Why this might matter to you
Polygenic risk scores (PRS) are the genetics story the consumer market is most excited about right now, and the excitement is running ahead of what the science supports for any one person. The pitch is seductive: add up many small genetic effects across the genome and get a single number for your risk of heart disease, or Alzheimer's, or diabetes. The honest version is more useful than the pitch, and it is the version Varia publishes. Varia explains polygenic scores and shows you, from your own file, why a consumer-grade file cannot produce a valid one. It never prints a risk number of its own.
What a polygenic score actually is
A PRS combines genotype information at hundreds to millions of common variants into one population-level estimate of disease risk. Each variant contributes a small effect; the score sums those effects using weights derived from large genome-wide association studies. The key words are population-level and estimate. A PRS describes patterns across research cohorts. It is not a diagnosis, and it is not a substitute for the cholesterol panels, blood pressure readings, and family history that clinicians already use.
Where the evidence is strongest, and where it is honest
Coronary artery disease is the best-validated application that exists. The American Heart Association's 2022 scientific statement concluded that a CAD polygenic score performs similarly to conventional risk factors and adds prognostic information in middle-aged adults PMID 35862132. The strongest real-world signal is about treatment: in trial reanalyses, statin benefit tracked with genetic risk, with relative risk reductions of 13, 29, and 48 percent across low, intermediate, and high genetic-risk tertiles PMID 25748612, and a 44 versus 24 percent reduction in high-genetic-risk versus other participants in a primary-prevention trial PMID 28223407.
That said, the European Society of Cardiology's 2025 clinical consensus is explicit that guidelines do not yet recommend routine PRS use in clinical practice (ESC 2025, Eur Heart J, DOI 10.1093/eurheartj/ehae649). The value concentrates in borderline or intermediate clinical risk, where genetic information might shift preventive planning. In people already high-risk by standard factors or already on intensive treatment, the added value is marginal.
Alzheimer's is the field's other headline. Here the honest framing matters even more, because APOE epsilon-4 is by far the largest common-variant effect and the marketing tends to imply a polygenic score replaces it. It does not. Recent benchmarking found the best prediction came from two predictors together, APOE plus the remaining polygenic score, at a modest AUC of 0.72 to 0.76 PMID 39762974. A polygenic score does add information even within epsilon-4 carriers, including epsilon-4/epsilon-4 homozygotes, where higher polygenic burden separated cases from cognitively healthy controls at phenotypic extremes PMID 34386572. The effect is real, modest, and population-level. It is not a personal prediction.
The caveat the pitch leaves out
Effect sizes trained in one ancestry group transfer poorly to others, and most polygenic scores were built almost entirely on European-ancestry data. Clinical use of current scores can widen rather than narrow health disparities (Martin et al., Nat Genet 2019; DOI 10.1038/s41588-019-0379-x). This is a property of the score's training data, not of any reader. Varia performs zero inference about your ancestry; it describes the training-data composition of the score being discussed.
What Varia does, and does not, do
Varia's polygenic-scores surface is editorial. It explains what validated scores exist, what the authorities say, and where a clinically validated result comes from. From an uploaded file, after unlock, Varia can report a factual coverage statement: how many of a published score's constituent variants are present in your file. That number is never a risk score. A valid clinical score additionally requires the full variant set, imputation, calibration to your ancestry, and CLIA-lab validation, none of which Varia performs. The point of showing coverage is to make the gap visible, not to suggest Varia is one step from closing it.
Read the full editorial coverage, including the coronary artery disease entry, at Polygenic Risk Scores. Varia describes what the published evidence reports. It does not tell you your risk, and it does not sell you a number.