Intended scope of molecular variation Expressions?

[jsstevenson]

I've noticed a few cases (not directly in the VRS docs, but in related places) where an Allele has multiple provided Expressions utilizing different molecule types than its own, such as the example below where the variation itself is on a transcript but the provided expressions include g. and p.. On the one hand, these are technically different things, but on the other, they are strongly implied by one another. Is this a big deal, or is this field intentionally broad? The schema itself doesn't have any additional description for the Expressions field (might be helpful if there is some narrow definition intended)

      {
         "id": "ga4gh:VA.gV7_dnvF8SQSeUdvgDFhU65zK_csc6VE",
         "type": "Allele",
         "digest": "gV7_dnvF8SQSeUdvgDFhU65zK_csc6VE",
         "expressions": [
            {
               "syntax": "hgvs.g",
               "value": "NC_000007.13:g.55259515T>G"
            },
            {
               "syntax": "hgvs.c",
               "value": "NM_005228.5:c.2573T>G"
            },
            {
               "syntax": "hgvs.p",
               "value": "NP_005219.2:p.Leu858Arg"
            }
         ],
         "location": {
            "id": "ga4gh:SL.LREsUiEYvOrRhwXW1rG72kXFPegvkNzI",
            "type": "SequenceLocation",
            "digest": "LREsUiEYvOrRhwXW1rG72kXFPegvkNzI",
            "sequenceReference": {
               "id": "refseq:NM_005228.5",
               "type": "SequenceReference",
               "refgetAccession": "SQ.d_QsP29RWJi6bac7GOC9cJ9AO7s_HUMN",
               "residueAlphabet": "na"
            },
            "start": 2833,
            "end": 2834,
            "sequence": "T"
         },
         "state": {
            "type": "LiteralSequenceExpression",
            "sequence": "G"
         }
      },

[korikuzma]

@jsstevenson I'm just now getting the the CIViC Molecular Profile refactor in CIViC. I do think this is wrong and should only include the hgvs.c expression in the above example...

This is the Cat Var that CIViCpy produces (at the moment, we can update if it's wrong):

{
          "id": "civic.mpid:12",
          "type": "CategoricalVariant",
          "name": "BRAF V600E",
          "description": "BRAF V600E has been shown to be recurrent in many cancer types. It is one of the most widely studied variants in cancer. This variant is correlated with poor prognosis in certain cancer types, including colorectal cancer and papillary thyroid cancer. The targeted therapeutic dabrafenib has been shown to be effective in clinical trials with an array of BRAF mutations and cancer types. Dabrafenib has also shown to be effective when combined with the MEK inhibitor trametinib in colorectal cancer and melanoma. However, in patients with TP53, CDKN2A and KRAS mutations, dabrafenib resistance has been reported. Ipilimumab, regorafenib, vemurafenib, and a number of combination therapies have been successful in treating V600E mutations. However, cetuximab and panitumumab have been largely shown to be ineffective without supplementary treatment.",
          "aliases": [
            "VAL600GLU",
            "V640E",
            "VAL640GLU"
          ],
          "extensions": [
            {
              "name": "CIViC Molecular Profile Score",
              "value": 1433.5
            },
            {
              "name": "expressions",
              "value": [
                {
                  "syntax": "hgvs.c",
                  "value": "NM_004333.4:c.1799T>A"
                },
                {
                  "syntax": "hgvs.p",
                  "value": "NP_004324.2:p.Val600Glu"
                },
                {
                  "syntax": "hgvs.g",
                  "value": "NC_000007.13:g.140453136A>T"
                },
                {
                  "syntax": "hgvs.c",
                  "value": "ENST00000288602.6:c.1799T>A"
                }
              ]
            },
            {
              "name": "CIViC representative coordinate",
              "value": {
                "chromosome": "7",
                "start": 140453136,
                "stop": 140453136,
                "reference_bases": "A",
                "variant_bases": "T",
                "ensembl_version": 75,
                "representative_transcript": "ENST00000288602.6",
                "reference_build": "GRCh37",
                "type": "coordinates"
              }
            }
          ],
          "mappings": [
            {
              "coding": {
                "system": "https://www.ncbi.nlm.nih.gov/snp/",
                "code": "rs113488022"
              },
              "relation": "relatedMatch"
            }
          ]
        }

I don't love that expressions is in extensions (ga4gh/cat-vrs#124 (comment))

[jsstevenson]

@ahwagner made an issue to update the docs on this matter (summary: expressions should refer to the specific VRS object, not a consequence of from another molecule type): #658