ahwagner
commented
2 months ago This seems like a duplicate of #16; I suggest we keep the discussion open but close the issue until there is consensus on developing this feature, as part of our GKS Technical Specification Maintenance process.
Currently there are two issues with the current structure of the schema that I would like to address.
First, the CanonicalAllele and Protein SequenceConsequence both exist and are more or less identical, but are simply meant to describe sequence variants at two different levels, genomic and protein, respectively. While classes are cheap, DRY. I want to explore the option to subsume these two classes under a more general SequenceVariant class (name can be contested) that keeps the canonicalized properties of both, while differentiating genomic from protien variants by some sort of required level: one of {"genomic","protein"} property.
Second, while sequence variants are handled by CanonicalAllele or the new SequenceVariant class , and CNVs are handled by the CategoricalCNV class (subject to ay changes as a result of issue-11), there is a coherent class of variants not covered so far: Function Variants. Here's some suCatVars in this category:
These variants all entail some sort of assayed quantitative change, often but not always associated with some sort of sequence change. That is, capturing this class of CatVars requires properties from both the CategoricalCNV class and the SequenceVariant class. We could either create a third class that just happens to contain the same classes as both of those, or I propose we make a third class that is a generalization of the notion of quantity change and sequence change variants, subsuming both CategoricalCNV and SequenceVariant into a larger class, which, by way of optional properties, can handle CNVs, sequence variants, AND intermediate cases like the function variants described above.