More flexible approach for ontology management

[lindaxiang]

Context:
During the data harmonization for HostSeq study, we were able to use 491/495 (99.2%) source variables for data mapping. However the following entities still need to add support for additional ontologies:

• Comorbidity_code: MONDO, icd10, snomedct, mesh
• Exposure_code: snomedct, ExO, LOINC, NCIT, icd10
• Measurement_code: LOINC, snomedct
• Phenotype: HP, MONDO, NCIT, snomedct
• Measurement unit code: UO, UCUM

Proposal:

• define one global permissible ontology list for the PCGL data model schema, rather than field-specific ontology restrictions for each *_code field.
• all ontology code fields share one permissible code list; field-level documentation provides usage guidance.

Rationale:

• Add new ontologies once; all fields automatically inherit
• Build comprehensive initial list to minimize future schema updates
• Easier to accommodates emerging ontologies and support ontology expands

[lindaxiang]

Decision on Jan14 DM meeting:

1. keep the current data model as it is
2. leave the data out if they can't fit into the model with the current support ontologies
3. revise the data model when more use cases come up