui, fdasub:drugbankId, fdasub:chebiId, fdasub:chemblId, fdasub:ecNumber, fdasub:meshId, fdasub:nciThesaurusCode, fdasub:innId. Use fdasub:dtxsid to join to biobricks-ice and other EPA CompTox graphs, fdasub:pubchemCid for PubChem-based graphs, and fdasub:unii for Wikidata (P652).

Every link resolves. 775 stub nodes (fdasub:isStub true) stand in for substances that other FDA graphs reference by UNII or GSRS UUID but GSRS does not list, so no cross-graph link points to an untyped IRI.

Related graphs. The other FDA graphs in this family link to these substance IRIs: drug recalls in fda-compliance-kg, applications, NDC products, labels and shortages in fda-drug-biologic-kg, food ingredient inventories in fda-food-cosmetic-kg (by CAS number) and FAERS reports in fda-drug-event-kg.

Source and license. GSRS and UNII data are public domain U.S. government works distributed by openFDA (https://open.fda.gov/license/). Refreshed from openFDA bulk downloads. Funded under NSF award 2535091 (Proto-OKN Theme 2: OKN-Fabric) through subaward 5136002. The graph is released under CC0 1.0.

provenance:
  - derivedFrom:
      label: openFDA other/substance (GSRS)
      uri: https://open.fda.gov/apis/other/substance/
      kind: Database
      license:
        data:
          label: Public Domain
          spdx: CC0-1.0
          uri: https://open.fda.gov/license/
  - derivedFrom:
      label: openFDA other/unii
      uri: https://open.fda.gov/apis/other/unii/
      kind: Database
      license:
        data:
          label: Public Domain
          spdx: CC0-1.0
          uri: https://open.fda.gov/license/