mgijax / mammalian-phenotype-ontology

Standard terms for annotating mammalian phenotypic data
Creative Commons Attribution 4.0 International
19 stars 6 forks source link

MP+HP together cause some MP unsats #3458

Open matentzn opened 3 years ago

matentzn commented 3 years ago

absent cochlear hair cells SubClassOf Nothing

absent cochlear outer hair cells SubClassOf Nothing

absent cochlear inner hair cells SubClassOf Nothing

Axiom Impact

Axioms used 3 times

Axioms used 2 times

Axioms used 1 times

Ontologies used:

@dosumis @balhoff can you pinpoint the problem?

matentzn commented 3 years ago

FYI @viomunoz, this is why the IMPC pipeline currently fails.

sbello commented 3 years ago

@matentzn are you looking at the latest version of the MP file? The eq I see for absent cochlear outer hair cell is: 'has part' some (absent and ('characteristic of' some 'outer hair cell') and ('has modifier' some abnormal))

absent inner hair cells is: 'has part' some (absent and ('characteristic of' some 'inner hair cell') and ('has modifier' some abnormal))

absent cochlear hair cells is 'has part' some (absent and ('characteristic of' some 'auditory hair cell') and ('has modifier' some abnormal))

dosumis commented 3 years ago

This issue has come up in a related context recently:

https://github.com/obophenotype/uberon/issues/1905

The problem is that the cochlea is classified as bone, but the sensory epithelium that lines it ('spiral organ of cochlea' AKA the organ of Corti) has a separate developmental origin (see https://journals.biologists.com/dev/article/147/12/dev162263/224272/Development-of-the-cochlea ). This breaks the spatial disjointness of skeletal system and nervous system.

I suspect that the reason Uberon is not inconsistent is because it breaks the part_of chain with a 'located in' https://github.com/obophenotype/uberon/issues/1513 (although I'm looking in to whether the placement of this is optimal). The phenotype terms add this back with a parthood relationship on the cell type.

Quick fix(ish): change MP/HP patterns. (Update - looks like Sue got there before me).

Agree with this:

cochlear outer hair cell is: 'has part' some (absent and ('characteristic of' some 'outer hair cell') and ('has modifier' some abnormal))

absent inner hair cells is: 'has part' some (absent and ('characteristic of' some 'outer hair cell') and ('has modifier' some abnormal))

For this - I would use:

absent cochlear hair cells is 'has part' some (absent and ('characteristic of' some ('auditory hair cell' and part of some 'spiral organ of cochlea') and ('has modifier' some abnormal))

This is redundant for now but CL is (as usual) much too mammal-centric here. Auditory hair cells exist in other contexts in vertebrates: https://en.wikipedia.org/wiki/Sensory_systems_in_fish. I'll open a ticket on the CL tracker to suggest broadening the definition or auditory hair cell and making a new cochlea subclass of this (following the composition pattern above).

For these two tickets

https://github.com/obophenotype/uberon/issues/1905 https://github.com/obophenotype/uberon/issues/1513

I think it might be sufficient to leave Uberon unchanged - or at most to move the located_in relationship down to 'organ of corti'

sbello commented 3 years ago

I'll make the change to the eq for absent cochlear hair cell and push out a new release ASAP.

matentzn commented 3 years ago

Awesome thank you all! :)

When you are done Sue, in case you have ROBOT installed, you can test like this:

robot merge -i mp.owl -I http://purl.obolibrary.org/obo/hp.owl reason -o x.owl

If not, I can also test it once you have made the changes.