Add back VariantClassification to disinguish from VariantStudyStatement

[korikuzma]

For my use case, I need to be able to represent CIViC evidence and assertions. I don't think we have a model that makes this distinction.

Back in March, we had VariantClassification and VariantStudySummary to represent CIViC assertions and evidence respectively:

va-spec/schema/va-spec/core-im/core-im-source.yaml

Lines 214 to 237 in 4c14e9f

	VariantClassification:
	inherits: VariantStatement
	description: >-
	A :ref:`VariantStatement` classifying the impact of a variant.
	heritableProperties:
	classification:
	oneOf:
	- $refCurie: gks.core:Coding
	- $refCurie: gks.core:IRI
	description: >-
	A methodological, summary classification about the impact of a variant.
	heritableRequired:
	- classification
	VariantStudySummary:
	inherits: VariantStatement
	description: >-
	A :ref:`Statement` summarizing evidence about the impact of a variant from
	one or more studies.
	heritableProperties:
	isReportedIn:
	extends: isReportedIn
	minItems: 1
	heritableRequired:
	- isReportedIn

It would be nice to do something similar again. We'd want to enforce having hasEvidenceLines and classification in VariantClassification, but these fields would not exist in VariantStudySummary. Both models would still inherit from Statement.

@mbrush @larrybabb @ahwagner thoughts?

[ahwagner]

This sounds correct to me; thanks for digging into this and recommending this. To add to this, I'm not aware of any datasets that would require hasEvidenceLines on a VariantStudySummary, but like the shared structure on subject predicate object qualifier that you would find on classification statements.

[mbrush]

Hi @korikuzma Just dropping a line to let you know I see this, but am currently focused on the MAVE Profile work. I will share my thoughts/proposals for how I think the SEPIO-VA model might handle this in the next couple weeks, if that is OK.

[ahwagner]

@mbrush maybe it would make sense for @korikuzma to propose a solution here so we aren't gated by your availability? In VRS we often pass off issues arising from implementations to the reporting product developers.

[mbrush]

Absolutely. I didn’t mean to imply that a proposal can’t be drafted in the meantime. Just that there’s a history of discussion and ideas around modeling CIViC data using SEPIO/VA that I wanted to review in light of recent developments, and share for consideration.

I’ll drop a few thoughts and references here over the next couple days, and we can discuss/compare notes as your model drafts develop.

As for the idea of bringing back some intermediate abstract classes, happy to revisit this if you all think it is useful. But we’ll want to be careful that this has general utility beyond the immediate use cases, and doesn’t impede broader utility of the models and framework.

[mbrush]

Hi @korikuzma - I updated some of our earlier CIViC data example diagrams to reflect outcomes of recent VA profiling work - in particular ongoing MAVE Functional Impact profile efforts - and copied some relevant examples into the slides here. These might be useful to consult as you draft additional profiles - in particular how content of a CIViC EID record might get split between VA Evidence Line and StudyStatement objects. Feel free to reach out to me with questions.

We will be discussing/finalizing the MAVE Functional Impact profile work on a call Tuesday 10-22 at 2pm PT / 5pm ET. You are welcome to attend this call if Alex thinks it would be useful for you. Thanks!

[korikuzma]

@mbrush in your example, couldn't implementers still incorrectly add hasEvidenceLines to the civic:EID11797_study_statement (evidence item)? I think that is why we would like to split out these objects.

[mbrush]

Yes - the questions here are (1) whether these are constraints we want in a Standard Profile that will need to serve some adopters for which they may be too restrictive; and (2) if so, whether defining various abstract Statement subtypes as you propose is the best way to achieve this - given the consequences/implications this would have. I don't know the answer here - just posing the questions.