Research area
Variants
Namespace
clinvar
Listed tools
6
Upstream source
ClinVar

Research tasks

  • Resolve a variant to a ClinVar record
  • Compare classifications, supporting submissions and review status
  • Preserve variant and condition-specific accessions

Start with tool discovery

The catalog lists clinvar_get_variant as a representative tool. After connecting your client, inspect the tools and input requirements returned by the service before submitting a query.

Identifiers to prepare

  • VCV accession
  • VariationID
  • dbSNP rsID
  • HGVS expression
  • AlleleID

All available tools (6)

The router surfaces these tools namespaced under clinvar_*:

get_variantResolve a single ClinVar variant by VCV accession, dbSNP rsID, HGVS expression, ClinVar AlleleID, or VariationID. A clean transcript-qualified HGVS resolves even without the (GENE) qualifier (e.g. NM_033380.3:c.1871G>A). Returns the normalized classification, review status, and 0-4 star rating plus a recommended_citation. Use this when you already have a variant identifier; if it fails to resolve, fall back to search_variants.

Resolve a single ClinVar variant by VCV accession, dbSNP rsID, HGVS expression, ClinVar AlleleID, or VariationID. A clean transcript-qualified HGVS resolves even without the (GENE) qualifier (e.g. NM_033380.3:c.1871G>A). Returns the normalized classification, review status, and 0-4 star rating plus a recommended_citation. Use this when you already have a variant identifier; if it fails to resolve, fall back to search_variants.

Input parameters

  • id_type(string) — How to interpret `identifier`. 'auto' (default) detects the shape from the value; the explicit types force one lookup and reject a value of the wrong shape.
  • identifier(string) [required] — A single ClinVar variant identifier. Accepts a VCV accession (VCV000007105), a dbSNP rsID (rs334), an HGVS expression (NM_000059.3:c.1234A>G — the (GENE) qualifier is optional), a ClinVar AlleleID, or a VariationID. Use id_type to force one interpretation; the default 'auto' detects the shape.
  • request_id(any) — Opaque id echoed back in _meta.request_id to correlate a response with server logs. Omit and the server mints one.
  • response_mode(string) — Payload verbosity, cheapest first: minimal = ids + classification + stars; compact (default) adds name, review status, traits; standard adds coordinates, RCVs, consequence; full adds all fields. Start compact, widen if needed.
get_variantsResolve MANY ClinVar variants in ONE call — the batch form of get_variant. Pass a list of identifiers (VCV / rsID / HGVS / AlleleID / VariationID, mixable); prefer this over looping get_variant when you have several. Each result row echoes its identifier and a found flag (misses are explicit, never dropped); requested / found_count / truncated summarize the batch. response_mode trims payload size and, in minimal/compact, hoists the citation to _meta.citation_template.

Resolve MANY ClinVar variants in ONE call — the batch form of get_variant. Pass a list of identifiers (VCV / rsID / HGVS / AlleleID / VariationID, mixable); prefer this over looping get_variant when you have several. Each result row echoes its identifier and a found flag (misses are explicit, never dropped); requested / found_count / truncated summarize the batch. response_mode trims payload size and, in minimal/compact, hoists the citation to _meta.citation_template.

Input parameters

  • id_type(string) — How to interpret `identifier`. 'auto' (default) detects the shape from the value; the explicit types force one lookup and reject a value of the wrong shape.
  • identifiers(array) [required] — A LIST of ClinVar variant identifiers resolved in one call (the batch form of get_variant). Shapes may be mixed (VCV / rsID / HGVS / AlleleID / VariationID). Each WELL-FORMED input yields one result row echoing its identifier and a `found` flag, so an absent record is an explicit miss, never silently dropped; a MALFORMED element fails the whole call with invalid_input naming its position. Capped at 100 per call.
  • request_id(any) — Opaque id echoed back in _meta.request_id to correlate a response with server logs. Omit and the server mints one.
  • response_mode(string) — Payload verbosity, cheapest first: minimal = ids + classification + stars; compact (default) adds name, review status, traits; standard adds coordinates, RCVs, consequence; full adds all fields. Start compact, widen if needed.
search_variantsFree-text search across ClinVar variant names, genes, and traits. Use this to locate a record when you only have a gene symbol plus a change or other loose text, then re-call get_variant with the returned vcv_accession. A gene symbol written in the query is applied as a filter automatically, so results never wander into an unrelated gene, and _meta.search declares what was inferred and whether the match degraded. Returns a paginated results list with total_count / has_more / next_offset; in minimal/compact mode the citation is hoisted once to _meta.citation_template (fill {variation_id}/{vcv_accession} per row).

