Skip to content

Hereditary cancer

Not a medical test — for research and educational use only

Yeliztli analyses consumer genotyping-array data (23andMe / AncestryDNA), which is not a clinical-grade test. Results are not diagnostic, are not clinically validated, and must not be used to make medical decisions. Array data is especially unreliable for rare, disease-causing variants. Always confirm any finding with an accredited clinical laboratory and discuss it with a qualified clinician or genetic counsellor before acting on it. See the Intended use & disclaimers page for the details and the evidence behind this warning.

The cancer module looks for inherited variants linked to hereditary-cancer syndromes, and separately estimates polygenic risk for a few common cancers.

What it looks at

A 28-gene hereditary-cancer panel, including BRCA1, BRCA2, TP53, PALB2, ATM, CHEK2, the Lynch-syndrome mismatch-repair genes (MLH1, MSH2, MSH6, PMS2), APC, MUTYH, PTEN, STK11, CDH1, and others. It also computes polygenic risk scores (PRS) for prostate (XY samples only), colorectal, and melanoma.

What you'll see

  • Monogenic findings — ClinVar Pathogenic/Likely-Pathogenic variants in panel genes, grouped by cancer syndrome, each with its ClinVar accession, review stars, inheritance pattern, and an evidence rating. For recessive genes, you're told whether you're a carrier versus affected. BRCA1/BRCA2 findings cross-reference the carrier module, where they also appear in a reproductive context. ClinVar lower-penetrance/risk-allele findings are stored under a distinct findings category from high-penetrance P/LP cancer-predisposition variants, although the current page displays both categories together.1
  • Polygenic scores — shown as a population percentile, never as a raw score, confidence interval, or absolute risk. This statement is specific to cancer PRS output: PRS percentiles are explicitly research-use-only and are kept separate from the optional monogenic absolute-risk context below.
  • Absolute-risk context (optional) — after explicit opt-in, the Cancer page can show a breast-cancer context card. It compares a general-population female lifetime-risk baseline from NCI SEER with published female-carrier penetrance estimates for BRCA1/BRCA2 Pathogenic/Likely-Pathogenic findings [1]. This is not a PRS-derived estimate and not a personalized clinical risk model.

Breast-cancer PRS is currently unavailable

The source-verified breast-cancer PRS77 [2] is not scored or reported. The bundled GRCh38 allele audit [3] records 39 multiallelic primary loci and 2 additional biallelic loci that are palindromic without trusted effect-allele strand provenance. The current sample schema does not preserve enough allele context to harmonize those loci safely. This is a runtime limitation, not an ancestry-calibration result; the percentile note below applies only to models that are scored. The Cancer page names this model beside the polygenic risk scores and says it is not currently scored or reported, so its absence is not mistaken for a missing result.

Percentiles can be withheld

A polygenic percentile is only shown when the score is properly calibrated for your inferred ancestry. When it isn't, Yeliztli withholds the percentile rather than show a misleading number. A score is also withheld when your array typed fewer than half of its variants; the page still lists that model and shows how many of its variants were typed, so a low-coverage score is not mistaken for a missing one.

Good to know

  • A negative result does not rule out hereditary cancer risk — arrays type only specific variants, not whole genes.
  • Polygenic scores are derived mostly in European-ancestry cohorts; an ancestry-mismatch warning is shown when your inferred ancestry differs, and the per-component evidence is capped low.
  • MUTYH single-copy carriers are framed as carriers, not as affected.

How to read the absolute-risk context

The absolute-risk context is hidden until you opt in because it displays sensitive disease-risk numbers. When shown, read it as context for clinical follow-up, not as a diagnosis or a complete personal risk calculation.

  • Population lifetime risk is the general-population baseline used for comparison. In this card it comes from NCI SEER's female breast-cancer statistics; it is not computed from your genotype.
  • Carrier penetrance is the estimated proportion of people with a specific pathogenic genotype who develop the condition over a defined time window.
  • Cumulative risk to age 80 means the estimated probability that a female BRCA1 or BRCA2 carrier is diagnosed with breast cancer by age 80 in the cited carrier-cohort study [1]. It is a group estimate, not your personal probability.
  • 95% CI means the study's 95% confidence interval around the estimate. It describes statistical uncertainty in the published estimate, not a guarantee that an individual's risk falls inside that range.

Sex-specific context matters. For XY/male samples, Yeliztli does not show the female BRCA1/BRCA2 cumulative-risk figures; it shows male-specific framing instead. If biological sex cannot be resolved from array data, sex-specific numeric penetrance is withheld. Prostate PRS is computed only when the sample's biological sex resolves to XY. It is not computed for XX samples, or when sex is ambiguous or cannot be determined, so those samples' results show no prostate score. The Cancer page names the missing score and says why: for an XX sample the model does not apply, while for an ambiguous or undetermined sex the score is withheld. Recording the individual's biological sex and re-running the analysis resolves the second case. The check uses the biological sex as it stands when the page loads, so if the recorded sex changes after the analysis ran, a prostate score computed under the earlier sex is withheld in the same way rather than shown as current. In the other direction, if the sex is recorded as XY after an analysis that skipped prostate, the page says the score has not been computed yet and that re-running the analysis will compute it.

References

[1] Kuchenbaecker KB, Hopper JL, Barnes DR, et al. Risks of breast, ovarian, and contralateral breast cancer for BRCA1 and BRCA2 mutation carriers. JAMA. 2017;317(23):2402-2416. PMID 28632866.

[2] Mavaddat N, et al. Prediction of breast cancer risk based on profiling with common genetic variants. Journal of the National Cancer Institute. 2015;107(5):djv036. PMID:25855707 (accessed 2026-08-03).

[3] Yeliztli cancer PRS panel version 1.2.0, breast-cancer PRS77 model_provenance.current_allele_audit in backend/data/panels/cancer_prs_weights.json (Ensembl Variation GRCh38 audit checked 2026-07-16; DOI:10.1093/database/bay119) (accessed 2026-08-03).

  1. Schmidt RJ, et al. Recommendations for risk allele evidence curation, classification, and reporting from the ClinGen Low Penetrance/Risk Allele Working Group. Genetics in Medicine. 2024;26(3):101036. PMID:38054408; DOI:10.1016/j.gim.2023.101036 (accessed 2026-07-31). ↩