RCV Statements¶

Overview¶

The rcv bundle section contains one record per condition-level aggregate classification. Each record is a Statement that aggregates individual SCV submissions for the same variant and condition into a hierarchical summary — combining classifications across submission levels to produce a single condition-specific result.

RCV statements differ from VCV statements in one important way: condition-scoped aggregation — each RCV is scoped to a specific condition (identified by trait_set_id), whereas VCV statements aggregate across all conditions for a variant.

RCV statements use #/ references for propositions, contributing SCVs, and lower-level RCV groupings. They are produced by the RCV Procedures.

Record Structure¶

Each record is a Statement with the following top-level fields:

Field	Type	Description
`id`	string	RCV layer ID — e.g., `RCV001781420.1-G-PATH-CP`
`type`	string	Always `Statement`
`proposition`	string	`#/proposition/{id}` reference to the aggregate proposition
`classification`	object	MappableConcept — the aggregate classification label. See Classification
`strength`	object	MappableConcept — the aggregate evidence strength
`direction`	string	`supports`, `disputes`, or `neutral` — derived from the aggregate classification
`confidence`	object	Concept struct with `conceptType: "Confidence"` and `name` (the submission level label, e.g., `criteria provided`, `expert panel`)
`extensions`	array of Extension	ClinVar-specific aggregate metadata (0..*). See Extensions
`hasEvidenceLines`	array	`#/evidenceLine/` references to contributing and non-contributing evidence. See Evidence Lines

The classification, strength, direction, and confidence fields follow the same structure and rules as VCV Statements.

Classification¶

RCV statements use the same MappableConcept classification as VCV:

{
  "conceptType": "Classification",
  "name": "Pathogenic/Likely pathogenic",
  "extensions": [
    {"name": "conflictingExplanation", "value": "Pathogenic(3); Likely pathogenic(2)"}
  ]
}

For somatic clinical impact (SCI) propositions, the classification label may include the tier, assertion type, and clinical significance.

Proposition¶

RCV propositions are stored in the proposition bundle section. Each RCV statement references its proposition via #/proposition/{id}.

A resolved RCV proposition contains:

Field	Type	Description
`type`	string	Proposition type matching the underlying SCVs (e.g., `VariantPathogenicityProposition`)
`id`	string	Proposition ID (e.g., `RCV001781420-G-PATH-CP`)
`subjectVariant`	string	`#/variation/clinvar:{id}` reference
`predicate`	string	Predicate matching the underlying SCVs (e.g., `isCausalFor`)
`objectCondition`	string	`#/condition/clinvar.trait:{id}` or `#/conditionSet/clinvar.traitset:{id}` reference — the specific condition for this RCV

The key difference from VCV propositions: the objectCondition references the specific condition for this RCV accession, sourced from the representative SCV's condition mapping. VCV propositions may carry multiple condition references from contributing SCVs.

Evidence Lines¶

Evidence lines work identically to VCV. The hasEvidenceLines field contains #/evidenceLine/ JSON pointer references to evidence line records in the evidenceLine bundle section:

{
  "hasEvidenceLines": [
    "#/evidenceLine/RCV001781420.1-G-PATH-CP.contributing",
    "#/evidenceLine/RCV001781420.1-G-PATH-CP.non-contributing"
  ]
}

At the Classification layer, evidence items within the referenced records point to SCVs via #/scv/. At the Priority and Aggregate Contribution layers, evidence items reference lower-level RCV groupings via #/rcv/.

See VCV Evidence Lines for the full field reference.

Extensions¶

Extensions carry aggregate metadata not part of the GA4GH VA-Spec statement model. Each extension follows the GA4GH Extension structure: { "name": "<name>", "value": <value> }. Extensions appear at two structural levels — on the top-level Statement and on the classification object.

RCV extensions follow the same patterns as VCV extensions. See RCV Extensions (Pipeline) for details on how these extensions are built during pipeline processing.

Statement Extensions¶

Extension Name	Value Type	Description
`clinvarReviewStatus`	`string`	The aggregate review status derived from the submission level and aggregation outcome. Same value set as VCV Statement Extensions. Always present.

Classification Extensions¶

Extensions on the classification MappableConcept within the Statement.

Extension Name	Value Type	Description
`conflictingExplanation`	`string`	A formatted breakdown of conflicting classification counts (e.g., `Pathogenic(3); Likely pathogenic(2)`). Present only when the classification is conflicting.

SCI Classification Label Format¶

For somatic clinical impact (SCI) propositions, the RCV aggregate classification label includes the tier, assertion type, and clinical significance:

<tier_label> - <assertion_type> - <clinical_significance> (<scv_count>)

Examples: Tier I - Strong - diagnostic - supports diagnosis (1), Tier II - Potential - prognostic - poor outcome (1)

Layer Hierarchy¶

RCV statements use the same multi-layer aggregation hierarchy as VCV, with condition (trait_set_id) as an additional grouping dimension at every layer.

Layer	ID Format	Aggregates By	Scope
Classification	`{RCV}.{ver}-{group}-{PROP}-{level}[-{TIER}]`	Classification label within submission level	All
Priority	`{RCV}.{ver}-{group}-{PROP}-{level}`	Tier priority within submission level	Somatic only
Aggregate Contribution	`{RCV}.{ver}-{group}-{PROP}`	Submission level (winner-takes-all)	All

Submission level ranking is PG > EP > CP > NOCP > NOCL > FLAG.

See RCV Procedures for implementation details.

Examples¶

Annotated JSONC examples of RCV statement records are available in the repository:

RCV statement examples