remove hash IRIs for properties

[balhoff]

Mondo has a number of annotation and object properties with hash IRIs that don't resolve. These kinds of IDs can't be compacted using the MONDO CURIE prefix and lead to confusion. I guess there may be an intent to submit these to RO at some point, but I suggest immediately giving them all real MONDO numeric IDs and later obsoleting if desired:

Annotation properties:

• http://purl.obolibrary.org/obo/mondo#excluded_subClassOf
• http://purl.obolibrary.org/obo/mondo#excluded_synonym
• http://purl.obolibrary.org/obo/mondo#pathogenesis
• http://purl.obolibrary.org/obo/mondo#related

Object properties:

• http://purl.obolibrary.org/obo/mondo#disease_causes_feature
• http://purl.obolibrary.org/obo/mondo#disease_has_basis_in_accumulation_of
• http://purl.obolibrary.org/obo/mondo#disease_has_basis_in_development_of
• http://purl.obolibrary.org/obo/mondo#disease_has_major_feature
• http://purl.obolibrary.org/obo/mondo#disease_responds_to
• http://purl.obolibrary.org/obo/mondo#disease_shares_features_of
• http://purl.obolibrary.org/obo/mondo#disease_triggers
• http://purl.obolibrary.org/obo/mondo#has_onset
• http://purl.obolibrary.org/obo/mondo#has_onset_before
• http://purl.obolibrary.org/obo/mondo#has_onset_during_or_after
• http://purl.obolibrary.org/obo/mondo#part_of_progression_of_disease
• http://purl.obolibrary.org/obo/mondo#predisposes_towards

[nicolevasilevsky]

thanks @balhoff, I'll bring this up with Nico on the technical call

[matentzn]

The essence of this request is this:

1. @cmungall has a mild aversion against properties having opaque ids, because they make it harder to read OBO files and therefore increase the complexity of reviewing pull requests
2. The ship for readable IRIs in general has sailed, and if we move these relationships to RO, they will become opaque anyways.

I think we should do the following:

• Prepare a submission of all these disease relations to RO asap. This involves creating a ROBOT template with at least 3 columns: ID, label, definition and parent relation (and any semantics we might have asserted in Mondo).
  • Getting these reviewed by Lynn
  • Adding them to RO.
• Drop non-standard annotations:
  • Remove two non-standard properties from Mondo #5621
  • Consider removing http://purl.obolibrary.org/obo/mondo#excluded_subClassOf and http://purl.obolibrary.org/obo/mondo#excluded_synonym from the release file, or alternatively using a standard formatted IRI do encode these (numeric MONDO:123 with a label for readability).
• Add a QC against these properties. We should have a strict policy for using only annotation properties that have been vetted.

[balhoff]

1. @cmungall has a mild aversion against properties having opaque ids, because they make it harder to read OBO files and therefore increase the complexity of reviewing pull requests

But keep in mind, with the right annotations you can make the OBO file use property names, e.g. has_primary_input here https://github.com/geneontology/go-ontology/pull/24379/files

[matentzn]

I didnt know you could do that! Or maybe I did, and just forgot. Thanks!

[nicolevasilevsky]

Note, from time to time, we need to review this. Raise this ~ once per month with Nico in your one-on-one meetings @sabrinatoro

[matentzn]

From the meeting, we decided:

1. Migrate the Object Properties asap (Q1 2023) into RO
2. Using http://purl.obolibrary.org/obo/mondo# only for temporal relations during editing, and stripping them out upon release.

Expected time of completion: March 2023.