UBERON cites SNOMED E-node codes - why not the structure codes?

[salai999]

Hi there – I am trying to figure out why UBERON cites the E-node (entire from SNOMED CT’s representation of anatomy Structure-Entire-Part (SEP) triples) instead of the structure codes in SNOMED-CT. For example, the breast in UBERON is 0000310 http://purl.obolibrary.org/obo/UBERON_0000310 and the cited SNOMED CT code is SCTID:181131000 (entire breast) while the higher level (which includes part-whole distinction) is SCTID:76752008. This is true of any terms that I have looked through in UBERON. The citation is e-node and not s-node. I wanted to see what was the rationale for this decision to better understand the relationship between UBERON and SNOMED-CT.

[cmungall]

For context for everyone not familiar with SNOMED SEP model: https://confluence.ihtsdotools.org/display/DOCEG/Anatomical+Concept+Model

I think the E node is the correct one, since the uberon class represents the entity. More importantly, equivalence axioms can be made between E nodes without incoherency.

For example, snomedct:181500002 ! Entire skin of areola (body structure) is a subclass of the s-node 76752008. But the equivalent of this in Uberon would not be a subclass of breast, it would be a part of the breast, and entailing it is a subclass would likely lead to unsatisfiable classes