Drug Approvals Knowledge Provider (DAKP) Reference Ingest Guide¶
Source Information¶
InfoRes ID: infores:multiomics-drugapprovals
Description: The Drug Approvals KP (DAKP) provides information on FDA-approved drugs and their approved indications, along with drug usage regardless of formal FDA approval, and drug contraindications. Knowledge is derived by the DAKP pipeline from DailyMed (FDA Structured Product Labeling), the FDA Adverse Event Reporting System (FAERS), and MEDI. Drug-disease edges carry FDA approval status, FDA approval (application) numbers, supporting evidence identifiers, and adverse-event case counts. DAKP is one of several data-derived Knowledge Providers contributing to NCATS Translator.
Citations: - Generating Biomedical Knowledge Graphs from Knowledge Bases, Registries, and Multiomic Data (preprint): https://pmc.ncbi.nlm.nih.gov/articles/PMC11601480/
Data Access Locations: - Direct download links for the latest version are given in the manifest file: https://github.com/multiomicsKP/drug_approvals_kp/blob/main/manifest.json - Processed KGX node/edge files: https://db.systemsbiology.net/gestalt/KG/
Data Provision Mechanisms: file_download
Data Formats: kgx
Data Versioning and Releases: New versions are released periodically (not on a fixed schedule). The current version is recorded in the manifest file (https://github.com/multiomicsKP/drug_approvals_kp/blob/main/manifest.json) and embedded in the downloaded file names (e.g. drug_approvals_kg_edges_v{version}.jsonl.gz).
Ingest Information¶
Ingest Categories: translator_knowledge_creator
Utility: DAKP provides curated drug-disease treatment relationships with FDA approval status and supporting evidence, which is essential for understanding approved therapeutic uses and clinical evidence. DAKP also provides information on drug use regardless of FDA approval (including off-label use, with adverse-event case counts) and information on drug contraindications. These edge types support Translator treatment-prediction and drug-safety reasoning use cases.
Scope: FDA-approved drugs and their treatment relationships with diseases/phenotypes, including approval status, FDA approval numbers, and supporting evidence; drug use in the presence or absence of FDA approval, with case counts derived from adverse-event reporting; and drug contraindications. Node and edge properties beyond those listed are out of scope.
Relevant Files¶
| File Name | Location | Description |
|---|---|---|
| drug_approvals_kg_edges.jsonl.gz | https://db.systemsbiology.net/gestalt/KG/ | Drug-disease treatment, usage, and contraindication edges processed from DailyMed, FAERS, and MEDI |
| drug_approvals_kg_nodes.jsonl.gz | https://db.systemsbiology.net/gestalt/KG/ | Node records (id, name, category) for the drug and disease/phenotype entities referenced by the edges |
Included Content¶
| File Name | Included Records | Fields Used |
|---|---|---|
| drug_approvals_kg_edges.jsonl.gz | Drug-disease/phenotype edges with predicates biolink:treats (FDA-approved indications), biolink:applied_to_treat (usage with or without formal approval, from FAERS), and biolink:contraindicated_in (contraindications). Edges missing subject, object, or predicate are skipped. | id, subject, object, predicate, category, knowledge_level, agent_type, sources, N_cases, clinical_approval_status, approvals, has_evidence, publications |
| drug_approvals_kg_nodes.jsonl.gz | Node records for the subject (drug/chemical) and object (disease/phenotype) of each ingested edge. | id, name, category |
Future Content Considerations¶
edge_property_content: The has_evidence field (DailyMed SPL identifiers) is currently merged into the publications edge property. Consider representing this supporting evidence with a dedicated evidence/has_evidence property rather than co-mingling it with publications. - Relevant files: drug_approvals_kg_edges.jsonl.gz