Free-text search across ClinVar variant names, genes, and traits. Use this to locate a record when you only have a gene symbol plus a change or other loose text, then re-call get_variant with the returned vcv_accession. A gene symbol written in the query is applied as a filter automatically, so results never wander into an unrelated gene, and _meta.search declares what was inferred and whether the match degraded. Returns a paginated results list with total_count / has_more / next_offset; in minimal/compact mode the citation is hoisted once to _meta.citation_template (fill {variation_id}/{vcv_accession} per row).

Input parameters

  • assembly(any) — Filter to variants that have coordinates on this reference assembly. GRCh38 or GRCh37 ('hg38'/'hg19' are accepted and normalized). Omit for either.
  • classification(any) — Filter by normalized ClinVar classification (accepted tokens are the enum). ClinVar's own wording is accepted and normalized ('Likely pathogenic' -> likely_pathogenic). Any other value is REJECTED with invalid_input — an unrecognized classification never silently returns zero rows. Omit for all.
  • count_mode(string) — exact (default) returns total_count (bounded by an internal scan cap, which sets total_count_capped when hit). none skips the count query for lowest latency.
  • gene_symbol(any) — Restrict the search to this HGNC gene symbol. An explicit value always wins; when omitted, a symbol in the query text is applied automatically (reported as _meta.search.gene_symbol_inferred) so loose text narrows within the gene, not into unrelated ones. An unknown symbol is not_found, never an empty page.
  • limit(integer) — Maximum rows to return in this page (1-100).
  • match_mode(string) — Token matching for the query text. auto (default) requires ALL terms and falls back to ANY only when that matches nothing (the fallback is declared in _meta.search). and/or force one mode.
  • min_stars(any) — Keep only variants with at least this many ClinVar review-status gold stars (0-4): 2 = multiple submitters, no conflicts; 3 = expert panel; 4 = practice guideline.
  • offset(integer) — Rows to skip before this page; use the response's next_offset to paginate. Rows beyond total_count return an empty page, never an error.
  • query(string) [required] — Free text matched against variant names, gene symbols and trait names — typically a gene symbol plus a change ('BRCA1 c.5266dup'). Terms are ANDed, degrading to OR (and then to gene-only) when nothing matches all of them; any degradation is declared in _meta.search, never presented as a confident ranking.
  • request_id(any) — Opaque id echoed back in _meta.request_id to correlate a response with server logs. Omit and the server mints one.
  • response_mode(string) — Payload verbosity, cheapest first: minimal = ids + classification + stars; compact (default) adds name, review status, traits; standard adds coordinates, RCVs, consequence; full adds all fields. Start compact, widen if needed.
get_gene_clinvar_summarySummarize a gene's ClinVar variant landscape: counts by clinical significance (pathogenic, likely pathogenic, VUS, benign, conflicting) and by review-status star rating, plus top associated traits. Use this for a gene-level overview before drilling into individual variants with get_variants_by_gene. Returns a recommended_citation.

Summarize a gene's ClinVar variant landscape: counts by clinical significance (pathogenic, likely pathogenic, VUS, benign, conflicting) and by review-status star rating, plus top associated traits. Use this for a gene-level overview before drilling into individual variants with get_variants_by_gene. Returns a recommended_citation.

