Are recipes exclusive?

[jsstevenson]

The ProteinSequenceConsequence description includes the following

The ProteinSequenceConsequence is a Categorical Variant with exactly one constraint:

A DefiningAlleleConstraint with the .relations array containing only a translation_of code. This constraint MUST refer to a protein variant for the allele.

I think this description seems pretty definitive that a DefiningAlleleConstraint with a translation_of relation AND something else (e.g. liftover_from) would not qualify (and the schema appears to reflect this as well). But I think at other times, I've heard other definitions or interpretations, e.g. that if I have a DefiningAlleleConstraint with .relations=["liftover_from","translation_of"] then that still meets the ProteinSequenceConsequence recipe. Is there a definitive ruling?

[brendanreardon]

Thanks for bringing this up! I agree, we should probably add liftover_to to Protein Sequence Consequence to include members on alternative genome builds. Flagging that this is a bit related to the open language question in this Pull Request regarding Recipes.

To directly answer your question, I don't think that it is definitive. I cannot find the citation right now within GitHub or on Slack, but there was a conversation with @DanielPuthawala and @larrybabb somewhere about where GKS is defining these terms that are not mapped to another ontology.
edit: For example, https://github.com/ga4gh/cat-vrs/blob/1.0/examples/canonicalAllele-ex1.yaml#L102-L104 liftover-to is currently a gks term but is not currently defined anywhere.