edge_property_content: clinical_approval_status values that are not valid Biolink ClinicalApprovalStatusEnum members (e.g. "?") are coerced to not_provided. Confirm the intended mapping with the source, and whether additional statuses should be preserved. - Relevant files: drug_approvals_kg_edges.jsonl.gz
Target Information¶
Edge Types¶
| Subject Categories | Predicate | Object Categories | Knowledge Level | Agent Type | UI Explanation |
|---|---|---|---|---|---|
| biolink:ChemicalEntity, biolink:SmallMolecule, biolink:MolecularMixture | biolink:Disease, biolink:PhenotypicFeature | knowledge_assertion | manual_validation_of_automated_agent | This edge asserts that a drug is an FDA-approved treatment for a disease or phenotype. DAKP derives these assertions by processing FDA Structured Product Labeling documents (via DailyMed) and FAERS, with manual validation of an automated pipeline. The edge carries the FDA approval status (clinical_approval_status), the FDA approval/application number(s) (FDA_regulatory_approvals), and DailyMed SPL evidence identifiers (recorded as publications). The strength of an FDA approval is best represented with the Biolink 'treats' predicate. | |
| biolink:ChemicalEntity, biolink:SmallMolecule, biolink:MolecularMixture | biolink:Disease, biolink:PhenotypicFeature | observation | manual_validation_of_automated_agent | This edge reports that a drug was applied to treat a disease or phenotype with or without formal FDA approval, based on usage observed in FDA Adverse Event Reporting System (FAERS) reports. FAERS is the primary knowledge source; the DAKP pipeline aggregates and manually validates the data. The number_of_cases property reports how many FAERS reports support the usage, and clinical_approval_status distinguishes approved-for-condition from off-label use. The observational, usage-based nature of this relationship is represented with the Biolink 'applied_to_treat' predicate. | |
| biolink:ChemicalEntity, biolink:SmallMolecule, biolink:MolecularMixture, biolink:Drug | biolink:Disease, biolink:PhenotypicFeature | knowledge_assertion | manual_validation_of_automated_agent | This edge reports that a drug is contraindicated in the presence of a disease or condition, based on warnings in FDA approval documents as captured by MEDI. MEDI is the primary knowledge source; the DAKP pipeline aggregates and manually validates the data, with DailyMed as supporting data. The contraindication relationship is represented with the Biolink 'contraindicated_in' predicate. |
Node Types¶
| Node Category | Source Identifier Types | Additional Notes |
|---|---|---|
| biolink:ChemicalEntity | CHEBI, CHEMBL.COMPOUND, DRUGBANK, MESH, UNII | |
| biolink:SmallMolecule | CHEBI, PUBCHEM.COMPOUND, UNII | |
| biolink:MolecularMixture | CHEBI, PUBCHEM.COMPOUND, UNII | |
| biolink:Drug | RXCUI | |
| biolink:Disease | MONDO, DOID, EFO, UMLS | |
| biolink:PhenotypicFeature | HP, EFO, NCIT, UMLS |
Future Modeling Considerations¶
node_properties: The ingest code maps a biolink:ComplexMolecularMixture category, but no such nodes are present in the current release. Revisit if/when the source begins providing them.
edge_properties: Confirm whether FDA_regulatory_approvals is the preferred Biolink property for FDA application numbers, or whether a more specific/standard representation should be used.
spoq_pattern: Object categories are currently Disease and PhenotypicFeature only. Confirm whether DiseaseOrPhenotypicFeature should be retained/used given downstream normalization.
Additional Notes: ['The previous draft included a target_info.infores_id ("infores:translator-dakp-kgx"); it was removed as part of this review (deprecated/placeholder). Confirm the correct target infores identifier for the DAKP output graph, if one is required.']
Provenance Information¶
Contributors: - Gwenlyn Glusman - code author, domain expertise, data modeling - Matthew Brush - data modeling - Sierra Moxon - code, data modeling
Artifacts: - RIG review issue: https://github.com/NCATSTranslator/translator-ingests/issues/416 - RIG review epic: https://github.com/NCATSTranslator/translator-ingests/issues/407