Open cmungall opened 5 years ago
It seems like we would need to define a property has part inhering in
or phenotype of
that would be a superproperty of 'has part' o 'inheres in'
, and then infer these restrictions using the EMR.
Oh, I guess UPheno already defines has phenotype affecting, which would work.
That would work, I believe it propagates over part-of too
This does not pertain to the patterns but to the release workflows of each pheno ontology.
Assuming that we have a shared release workflow where a basic obo file is available, as well as a complete OWL file:
Normally in releases we run
robot relax
which includes a basic subclass axiom for every complex equivalence axiom. E.g. in the cl release we include `hippocampal neuron SubClassOf part-of some hippocampus' (based on a logical definition). This is very useful for an increasing number of semi-advanced consumers who are interested in complete knowledge graphs but are still not facile with complex OWL.Unfortunately relax will not provide useful results over pheno ontologies due to the nesting pattern/subq. We can't infer
'increased bone density' SubClassOf 'inheres_in' some bone
from the logical definitions.We could just have custom SPARQL queries to make tables but this is unsatisfactory and ad-hoc, ideally we could have basic existential graphs that contain information of use
(to see what I mean by basic existential graph and for my general philosophy on this see https://douroucouli.wordpress.com/2016/10/04/a-developer-friendly-json-exchange-format-for-ontologies/)
Any ideas how to deal with this @dosumis @matentzn @balhoff? Shortcut relations that can be inferred?
Or should we just accept that we need to use an intermediate blank node?
This is increasingly necessary for projects like Translator and IDG-KG, cc @deepakunni3