Closed cmungall closed 3 years ago
Should
We can also include opposites within the model itself
We can also define fuzzy property chains:
be their own issues?
The concept of "opposing phenotypes" has come up in drug repurposing (DR) use case discussions. Would these predicates be valid for making a query for finding drugs that might be repurposed because they have adverse effects that may be used to treat a disease with an opposing phenotype? I'm also wondering how to handle some phenotypes which don't have an obvious "opposite." In this case I'm sure the answer is just that the predicate won't exist, so that's probably going to be a problem for the DR use case.
I'm also curious how the inference logic would follow for this relation. Would children of node A which is opposite_of node B also be opposite of node B?
inverse of is precisely defined,'opposite of' might be less precisely defined? (TODO: sierra make linkml ticket and link it here to discuss in linkml). domain predicate in linkml, but first class object in BMT
https://github.com/biolink/biolinkml/pull/396 - proposal for adding antonymy to the linkml meta model so that we can set 'opposite' flags (similar to symmetric or inverse).
Colleen's PR #681 which includes inverses could be relevant to this discussion. As Sierra mentioned, we could add "opposite" as a new attribute in addition to "inverse" as in:
increases_activity_of / activity_increased_by (inverse) decreases_activity_of / activity_decreased_by (inverse) increases_activity_of / decreases_activity_of (opposite) activity_increased_by / activity_decreased_by (opposite)
Exact mapping: RO:0002604
see_also: https://doi.org/10.1101/108977
Example of how this is used for phenotypes:
http://www.ontobee.org/sparql SELECT ?x ?y WHERE {?x http://purl.obolibrary.org/obo/RO_0002604 ?y}
Note: many people confuse inverseOf and opposite-of. Remember to clearly state this
Examples of opposites between nodes:
We can also include opposites within the model itself (made a separate ticket for this #657)
We can also define fuzzy property chains: