Skip to content

Polygenic score (PGS/PRS) vs Genetic risk

Definition

A polygenic score (PGS) or polygenic risk score (PRS) is a *constructed predictor*: a weighted sum of allele dosages from GWAS, usually trained in one sample and evaluated in another. Genetic risk is informal language for how much of a trait’s or disease’s probability or liability is attributable to inherited variation—conceptually broader than any single score, and not tied to one training algorithm.

How they differ

PGS / PRS Genetic risk (informal)
Nature A specific model: weights, SNP set, ancestry handling, clumping and thresholding choices. A conceptual umbrella: heritable contribution vs environment and chance.
Output A continuous score per person (often relative, e.g. SD units). May mean absolute risk, relative risk, heritability, or liability—depends on context.
Validation Measured by AUC, , calibration plots, transfer across cohorts. Requires clear definitions of trait, time horizon, and population when translated to risk.

Rule of thumb: PRS is one operationalization of part of genetic risk for complex traits; “genetic risk” without a definition is ambiguous—ask whether the speaker means the score, SNP heritability, familial risk, or penetrance of a Mendelian variant.

References

  • Choi SW, Mak TS, O’Reilly PF. (2020). Tutorial: a guide to performing polygenic risk score analyses. Nat Protoc.