atom disjointWith molecular entity infects multiple ontologies with incoherency

[cmungall]

Integration test for OBI currently fails:

This is obv not OBI's fault, it is coming from CHEBI. We can make OBI pass by testing its base file.

However, the fact remains that the main OBI product (which includes a subset of CHEBI) is incoherent w.r.t COB. This will be true for many others, e.g GO.

We can

• work with chebi to rework their upper level
• relax the disjointness in cob
• remove the equivalence in cob to external
• do nothing for now, and accept that no one can use cob in the short term except as a means of reporting these kinds of issues

[bpeters42]

Does Chebi by itself pass reasoning with COB? Presumably it shouldn't? That would just be the more direct way to report the problem. I would argue that currently we should 'in the short term use COB to reporting these kind of issues', and let the Chebi people know about it. Who can then either work on their upper level, or make the argument that COB should be updated. Or they do nothing, which if it persists would indicate that Chebi is not able to react to community requests sufficiently to be considered a OBO foundry reference ontology. - Bjoern

…

On Wed, May 6, 2020 at 11:10 PM Chris Mungall ***@***.***> wrote: Integration test for OBI currently fails: [image: image] <https://user-images.githubusercontent.com/50745/81260079-21c88800-8fee-11ea-8e5b-aebcac7236a4.png> This is obv not OBI's fault, it is coming from CHEBI. We can make OBI pass by testing its base file. However, the fact remains that the main OBI product (which includes a subset of CHEBI) is incoherent w.r.t COB. This will be true for many others, e.g GO. We can - work with chebi to rework their upper level - relax the disjointness in cob - remove the equivalence in cob to external - do nothing for now, and accept that no one can use cob in the short term except as a means of reporting these kinds of issues — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#71>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADJX2IVSYRZ6AIQRM6V5RVDRQJGE3ANCNFSM4M3BDKUQ> .

-- Bjoern Peters Professor La Jolla Institute for Allergy and Immunology 9420 Athena Circle La Jolla, CA 92037, USA Tel: 858/752-6914 Fax: 858/752-6987 http://www.liai.org/pages/faculty-peters

[beckyjackson]

@jannahastings could you please follow up with ChEBI on reworking their upper level structure? Thank you!

[jannahastings]

@jannahastings could you please follow up with ChEBI on reworking their upper level structure? Thank you!

Okay, I'll ping them :-)

[jannahastings]

The ChEBI team (Adnan) sent me this proposal, which I believe was also sent to @cmungall earlier:

With regards to your suggestion to the uniting of atoms and their corresponding ions in ChEBI. If I remember correctly, your suggestion was to create a class in ChEBI called atom (E.g. Nickel atom), and then have child classes such as the 'uncharged nickel atom' and its corresponding 'nickel ion' linked to nickel atom. This way the atom and its corresponding ion are in the same branch of the ontology tree. After discussing your idea with others in my group, it was decided that rather than undertaking a major restructuring of this area of the ontology, it would be a lot easier to create for each atom that has ionic forms new relationships such as 'is charged particle formed from' and 'forms charged particle'.
e.g.
nickel(1+) is_charged_particle_formed_from nickel atom
and the reverse:
nickel atom forms_charged_particle nickel(1+)

Comments?

[cmungall]

@jannahastings @amalik01 you already know this, but filling in details for everyone.

My atom proposal was discussed at the 2020 COB workshop, I also included it in the slidedeck I have on simplifying CHEBI:

https://docs.google.com/presentation/d/1R3NRzH70ERjwebqecgt2OYKC8sIuB1_Xs7ENyWAXgjc/edit#slide=id.g2e88609578d_0_11

When I discussed this with @amalik01 and colleagues they seemed open to fixing the atom hierarchy in CHEBI