Skip to content

Ontology Policy

Principles

  1. Prefer standard ontologies over ad-hoc identifiers (MONDO for diseases, HPO for phenotypes, GO/Reactome for biology, Uberon for anatomy).
  2. Never merge ontologies silently — imports create versioned snapshots in the biomed store.
  3. Preserve source CURIEs in disease profiles and claims.
  4. Document upstream licenses in docs/legal/data-licenses.md.

Supported ontologies (import adapters)

Ontology Use Import module
MONDO Disease identifiers biomed/imports/mondo
HPO Phenotypes biomed/imports/hpo
HPOA Phenotype associations biomed/imports/hpoa
GO Biological processes biomed/imports/go
Reactome Pathways biomed/imports/reactome
Uberon Anatomy biomed/imports/uberon
ClinVar Variants biomed/imports/clinvar
openFDA Drug labels biomed/imports/openfda

Disease module identifiers

  • Scaffold modules map to MONDO/EFO via Open Targets bulk harvest
  • Curated modules include kg_node_id and EFO/MONDO references in profile.json
  • Slug (disease_id) is the stable internal key — not always equal to MONDO label

Interoperability status

Standard Status
MONDO CURIEs STABLE
HPO term IDs STABLE
FHIR Condition export NOT_IMPLEMENTED
OMOP concept mapping NOT_IMPLEMENTED
Phenopackets NOT_IMPLEMENTED

Curation rules

  • Do not invent ontology IDs; use resolver scripts or Open Targets EFO mappings
  • When ontology term is unavailable, document limitation in SCREENING_PROFILE.limitations
  • Cross-disease comparison uses HPO-aware logic in biomed/comparison/

Update policy

Ontology snapshots are refreshed via scripts/setup_biomed_imports.py. Pin snapshot dates in import reports for reproducibility.