A Recipe for cDNA variants

[brendanreardon]

Summary of description of the proposed feature

While we currently have Recipes for genomic and protein variants, Canonical Allele and Protein Sequence Consequence, we do not have a Recipe for cDNA variants.

The Recipe should require:

• At least a Defining Allele Constraint
• Agreed upon relations for when a member is a genomic DNA variant, cDNA variant on another transcript, or protein variant

Product implementer

This Recipe is being requested by the MetaKB, a VICC driver project.

Justification for development with relevant use cases

The following cDNA variants are listed within our Test Set Variants

• NM_000410.4(HFE):c.187C>G (p.His63Asp)
• NM_000551.4(VHL):c.499C>T(p.R167W)
• NM_000245.4(MET):c.3028+1G>A

The MetaKB would like a Recipe with agreed upon relations because they need to convert all variants within CIViC that contain a cDNA change.

Example use cases, modeled using the current specification

Below is an example Categorical Variant for NM_000410.4(HFE):c.187C>G (p.His63Asp)

{
  "id": "clinvar:10",
  "type": "CategoricalVariant",
  "name": "NM_000410.4(HFE):c.187C>G (p.His63Asp)",
  "description": "",
  "aliases": [],
  "extensions": [],
  "constraints": [
    {
      "type": "DefiningAlleleConstraint",
      "allele": {
        "id": "ga4gh:VA.Vqmy0hqZKwLea6TPXsu2Zs2Hnl6EiqLu",
        "type": "Allele",
        "digest": "qmy0hqZKwLea6TPXsu2Zs2Hnl6EiqLu",
        "location": {
          "id": "ga4gh:SL.hrHdRAI_3mBVhT0rDCpsZXj9fRkkxLXP",
          "type": "SequenceLocation",
          "digest": "hrHdRAI_3mBVhT0rDCpsZXj9fRkkxLXP",
          "sequenceReference": {
            "name": "NM_000410.4"
            "type": "SequenceReference",
            "refgetAccession": "SQ.E2bJY0xOZpH26HWMo8vDf7y9tFd3xcc2"
          },
          "start": 198,
          "end": 199,
          "sequence": "C",
        },
        "state": {
          "type": "LiteralSequenceExpression",
          "sequence": "G"
        }
      },
      "relations": []
    }
  ]
}

Explanation of why the product cannot currently support the use case

Current officially supported Recipes are only satisfied for genomic DNA (Canonical Allele) and protein variants (Protein Sequence Consequence).

A link to the associated project board

They have an Issue open on their GitHub repository. There, Daniel and I previously workshopped a draft Recipe with them.

Anticipated open questions and design decisions

The two primary design decisions for this feature are likely:

• Do we want to add the restriction that this Recipe requires 1 and only 1 Constraint, as is currently implemented with the Canonical Allele and Protein Sequence Consequence?
• What relations should the Recipe require? This Discussion includes recommended relations for Defining Alleles depending on what molecule type the Defining Allele is and what molecular type the members are. There is another Discussion here that explores recommendations for what relations to use based on what the Defining Allele is and what a member is.

An example of a Recipe previously written for this use case with the relations of the following was included in the MetaKB Issue:

• translated_through_from (for p. members)
• transcribed_from (for cDNA members)
• gDNA_from

  cDNAProteinSequenceConsequence:
    maturity: trial use
    description: >-
      A change that occurs in a cDNA sequence that results in a protein sequence.
      The cDNA protein sequence consequence, like a :ref:`CanonicalAllele`, is defined by an
      :ref:`Allele`.
    allOf:
      - $ref: CategoricalVariant
      - properties:
          constraints:
            contains:
              allOf:
                - $ref: DefiningAlleleConstraint. This will be a c. variant
                - type: object
                  properties:
                    relations:
                      anyOf:
                        - contains:
                            type: object
                            properties:
                              primaryCoding:
                                code:
                                  const: translated_through_from (Protein > RNA > DNA) This is the p. variant member
                                system:
                                  const: _http://www.sequenceontology.org (Not actually an SO term that I know of)_
                              # This would be for a r. variant member
                              # primaryCoding:
                              #  code:
                              #    const: transcribed_from (RNA > DNA)
                              #  system:
                              #    const: http://www.sequenceontology.org
                              primaryCoding:
                                code:
                                  const: gDNAs_from (gDNA > cDNA) This is for a g. variant member
                                system:
                                  const: http://www.sequenceontology.org

Given the current relations of Canonical Allele and Protein Sequence Consequence, I think that appropriate relations to use when the cDNA is the Defining Allele would be:

• transcribed_to (if member is genomic DNA)
• transcribed_to (if member is cDNA on a different transcript)
• translated_to (if the member is a protein variant)

Maybe call the Recipe Transcript Sequence Consequence or Transcript Allele?