taxon disjoints missing from locally built uberon.owl

[balhoff]

Please correct me if I'm missing it, but as far as I can tell, the Uberon build doesn't use taxslim-disjoint-over-in-taxon.owl anywhere, or generate the axioms it contains, like:

('in taxon' some Opisthokonta) DisjointWith ('in taxon' some Viridiplantae)

Without these axioms, taxon constraints don't do anything in OWL EL. I would expect to find these GCIs in the final Uberon product. I'm running into this while trying to test some edits for #2349.

[balhoff]

I do see a disjoint axiom like that in #2367; can someone explain where they're used?

[balhoff]

Now I see that these axioms are present in the released ext.owl, but not the version I built locally by running make ext.owl. What am I doing wrong?

[matentzn]

This is so weird. In ext.owl they must be coming in from through CL or something..

Can you summarise the functional problem with these axioms missing? I think that the for the subset generation, which is the place using the taxon constraints are used in Uberon, some other owltools based step is invoked after the fact, but the disjointness axioms are not shipped out.. From your understanding, should the taxon disjointness axioms be present in the release? So that it works with, for example, your OBO taxon plugin?

[balhoff]

I think the taxon constraints should be working for any DL coherency check, not just to generate subsets. If you don't have axioms like ('in taxon' some X) DisjointWith ('in taxon' some Y), then there is no way to infer an unsatisfiable class based on taxon constraints (using OWL EL). Personally I think the taxon disjointness axioms should be in the release. They're definitely needed for the Protege taxon constraints plugin to work.

To go further in helping out ELK here, we also need ('in taxon' some X) DisjointWith ('in taxon' some (not X)) (at GO).

[matentzn]

@cmungall Do you see any reason not to release the unfolded taxon constraints in Uberon?

[cmungall]

Do you see any reason not to release the unfolded taxon constraints in Uberon?

To be pedantic, axioms like

('in taxon' some X) DisjointWith ('in taxon' some Y)

are not themselves taxon constraints. This is part of the axiomatization of NCBITaxon.

The documentation for this is very buried:
https://github.com/obophenotype/ncbitaxon/blob/master/subsets/README.md

I have no problem in incorporating these in a merged release file accompanying the taxon constraints themselves. But be aware from the point of view of assembly non-redundant subsets of ontologies using base files, these would not be included. They should come from some kind of ncbitaxon-base, were such a thing to exist

[shawntanzk]

Action item:

• Figure out where ('in taxon' some X) DisjointWith ('in taxon' some Y) live (look through uberon.makefile)
• Specify new component for these axioms (make sure to only import needed axioms - ie related to taxa we use)
• Make PR and see what breaks

[github-actions]

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

[anitacaron]

• Figure out where ('in taxon' some X) DisjointWith ('in taxon' some Y) live (look through uberon.makefile)

This is not in the pipeline right now. We need to merge the taxslim-disjoint-over-in-taxon.owl as Jim said.

A related issue that will fix this #2707

[anitacaron]

After refactoring the Makefile, the ext.owl artefact was removed because it was the same as uberon.owl. For now, the disjoints are only added in the QC.

[github-actions]

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

[gouttegd]

Taxon disjointness axioms are now included in uberon.owl, for all taxa that are imported by Uberon. Closing here.