Should the JSON be able to represent different types of mapping?
There are different schemes available
Skos
exact, broad, narrow, related
Note in skos exact is weaker than true equivalence. It is not transitive.
OWL
EquivalentClass, SubClassOf
Note that OWL equivalence is strict logical equivalence. Symmetry and transitivity hold. This means if A Equiv B, B Equiv C, then A Equiv C
Xrefs
obo format provides xrefs but these are not logical relations and are not precisely defined. Different ontologies use xref differently, and in some ontologies the meaning of xref varies tremendously within that ontology.
Different ontologies support a mix of each, as each type of mapping has it's own consumers. For OWL reasoning logical mappings are necessary
MONDO provides logical equivalencies, skos mappings and xrefs
Uberon provides logical mappings and xrefs; skos can be added on demand
An additional wrinkle with Uberon is that Uberon classes are not strictly equivalent to FMA classes, since Uberon is species-neutral and FMA is species-specific. Instead we represent using OWL axioms of the form:
FMA:Heart = UBERON:Heart and part_of some NCBITaxon:9606
Should the JSON be able to represent different types of mapping?
There are different schemes available
Skos
exact, broad, narrow, related
Note in skos exact is weaker than true equivalence. It is not transitive.
OWL
EquivalentClass, SubClassOf
Note that OWL equivalence is strict logical equivalence. Symmetry and transitivity hold. This means if A Equiv B, B Equiv C, then A Equiv C
Xrefs
obo format provides xrefs but these are not logical relations and are not precisely defined. Different ontologies use xref differently, and in some ontologies the meaning of xref varies tremendously within that ontology.
Different ontologies support a mix of each, as each type of mapping has it's own consumers. For OWL reasoning logical mappings are necessary
An additional wrinkle with Uberon is that Uberon classes are not strictly equivalent to FMA classes, since Uberon is species-neutral and FMA is species-specific. Instead we represent using OWL axioms of the form:
FMA:Heart = UBERON:Heart and part_of some NCBITaxon:9606
These are distributed here: http://uberon.github.io/downloads.html#bridge
Docs are in the uberon paper and on https://github.com/obophenotype/uberon/wiki/inter-anatomy-ontology-bridge-ontologies
Formats
Choices here are
The obo json format is described here https://github.com/geneontology/obographs/ And in this blog: https://douroucouli.wordpress.com/2016/10/04/a-developer-friendly-json-exchange-format-for-ontologies/
I think the obo json makes a good choice for representing subsets of ontologies, including mappings. Should we adopt this here?