search

ZFIN / zebrafish-anatomical-ontology

Ontology describing the anatomy of Danio rerio from a single cell to adult fish. The namespace of the ontology is zebrafish_anatomical_ontology the prefix is ZFA. Updates to ZFA are released every other month. A subset of the ZFS http://purl.obolibrary.org/obo/zfs.obo is packaged with this ontology.
Other
6 stars 6 forks source link

immaterial entities and RO compatibility #137

Closed balhoff closed 6 years ago

balhoff commented 6 years ago

Hi,

I'm wondering what the prospects are for making any updates to ZFA to increase compatibility with MP and HPO (via Uberon mappings). For example, this issue highlights a problem when ZFA 'anatomical space' is equated to Uberon 'anatomical space', mostly due to high level domain and range constraints in RO:

https://github.com/obophenotype/upheno/issues/245

For this specific case, it might require adding 'located in' to ZFA and changing some 'part of' relationships to 'located in'. But more generally, I think there are some additional incompatibilities that will need to be addressed to make everything satisfiable.

What do you think? I would be happy to propose some changes if it seems feasible to incorporate them.

cerivs commented 6 years ago

Changing the partonomy will cause issues with the search functions at ZFIN. We don't have the personal free to look at the implications until mid 2018 at the earliest some changes may be possible but review of each proposal will need to be done to make sure we aren't loosing site functionality by making the change.

balhoff commented 6 years ago

Hi @cerivs - thanks for the feedback. I think we have a workaround now which won't require mixing these ZFA axioms with incompatible Uberon axioms. I created a zfa_import module for use with Upheno.

balhoff commented 6 years ago

Hi @cerivs and @ybradford,

I just wanted to leave one more comment here about another aspect of this issue. The ZFIN data files (e.g. http://zfin.org/downloads/phenoGeneCleanData_fish.txt) allow some post-composition and thus provide another spot for this issue to creep in. Within Upheno and also Phenoscape we're translating ZFIN phenotype annotations into OWL classes, and so any phenotypes that specify a "subterm" entity which is part of a "superterm" anatomical space will end up being unsatisfiable if ZFA gets bridged to BFO/RO.

Please let me know if there's a better place to leave this comment—I just wanted to provide the feedback so that in case you make any changes to the ontology you might look at the data also. For the time being I may add a check to our tool which swaps in located in when the superterm is an anatomical space, to make things work better when all these ontologies are combined together.

cc @sbello