New PGS framework adds gene–context interactions to improve polygenic score accuracy
A preprint from UK Biobank data introduces PGSC, a method that incorporates sex, age, and medication context into polygenic scores, outperforming standard additive models.
Polygenic scores (PGS) — numerical summaries of an individual's inherited genetic risk for a trait — are constructed from large genome-wide association studies and typically assume that each genetic variant contributes independently and additively to a trait. A preprint posted to bioRxiv on 28 August 2026 challenges that assumption by introducing PGSC, a framework that captures locus-specific gene–context interaction effects (GxC) to refine polygenic predictions.
The authors developed PGSC by analysing UK Biobank data, using sex, age, and statin treatment status as contextual variables. Simulations reported in the preprint show PGSC remains robust under a purely additive model — so it does not degrade performance when interactions are absent — while outperforming standard PGS in realistic scenarios where context-dependent effects are present.
The preprint has not yet been peer-reviewed. If the findings hold up to scrutiny, PGSC could offer a more nuanced approach to polygenic prediction in research contexts, particularly for traits whose genetic architecture is known to vary with biological or pharmacological context. The work is directly relevant to ongoing debates about the assumptions underlying current PGS methodology and their implications for research-grade applications of polygenic scores.
Plain-language version
For patients, families, and general readers. Educational only — not medical advice.
Researchers have described a new way to calculate 'polygenic scores' — a type of genetic calculation that estimates how much a person's inherited DNA contributes to the chance of having a particular trait or condition. Current methods add up the effect of each genetic variant as if they work the same way in everyone. The new method, called PGSC, takes into account that some genetic variants may behave differently depending on a person's sex, age, or whether they take certain medications. The study used a large UK database of genetic and health information. The findings are from a preprint — meaning the research has not yet been checked by independent scientific reviewers — so the results should be treated with caution. This is an educational summary, not medical advice. If anything here raises questions for you, please speak with your GP or a clinical professional.
Sources
Read the original reporting — these are the public sources this summary draws from.
-
Primary sourcePreprint bioRxiv (Cold Spring Harbor Laboratory) · 2026-08-28Locus-specific gene-context interactions improve polygenic prediction