Input parameters

  • gene_symbol(string) [required] — An HGNC gene symbol (case-insensitive), e.g. BRCA1, TP53, CFTR. A symbol with no ClinVar variants in the local index returns not_found, never an empty success.
  • request_id(any) — Opaque id echoed back in _meta.request_id to correlate a response with server logs. Omit and the server mints one.
  • response_mode(string) — Payload verbosity, cheapest first: minimal = ids + classification + stars; compact (default) adds name, review status, traits; standard adds coordinates, RCVs, consequence; full adds all fields. Start compact, widen if needed.
get_variants_by_geneList the ClinVar variants for a gene as per-variant rows (each with classification and star rating). Use this after get_gene_clinvar_summary to enumerate individual records; narrow with classification / min_stars and paginate with limit / offset (response carries total_count / has_more / next_offset). classification takes the normalized tokens in its enum — ClinVar's own wording ("Likely pathogenic") is accepted and normalized, and any unrecognized value is REJECTED rather than silently returning zero rows. Default sort is stars_desc (highest review confidence first). In minimal/compact mode the citation is hoisted once to _meta.citation_template instead of repeated per row.

List the ClinVar variants for a gene as per-variant rows (each with classification and star rating). Use this after get_gene_clinvar_summary to enumerate individual records; narrow with classification / min_stars and paginate with limit / offset (response carries total_count / has_more / next_offset). classification takes the normalized tokens in its enum — ClinVar's own wording ("Likely pathogenic") is accepted and normalized, and any unrecognized value is REJECTED rather than silently returning zero rows. Default sort is stars_desc (highest review confidence first). In minimal/compact mode the citation is hoisted once to _meta.citation_template instead of repeated per row.

Input parameters

  • classification(any) — Filter by normalized ClinVar classification (accepted tokens are the enum). ClinVar's own wording is accepted and normalized ('Likely pathogenic' -> likely_pathogenic). Any other value is REJECTED with invalid_input — an unrecognized classification never silently returns zero rows. Omit for all.
  • gene_symbol(string) [required] — An HGNC gene symbol (case-insensitive), e.g. BRCA1, TP53, CFTR. A symbol with no ClinVar variants in the local index returns not_found, never an empty success.
  • limit(integer) — Maximum rows to return in this page (1-100).
  • min_stars(any) — Keep only variants with at least this many ClinVar review-status gold stars (0-4): 2 = multiple submitters, no conflicts; 3 = expert panel; 4 = practice guideline.
  • offset(integer) — Rows to skip before this page; use the response's next_offset to paginate. Rows beyond total_count return an empty page, never an error.
  • request_id(any) — Opaque id echoed back in _meta.request_id to correlate a response with server logs. Omit and the server mints one.
  • response_mode(string) — Payload verbosity, cheapest first: minimal = ids + classification + stars; compact (default) adds name, review status, traits; standard adds coordinates, RCVs, consequence; full adds all fields. Start compact, widen if needed.
  • sort(string) — Row order. stars_desc (default) puts the highest review confidence first; stars_asc reverses it; name sorts by variant name; variation_id by ClinVar ID.
get_server_capabilitiesUse this for orientation in a new session: the supported tool surface, response modes, filter vocabularies (the exact accepted values for classification / assembly / sort), recommended workflows, the live ClinVar release date, error codes, and current limitations. Returns ~3kB.

Use this for orientation in a new session: the supported tool surface, response modes, filter vocabularies (the exact accepted values for classification / assembly / sort), recommended workflows, the live ClinVar release date, error codes, and current limitations. Returns ~3kB.

Input parameters

  • request_id(any) — Opaque id echoed back in _meta.request_id to correlate a response with server logs. Omit and the server mints one.

Database & release provenance

Data mode
external-reference
Attestation
attested-reference
Data release tag
bundle-2026-08-31
Release version
v0.5.9
sha256:70e8fc1e8c2edad8c6f9bae91606ec86b3cbf6118f7ac8da1cb356b1a36855d6
ghcr.io/berntpopp/clinvar-link@sha256:0c8c0e64a6d91ca21a8c9ecfdc24381ca9a09539a40cffb0b6f65091a6062ccf

Review the response

Keep the source record link and submitted identifiers with your notes. Record the dataset version and retrieval date when available. This page does not contain a captured ClinVar response.