Current use cases for term lists

[mccalluc]

Term lists (typically as yaml files) are referenced from many of the repos. As we move these lists into an ontology service, there needs to be a plan to make sure the current needs are still being satisfied. That might be periodic exports from the service and updates to the repos, or something at run-time. TBD.

Under search-api:

• We map from codes in neo4j to human-readable strings for Elasticsearch.
• We generate JSON Schema to validate entities.

Under ingest-validation-tools:

• We enumerate the assay types, analyte classes, etc.
• Individual assays will allow only particular values from the list, and may introduce their own enums.
• Assay metadata field names are gathered into a single list and this is included as a submodule in portal-ui so we can provide descriptions in the UI.

For the current state of things regarding assay names and codes (in October 2020) there's this diagram. It's a mess.

[computationdoc]

With the first KG now complete (an instance of a system of ontology ingestion and integration and associated APIs) we can now see precisely the format required for codifying and ingesting all HuBMAP application enumerated lists as the HuBMAP application ontology. It will be set up for editing and ingestion and use. To ingest and use in KG APIs it only requires nodes and edges as two tsv files with all fields ASCII strings. Nodes have these fields (only required are id and label): id, label, definition, synonyms (| separated list), dbxrefs (| separated list of general biomedical ontology ids like UBERON:123456 or UBERON 123456). Edges have these fields (all required): subject_label, relation_label, object_label. Some examples of nodes rows (stylized here with \t to show tabs):

HuBMAP H00000\tHuman BioMolecular Atlas Program\tNIH Common Fund program and consortium\tHuBMAP|HuBMAP Consortium\t HuBMAP H00001\torgan\tHuBMAP organs\t\t HuBMAP H00002\tkidney\tHuman renal organ\tKidney\tUBERON:0002113 HuBMAP H00003\tleft kidney\t\tLeft Kidney\t

Some examples of edges rows (stylized here with \t to show tabs): organ\tdomain_of\tHuman BioMolecular Atlas Program kidney\tisa\torgan left kidney\tisa\tkidney

This is now unblocked per se because the method is clear, and can proceed in assembling all related artifacts and building the HuBMAP applications ontology into the KG.

[computationdoc]

Note that ontology-api enables programmatic lookup of string (term) to obtain general biomedical ontology CodeID (id) such as kidney -> UBERON 0002113

[AlanSimmons]

@computationdoc is this still a thing to do? If not, I'd like to close this issue so that we can archive this repo.