Model with activity enabled_by compound

geneontology / minerva

BSD 3-Clause "New" or "Revised" License

6 stars 8 forks source link

Model with activity enabled_by compound #302

Closed lpalbou closed 4 years ago

lpalbou commented 4 years ago

In this model, @tmushayahama was able to create a model with a compound as an enabler of an activity (see Kanamycin A on the right).

[ ] autocomplete should not allow for the selection of a compound (not using CHEBI molecular_entity) as an enabler but probably only biomacromolecule (CHEBI_33694) @cmungall ?
[ ] would shex currently catch that mistake or do we have to update the schema ? (@goodb ?)

ukemi commented 4 years ago

If you run the reasoner the little red error box appears in the upper left showing a Shex failure. If you click on the box you get an unfriendly but accurate description of why it failed.

ukemi commented 4 years ago

I should probably also point out that this works on development.

lpalbou commented 4 years ago

Thanks for the feedback, two comments:

The autocomplete probably should still not allows curators to enter compound as enabler
I don't think we have a SHEX validator on noctua form ? (@tmushayahama). I thought that each time the model is updated, noctua form was receiving a response with possible errors to highlight in the UI ?

tmushayahama commented 4 years ago

@lpalbou

That is the same closure on shex and there is no validation on noctua form

vanaukenk commented 4 years ago

@tmushayahama We should restrict the 'Gene Product' field to only allow for an InformationBiomacromolecule (CHEBI:33695) OR a ProteinContainingComplex (GO:0033291)

tmushayahama commented 4 years ago

@vanaukenk thanks, so I will put also OR protein containing complex.

@lpalbou @goodb double check to align with shex, Golr will be ... fq: document_category:"ontology_class" fq: isa_closure:"CHEBI:33695" OR isa_closure:"GO:0032991" q: keyword ...

Thanks @thomaspd for the catch

lpalbou commented 4 years ago

@tmushayahama as I understand, you are doing GOLr queries on the fly to fill your autocomplete ? It seems you were using a closure of CHEBI_23367 (molecular_entity), so indeed changing it to CHEBI_33695 would do the trick and with a OR:

fq=isa_closure:("CHEBI:33695" "GO:0032991")

This issue is related to the ticket on autocomplete: https://github.com/geneontology/minerva/issues/236 . I would suggest autocomplete to become a priority for the next cycle after May as we need to have something more unified across UIs and SHEX.

vanaukenk commented 4 years ago

@lpalbou

We are working towards aligning the autocompletes on the form with the ShEx.

See, for example: https://github.com/geneontology/noctua-form/issues/66

It is definitely on our radar, although with the caveat that the ShEx has still been under discussion a bit on the GO-CAM specs calls.

Thx.

vanaukenk commented 4 years ago

Tested on noctua-dev and appears to be working as expected.