Possible invalid axioms for 'mesenchymal stem cell'

[balhoff]

When the full versions of go-plus and Uberon ext are used together, these two axioms result in unsatisfiable classes:

'mesenchymal stem cell' SubClassOf 'capable of' some 'cardiac muscle cell differentiation'
'mesenchymal stem cell' SubClassOf 'capable of' some 'osteoblast differentiation'

[cmungall]

I'm just going to paste the equivalent class here to give @dosumis nightmares:

'connective tissue cell'
 and ('has plasma membrane part' some endoglin)
 and ('has plasma membrane part' some 5'-nucleotidase)
 and ('has plasma membrane part' some 'Thy-1 membrane glycoprotein')
 and ('capable of' some 'osteoblast differentiation')
 and ('capable of' some 'cell migration')
 and ('capable of' some 'neuron differentiation')
 and ('capable of' some 'epithelial cell differentiation')
 and ('capable of' some 'fat cell differentiation')
 and ('capable of' some 'cardiac muscle cell differentiation')
 and ('capable of' some 'chondroblast differentiation')
 and (lacks_plasma_membrane_part some 'MHC class II protein complex')
 and (lacks_plasma_membrane_part some 'CD34 molecule')
 and (lacks_plasma_membrane_part some 'receptor-type tyrosine-protein phosphatase C')

[mellybelly]

Maybe we should consider a refactor where such things move to an annotation file. On Jun 3, 2017, at 5:06 PM, Chris Mungall <notifications@github.com<mailto:notifications@github.com>> wrote: I'm just going to paste the equivalent class here to give @dosumis<https://github.com/dosumis> nightmares: 'connective tissue cell' and ('has plasma membrane part' some endoglin) and ('has plasma membrane part' some 5'-nucleotidase) and ('has plasma membrane part' some 'Thy-1 membrane glycoprotein') and ('capable of' some 'osteoblast differentiation') and ('capable of' some 'cell migration') and ('capable of' some 'neuron differentiation') and ('capable of' some 'epithelial cell differentiation') and ('capable of' some 'fat cell differentiation') and ('capable of' some 'cardiac muscle cell differentiation') and ('capable of' some 'chondroblast differentiation') and (lacks_plasma_membrane_part some 'MHC class II protein complex') and (lacks_plasma_membrane_part some 'CD34 molecule') and (lacks_plasma_membrane_part some 'receptor-type tyrosine-protein phosphatase C') — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub<#474 (comment)>, or mute the thread<https://github.com/notifications/unsubscribe-auth/AAwn3PPoHu5LbIaXLKt23h8UhSGuxQutks5sAXZagaJpZM4NvGYz>.

[dosumis]

When the full versions of go-plus and Uberon ext are used together, these two axioms result in unsatisfiable classes:

'mesenchymal stem cell' SubClassOf 'capable of' some 'cardiac muscle cell differentiation'
'mesenchymal stem cell' SubClassOf 'capable of' some 'osteoblast differentiation'

I suspect the problem is the taxonomic broadness of cardiac muscle cell def. Do you have an explanation handy?

I'm just going to paste the equivalent class here to give @dosumis nightmares

You succeeded. I'm skeptical of the biology, but even if true for mice or humans, the EC expression is still completely useless for automated classification.

Maybe we should consider a refactor where such things move to an annotation file.

This has been suggested in the past. I'd certainly like to move the membrane markers out into some separate file marked as applicable to mouse &/or human.

[cmungall]

As a first pass I was going to weaken to a taxon-specific equivalence, e.g.

(CL and part-of some NCBITaxon) EquivalentTo (....long complex expression...)

@davidos criticisms still apply, but at least we avoid outright logical incoherency and we can rapidly implement this, and it's easy to apply future refactoring

[balhoff]

@dosumis here is the example I had put in my original oborel issue.

[dosumis]

Ta. Looks like very different usage of the term 'mesenchymal stem cell' between flies and mammals.