Final Statements (Step 7)¶

Overview¶

Step 7 of gkm_scv_statement_proc performs the final assembly of complete VA-Spec Statement records into gkm_dict_scv. This step joins all previously built components -- SCV records, propositions, conditions, citations, and assertion methods -- into the output structure consumed by VCV procedures and the JSON export pipeline.

CTEs¶

The step uses several CTEs to prepare intermediate data before the final SELECT.

scv_condition_name¶

Derives a single human-readable condition name string from gkm_scv_condition_sets for use in the classification description extension:

• Named submitted condition -- uses the submitted condition's name directly
• ConditionSet with 2+ conditions -- produces "N conditions" (e.g., "3 conditions")
• Single submitted condition with no name (e.g. a coded-only submission) -- uses the name of the ClinVar condition it maps to in gkm_dict_condition, the name ClinVar shows (from 1.1.0, #162)
• Otherwise -- falls back to "unspecified condition"

temp_scv_citation_docs¶

Builds one VA-Spec Document per ClinVar citation. The same rules apply to the submission's citations (reportedIn) and to the assertion method's citations (specifiedBy.reportedIn). A ClinVar citation can list several ids for one publication; ACMG 2015, for example, lists a DOI, a PubMed id and a PMC id. They become one Document:

urls only ever holds http(s) URLs, never a bare id or CURIE. Citations with only a URL become {urls: [url]}.

ClinVar's original values are never changed silently. Only invalid values are repaired, and the Document's extensions keep the originals:

The repairs:

• DOI with a prefix or stray characters: whitespace, a leading https://doi.org/, http://dx.doi.org/, doi.org/, doi:, doi/ or org/, and a trailing . are removed. The result must match the DOI pattern, or the id is unusable (invalidClinvarCitation).
• DOI source with a PubMed or PMC id: a DOI-sourced id that is all digits becomes pmid; one shaped PMC<digits> becomes a PMC URL. Both keep originalClinvarCitationSource.

Output: temp_scv_citation_docs — internal temporary table consumed by temp_scv_citations and temp_scv_method.

temp_scv_citations¶

Collects the submission's citation Documents into the reportedIn array, ordered by PubMed id and then by content. Identical Documents collapse to one.

temp_scv_method¶

Builds the assertion method record for the specifiedBy field: the method type, the method name and, when the assertion method cites a publication, its Document. Method.reportedIn holds one Document. When ClinVar lists more than one citation for the method, the first, in ClinVar's order, goes on the method. The rest go in the Method's own extension additionalClinvarMethodCitations, whose value is an array of Documents built by the same rules:

{
  "type": "Method",
  "methodType": "research",
  "name": "Submitter's publication",
  "reportedIn": { "type": "Document", "pmid": "29116469", "urls": ["https://pubmed.ncbi.nlm.nih.gov/29116469"] },
  "extensions": [
    {
      "name": "additionalClinvarMethodCitations",
      "value": [ { "type": "Document", "pmid": "29907814", "urls": ["https://pubmed.ncbi.nlm.nih.gov/29907814"] } ]
    }
  ]
}

Output Record Structure¶

Extensions¶

The extensions array on each statement includes:

• clinvarScvId -- the SCV accession identifier
• clinvarScvVersion -- the SCV version number
• clinvarScvReviewStatus -- the review status assigned by ClinVar
• submittedScvClassification -- the original submitted classification label
• submittedScvLocalKey -- the submitter's local key for the assertion
• submissionLevel -- the submission level code and label

Classification Description Extension¶

The classification.extensions array contains a formatted description string assembled from multiple fields:

for <condition_name>
Classification is based on the <submission_level_label> submission
<evaluated_date> by <submitter_name>

This provides a human-readable summary of the classification context, including the condition, submission level, evaluation date, and submitting organization.

Somatic Evidence Lines¶

For somatic clinical impact assertions, the hasEvidenceLines field contains evidence line records that wrap the target proposition from Step 6. Each evidence line includes the target proposition (with its somatic-specific predicate and therapy/condition references) and the outcome classification.

Output¶

gkm_dict_scv -- one row per SCV with the complete VA-Spec Statement record. Pipeline table

Consumed by gkm_vcv_proc and gkm_vcv_statement_proc for VCV aggregation, and assembled directly into the published bundle by the export pipeline.

Dependencies¶

• Upstream Steps: Step 1 (temp_gkm_scv), Step 5 (temp_gkm_scv_proposition), Step 6 (temp_gkm_scv_target_proposition)
• Upstream Procedures: gkm_scv_condition_proc (provides gkm_scv_condition_sets)
• Downstream Consumers: gkm_vcv_proc, gkm_vcv_statement_proc, change-log / delta build (gkm_change_log), export pipeline