Quantifying the susceptibility of polygenic scores to ancestry stratification
Published in bioRxiv, 2025
Principal component analysis is the standard correction for population stratification in genome-wide association studies, and polygenic scores are built on the resulting summary statistics. When that correction is imperfect — as it is in large, ancestry-diverse biobanks and meta-analyses — the residual structure does not simply add noise. It propagates into the scores and into every test built on them.
This work quantifies that susceptibility, characterizing how much residual ancestry stratification survives standard correction and how strongly it biases downstream inference, including tests for selection on polygenic traits.
Recommended citation: Blanc, J., Mawass, W., and Berg, J. J. (2025). "Quantifying the susceptibility of polygenic scores to ancestry stratification." bioRxiv, 2025.12.04.692430.