A polygenic score is only as good as its ancestry
A score that means average in one population can mean high in another.
Polygenic risk scores (PRS) summarise the combined effect of thousands of variants into a single number. They are one of the most promising tools in preventive genomics — and one of the easiest to get dangerously wrong.
The portability problem
Most PRS were derived from genome-wide association studies in predominantly European-ancestry cohorts. Applied to a person of a different ancestry, the same raw score can land at a completely different percentile of true risk. A score that means “average” in one population can mean “high” in another. Reporting it without calibration is not just imprecise — it is misleading.
Calibrating explicitly
Innovare computes polygenic scores against reference panels and infers ancestry explicitly using an approach (PCA against reference panels). The inferred ancestry then drives calibration, so a percentile means the same thing regardless of background — with particular attention to the Iberian and admixed Latino populations our clients serve, which are underrepresented in most published scores.
Reproducible, layered, reviewable
Every score is traceable: the reference used, the ancestry inferred, the calibration applied. Coverage gates decide whether a score is reportable at all. The goal is a PRS a laboratory can stand behind — not a black-box number that happens to look precise.
See it run on your data
Request a demo of the full chain — from raw reads to a reviewable, signable report.