What people mean by “the BRCA gene”
There is no single BRCA gene. BRCA1 and BRCA2 are two large DNA-repair genes. When they work, they help fix breaks in DNA. When a variant knocks that function out, lifetime risk of some cancers can rise. That is why families talk about “the BRCA gene” as if it were one switch.
It is not one switch. There are thousands of BRCA1 variants in ClinVar. Some are pathogenic. Many are benign or uncertain. Founder variants exist in some populations. A genealogy chip looks at a handful of sites and misses most of the gene.
A clinical BRCA test sequences the genes and looks for copy-number changes. A 23andMe file is not that test. “Not in my raw data” is not a negative BRCA result. It is a sampling gap.
- BRCA1 and BRCA2: DNA-repair genes, not one gene
- Clinical test: sequencing plus copy-number analysis
- Consumer chip: a thin sample of common SNPs
BRCA1 variant lookup is a catalog or a site, not a test
BRCA1 variant lookup in genome.sh is identifier work. genome query BRCA1 lists known variants for the symbol. That is a catalog from public databases, not a report about a person.
A specific site is an rsID or an HGVS string. Only a local VCF, a clinical assay, or a called chip probe tells you whether that site was seen in a sample. The HTTP API will not take the file.
BRCA1 sits on chromosome 17. NCBI Gene 672 is the locus record. Pathogenic variation includes many frameshifts and splice sites, not only famous founder SNPs. A clean consumer file does not rule those out.
- Gene query: catalog of known variants for the symbol BRCA1
- Site query: one rsID or HGVS string
- Personal result: only from a called local file or a clinical assay
Gene query versus founder rsID versus HGVS
genome query BRCA1 --format json is how you inspect the catalog. Limit it in jq. Do not paste the whole catalog into a PDF as if those sites were tested.
Founder variants that appear in tutorials still need an rsID or HGVS to look up one site. rs80357906 is a ClinVar-heavy BRCA1 example used in genome.sh docs. It is not a complete founder panel.
HGVS is required when you must name a transcript-specific change. BRCA1 has several transcripts. A protein string without an accession is a nickname. See the HGVS guide before you invent a coding HGVS string.
Consumer chips miss most pathogenic BRCA1 variants
Genotyping arrays cover a few hundred thousand sites. BRCA1 pathogenic catalogs include many indels and sites that never made the array. Not in my 23andMe file is not a negative BRCA1 result.
If a founder rsID is on the chip and called, you can look up the public ClinVar record and the genotype letters. If the row is missing, write not called. Missing is not wild type and not BRCA-negative.
Clinical BRCA testing sequences the gene, looks at rearrangements, and is interpreted in a lab. genome.sh will not replace that with a chip grep.
Some consumer products added a handful of BRCA sites in later chip versions. That still is not gene sequencing. Treat every unlisted site as not called, including rearrangements no array can see.
Read ClinVar review status on BRCA1 records
BRCA1 has thousands of ClinVar rows. Pathogenic, likely pathogenic, VUS, likely benign, and benign all appear. Conflicts appear too.
Review status (the stars) is consensus, not severity. A VUS in BRCA1 from a consumer file is not a finding to act on. A pathogenic label is a submitted assertion about a variant, not a diagnosis of a person.
AlphaMissense can score missense substitutions. Many BRCA1 pathogenic variants are not missense. Do not use a missense prior as a BRCA screen.
If you filter JSON to Pathogenic only, keep review status on those rows. A single-submitter BRCA1 assertion is not the same public object as an expert-panel assertion. genome.sh will not hide that difference unless your jq does.
- Keep significance and review status together
- Do not flatten conflicts to a single emoji
- Do not treat chip absence as a negative gene test
Query BRCA1 with the CLI and the open API
Install genome-sh, then lite or a larger tier. Query the gene or a site. JSON is for scripts. Human output is for one terminal.
curl https://api.genome.sh/v1/query/BRCA1?limit=5 is the hosted catalog peek. GET /v1/gene/BRCA1 is the gene endpoint. No API key. No VCF upload.
For a local file, genome annotate sample.vcf.gz --filter clinical --format json. That still only annotates sites the file called.
cargo install genome-sh genome db install lite genome query BRCA1 --format json | jq '.[0:5]' genome query rs80357906 curl -s 'https://api.genome.sh/v1/query/BRCA1?limit=5' | jq .
BRCA2 is a different gene
BRCA2 is a separate symbol with its own ClinVar catalog. Query it separately. The same chip limits apply. The same not-called rule applies.
People say BRCA when they mean both genes and a clinical panel. genome.sh will not merge them into a single risk score. It will look up identifiers you give it.
Point people who want personal risk to genetic counseling and a clinical test. Point scripts at JSON.
genome query BRCA1 --format json | jq 'length' genome query BRCA2 --format json | jq 'length' curl -s https://api.genome.sh/v1/query/rs80357906 | jq .
What to write when a chip site is missing
Write not called. Do not write negative. Do not write wild type. Do not write no BRCA mutation.
If you produce a PDF from a consumer export, state that a genotyping chip is not a BRCA test. List only sites that were called, with ClinVar review status.
genome.sh prints public annotations. It is not a medical device. Clinical decisions need a clinician and an appropriate assay.
Local annotation still is not a BRCA panel
If you have a VCF from a clinical or research pipeline, genome annotate sample.vcf.gz --filter clinical --format json will attach ClinVar to called BRCA1 sites. Called is the important word.
Copy-number variants and large rearrangements are often absent from a small VCF of SNPs. Do not write BRCA-negative because annotate returned no pathogenic SNP.
Keep the file on disk. The HTTP API rejects uploads. Point people who need a personal result to genetic counseling and a clinical test, not to a catalog length from jq.
genome annotate sample.vcf.gz --filter clinical --format json | jq '[.[] | select(.gene=="BRCA1")]' grep -w rs80357906 23andme.txt || echo not called
Questions
How do I do a BRCA1 variant lookup?
genome query BRCA1 --format json for the catalog, or genome query with an rsID or HGVS for one site. curl https://api.genome.sh/v1/query/BRCA1?limit=5 is the hosted form. Neither is a clinical BRCA test.
Can genome.sh tell me if I have a BRCA1 mutation?
It can annotate sites that were called in a local file. It cannot replace a clinical BRCA test with sequencing of the gene.
Will 23andMe find BRCA1 founder variants?
Only if those rsIDs are on the chip and called. Many important BRCA1 variants are not on consumer arrays. Missing is not called, not negative. A later chip version that added a few founder sites still is not a BRCA test.
Should I query BRCA1 or BRCA2?
They are different genes. Query each symbol separately. The same chip limits apply to both.
What rsID should I use for BRCA1?
There is no single BRCA1 rsID. Use a specific site such as rs80357906 when you mean that site. Use the gene query when you want the catalog.
Does a VUS in BRCA1 mean I should act?
Not from a consumer file. Uncertain significance means the evidence was insufficient to classify. That is a question for a clinician and an appropriate assay.
Can I upload a VCF for BRCA1 annotation?
Not to the HTTP API. Use genome annotate file.vcf.gz --filter clinical --format json on disk, or the in-browser importer.
Is BRCA1 variant lookup medical advice?
No. genome.sh reports public ClinVar and related annotations. Personal risk belongs with genetic counseling and a clinical test.