search

obophenotype / mouse-anatomy-ontology

Ontologies for mouse anatomy and development
Other
5 stars 3 forks source link

Update hash relationships #127

Open cthoyt opened 1 year ago

cthoyt commented 1 year ago

There are several hash relationships in EMAPA (i.e., ones locally defined within the ontology but not using the designated URI prefix http://purl.obolibrary.org/obo/EMAPA_).

This is a problem for interoperability as these are not well defined. This is also a problem for parsing since it's not clear how to incorporate these hash relationships into the EMAPA identifier space.

Some examples (non-exhaustive) I got when parsing EMAPA with bioontologies:

http://purl.obolibrary.org/obo/emapa#Tmp_new_group                                                                      
http://purl.obolibrary.org/obo/emapa#ends_at                                                                            
http://purl.obolibrary.org/obo/emapa#group_term                                                                         
http://purl.obolibrary.org/obo/emapa#starts_at
http://purl.obolibrary.org/obo/emapa#is_a

The last one is particularly worrying since an "is a" relationship is well-defined in OBO world

To Do

  1. http://purl.obolibrary.org/obo/emapa#is_a, http://purl.obolibrary.org/obo/emapa#starts_at, and http://purl.obolibrary.org/obo/emapa#ends_at probably map onto things in the relation ontology (this is already covered by #125)
  2. http://purl.obolibrary.org/obo/emapa#Tmp_new_group and http://purl.obolibrary.org/obo/emapa#group_term aren't qualified with a namespace and are completely unused. These are removed in #128.
cthoyt commented 1 year ago

I emailed Terry about this on May 4, 2023

matentzn commented 1 year ago

Just to note, that there is a bit of a steep priority issue here:

Level 1: OWL constructs (Importance 10/10)

http://purl.obolibrary.org/obo/emapa#is_a <- is precludes any kind of logical processing and should be fixed as a priority

Level 2: Object properties that can be aligned with RO (Importance 6/10)

http://purl.obolibrary.org/obo/emapa#starts_at <- aligning this with RO is necessary to facilitate interoperability with other stage ontologies

Level 3: Fixing strange class IRIs (Importance 4/10)

http://purl.obolibrary.org/obo/emapa#group_term This smells like a subset declaration gone wrong in EMAPA. So basically, someone wanted to declare a subset property (which is an annotation property) and ended up accidentally creating a class declaration. I guess the question is if these are even used?

cthoyt commented 1 year ago

For 3: good idea, I just checked and see these actually aren't used at all