Open caufieldjh opened 1 year ago
I would be curious to understand what the biolink vision for this is.
Note that some people in Translator actively want conflation between e.g. a gene, the corresponding reference protein, and the corresponding OMIM mendelian disease entry, which are often all in 1:1. I don't think this is good, but biolink is currently quite lenient when it comes to formal disjointness declarations.
For our purposes, we want to treat the main is-a hierarchy as pairwise disjoint.
of course, something can be both a pathway and a mixin like OntologyClass
even in cases where we do want to conflate e.g gene to refP, we should pick one, so we should treat the is-a hierarchy as pairwise disjoint
A graph may be assigned multiple Biolink (or other) categories, but these should adhere to a model. Including both a child and a parent category is not a violation of the model. Including two or more categories exclusive of each other (like
biolink:Disease|biolink:Device
) would be a violation. But how do we define this? At minimum, the category should be the same as that of the type entity. If a type entity isn't available, we should be able to find a path between the two categories without involving a root.