UBERON:0007685 'region of nephron tubule' is both part of and subclass of 'nephron tubule'

[paolaroncaglia]

Uberon term UBERON:0007685 'region of nephron tubule'

Bug description Is both (asserted) part of nephron tubule and (inferred) subclass of nephron tubule Explanation in screenshot attached:

I came across this by chance.

• [ ] Are Uberon classes allowed to be both subclass of and part of the same superclass?
• [ ] If not, as I suspect, are there other instances in Uberon that may need fixing? And
• [ ] Is there a post-reasoning check in place to flag such instances?
• [ ] For this specific case, replace 'region of nephron tubule' subclass of 'epithelial tube' with subclass of 'organ part' (or a more granular term if anyone would like to suggest one). Thanks.

[paolaroncaglia]

@matentzn could you please comment on the first 3 checkboxes or tag others as appropriate, and I'll address the 4th one (but I'll leave it as is for now in case it's useful to run checks). Thank you.

[matentzn]

Are Uberon classes allowed to be both subclass of and part of the same superclass?

Definitely not

Is there a post-reasoning check in place to flag such instances?

Seems not! This is an important and quite advanced qc query for @anitacaron. Basically:

Look for

• all UBERON ids
• that have an existential restriction with a part-of property to some UBERON:A
• and are children (whether direct or indirect, use *) of UBERON:A

[matentzn]

One problem here is that we, again, have no access to the part of closure.. Maybe this can be solved with souffle rather than SPARQL?

[paolaroncaglia]

Are Uberon classes allowed to be both subclass of and part of the same superclass?

Definitely not

Thought so ;-) thanks @matentzn and @anitacaron for addressing this.

[anitacaron]

One problem here is that we, again, have no access to the part of closure.. Maybe this can be solved with souffle rather than SPARQL?

Actually, it's relation-graph that materialise and infer the existential axioms

[matentzn]

Its I think solvable with this: https://github.com/obophenotype/uberon/pull/2125

[balhoff]

I might be misunderstanding the issue, but I think it would be quite common to be part of and also a subclass of certain classes, if you are including inferences in this. A bone is both part of an 'anatomical structure' and also is an 'anatomical structure'.

[matentzn]

Hmmm you are right.. Sorry it was I who was too hasty here. You can indeed have the same parent and be part of it - its just not very often.. Sorry. Maybe there is, after all, no general check we can have that protects us here..

[paolaroncaglia]

Hi @balhoff and @matentzn , I agree that at a high level all UBERON classes will be anatomical structures/entities via is_a as well as part_of. Is there a way to have a check in place at some lower level in the ontology? The example that spurred this ticket is just intuitively incorrect, a region of a tubule can't really be a type of tube. :-)

[matentzn]

I think investigating this is worth it, but I have a hard time figuring out how to draw the line between "upper" and "lower" level. Can a limb be part of a limb? Very hard to be sure..

It is probably better to solve this with logic, i.e. you identify some mid level classes for which the above is definitely false, and then we add an axiom saying: limb disjoint with part of limb.

[paolaroncaglia]

This ticket is in the Todo list of the Tech board now, so I'll remove my assignment (won't edit the example I provided in case it's useful to run checks and also because a consensus hasn't been reached yet).

[github-actions[bot]]

This issue has not seen any activity in the past 6 months; it will be closed automatically in one year from now if no action is taken.

[github-actions[bot]]

This issue has been closed automatically because it has not been updated in 18 months. Please re-open if you still need this to be addressed.