Quantifying portable genetic effects and improving cross-ancestry genetic prediction with GWAS summary statistics

Abstract

Polygenic risk scores (PRS) calculated from genome-wide association studies (GWAS) of Europeans are known to have substantially reduced predictive accuracy in non-European populations, limiting its clinical utility and raising concerns about health disparities across ancestral populations. Here, we introduce a novel statistical framework named X-Wing to improve predictive performance in ancestrally diverse populations. X-Wing quantifies local genetic correlations for complex traits between populations, employs a novel annotation-dependent estimation procedure to amplify correlated genetic effects between populations, and combines multiple population-specific PRS into a unified score with GWAS summary statistics alone as input. Through extensive benchmarking, we demonstrate that X-Wing pinpoints portable genetic effects and substantially improves PRS performance in non-European populations, showing 18.7%-122.1% gain in predictive R2 compared to state-of-the-art methods based on GWAS summary statistics. Overall, X-Wing addresses critical limitations in existing approaches and may have broad applications in cross-population polygenic prediction.

Optimizing and benchmarking polygenic risk scores with GWAS summary statistics

“…First, we allow our method to incorporate LD information. We note that this extension is similar to some recent work built on our initial PUMAS paper 37,39 . Second, we allow the method to partition full GWAS summary statistics into more than two datasets for various analytical purposes.…”

“…It is an important future direction to systematically optimize and benchmark PRS for diverse ancestral populations which would require incorporation of multiple sets of ancestry-specific GWAS and LD references. Although we did not explore this topic in this paper, our recent work introduced parallel ideas to tackle the challenges in multi-ancestry genetic risk prediction 39 . Second, analyses in this study were limited to GWAS summary statistics computed from independent samples.…”

“…We base our statistical framework on PUMAS, a method we recently introduced to perform Monte Carlo cross-validation (MCCV) using GWAS summary statistics 35 . We have shown that PUMAS can effectively fine-tune PRS models with clumped SNPs 36 and the approach has since been adopted in other applications [37][38][39] . Here, we first demonstrate that PUMAS can fine-tune and benchmark state-of-the-art PRS models without SNP pruning.…”

Global biobank analyses provide lessons for developing polygenic risk scores across diverse cohorts

“…We also found that prediction performance can be improved when using ancestry-matched tuning cohorts for PRS construction to fine-tune hyper- parameters and avoid overfitting, such as P+T and the PRS-CS grid models explored in this study. While other studies have also explored options such as pseudo-validation when no additional tuning cohort is available 40,41…”

“…While other studies have also explored options such as pseudo-validation when no additional tuning cohort is available 40,41 Third, the practical considerations for target populations involved in PRS analyses are quite consistent between using homogenous GWAS and multi-ancestry GWAS. In this study, we used biobanks with various ancestry compositions and recruitment strategies as the target cohorts 19 .…”