What ClinVar clinical significance actually records
ClinVar clinical significance is a public assertion about a variant, not a statement about a person. A lab or expert group submits a relationship between a change and a condition. NCBI aggregates those submissions onto a variation record.
The aggregate word is what most CLIs print first. It is a summary of submissions, not a new classification computed by genome.sh. Dates, transcripts, and evidence cutoffs differ across submitters, so the summary can move when a snapshot updates.
Absence of a ClinVar row is the usual state of the genome. Most sites have no clinical assertion at all. Missing from ClinVar is not evidence of benignity, and present in ClinVar is not a personal result.
- Assertion: submitted relationship to a condition or drug response
- Review status: how that assertion was reached (the stars)
- Aggregate: NCBI’s combination of submissions on one variation id
The five ACMG/AMP words and the extra buckets
Pathogenic, likely pathogenic, uncertain significance (VUS), likely benign, and benign follow ACMG/AMP language. Those five words are the ones people screenshot. They are not the whole controlled vocabulary.
ClinVar also stores drug response, risk factor, association, protective, affects, and other labels that are not Mendelian pathogenicity. A risk-allele record is a different kind of statement from a pathogenic loss-of-function in a recessive gene.
Conflicting interpretations of pathogenicity is its own aggregate state. Flattening it to Pathogenic or to VUS in a script hides the disagreement. Keep the conflict flag next to whatever word you display.
- Pathogenic / likely pathogenic: submitted disease-causing in the stated context
- VUS: evidence insufficient to classify benign or pathogenic
- Likely benign / benign: submitted not disease-causing for the stated condition
- Other: drug response, risk factor, association, protective, affects
Review status is consensus, not severity
ClinVar gold stars describe review status. They do not grade how dangerous a variant is. A four-star practice guideline and a one-star single submitter can both say Pathogenic. The word is the same, but the public weight is not.
Criteria provided, multiple submitters, no conflicts is stronger than a single submitter without assertion criteria. Expert panel and practice guideline sit above both. No assertion provided is a record that is barely a classification.
genome.sh prints significance and review text when the local ClinVar row has them. If you keep only the first word in JSON, you threw away the part that tells you whether to trust the word.
Conflicts, conditions, and zygosity in the public record
Two labs can submit opposite labels on the same variation id. The aggregate then shows a conflict. That is common for missense in large genes and for older submissions that never used ACMG/AMP criteria.
Conditions matter. A variant can be pathogenic for a recessive condition and still sit at appreciable frequency. Heterozygous carriers are not the affected homozygotes the assertion describes. ClinVar does not know which genotype a person has.
Penetrance is not a star rating. Low-penetrance risk alleles and high-penetrance Mendelian variants can share a vocabulary if you only print one word. Read the condition list and the review text before you promote a row into a report.
Query significance from the genome CLI
Install the crate genome-sh. The binary is genome. lite is the ClinVar-oriented database tier. standard and full still include ClinVar and add gnomAD.
Query by rsID when you have one, by gene when you want the catalog, by coordinates when the site has no rs number. Human output is for one terminal. JSON is for filters. compact is the short form.
rs334 is a clean test because the sickle allele has a strong public ClinVar record. BRCA1 as a gene query is a catalog, not a personal report. Open the ClinVar VCV page when you need the full submission table.
cargo install genome-sh genome db install lite genome query rs334 genome query rs334 --format json | jq . genome query BRCA1 --format json | jq '.[0]'
Filter JSON without dropping review status
Scripts that keep only significance == Pathogenic will drop conflicts, drug-response rows, and review text. That is how a pipeline invents certainty. Filter in jq, but keep the review field in the output you store.
The HTTP API at https://api.genome.sh/v1/query/{id} returns the same public identifier lookup without a key. It will not accept a VCF. For a file, use genome annotate file.vcf.gz --filter clinical --format json on disk.
Refresh the local snapshot when you need a newer ClinVar release. A local index is not live NCBI. Cite the snapshot date from genome db stats or GET /v1/sources.
genome query rs334 --format json | jq '{significance, review_status}'
genome annotate sample.vcf.gz --filter clinical --format json
curl -s https://api.genome.sh/v1/query/rs334 | jq .What a missing ClinVar row, a VUS, and a chip gap mean
No ClinVar record means no submitted assertion in the snapshot, not a clean bill of health. Novel VCF alleles often have no rsID and no ClinVar row. Query those as chr:pos:ref:alt.
A VUS is not a finding to act on from a consumer file. It means the evidence was insufficient to classify. Do not treat it as pathogenic-lite in a PDF.
A genotyping chip is not a clinical exome. Most pathogenic ClinVar variants are not on consumer arrays. If the rsID is absent from the export, write not called. Missing is not wild type.
What genome.sh will not call a diagnosis
genome.sh prints public annotations. It does not apply ACMG/AMP criteria to your genotypes. It does not replace a classified report from a clinical lab.
AlphaMissense is a missense predictor sitting next to ClinVar, not a substitute for review status. Frequency from gnomAD is evidence, not a second clinical significance column.
Clinical decisions need a clinician and an appropriate assay. Use the CLI to find the public record fast. Use ClinVar’s own page when the submission table matters.
Questions
What does pathogenic mean in ClinVar?
Labs submitted that the variant can cause a condition in the stated zygosity and clinical context. It is not a personal diagnosis and it is not a prediction that a person will develop disease.
What is a VUS in ClinVar?
Uncertain significance: the evidence is insufficient to classify benign or pathogenic. Do not treat a VUS as a finding to act on from a consumer chip file.
Do ClinVar gold stars mean the variant is more severe?
No. Stars describe review status and consensus, not severity. A one-star pathogenic and a four-star pathogenic share a word and not a review process.
Why do two labs disagree on ClinVar clinical significance?
Different evidence cutoffs, dates, transcripts, and whether ACMG/AMP criteria were used. genome.sh shows the public aggregate, not a new classification.
How do I query ClinVar clinical significance from the command line?
cargo install genome-sh, genome db install lite, then genome query rs334 --format json. Keep significance and review status together when you filter.
Does absence from ClinVar mean a variant is benign?
No. Most genomic sites have no ClinVar record. Absence is not evidence of benignity.
Can I trust Pathogenic from a 23andMe file?
You can look up the public record for rsIDs the chip called. A consumer chip is not a clinical test, and a missing marker is not called, not wild type.
Is ClinVar clinical significance medical advice?
No. genome.sh prints public annotations. Clinical care needs a clinician and the right assay.