Home > Archive > Volume 65, No. 1 > This paper
Predictive Accuracy of Polygenic Scores from European GWAS among Chinese Provinces
Davide Piffer and Emil O. W. Kirkegaard
Published: 2024/09/01
Abstract
This study assesses the predictive accuracy of polygenic scores (PGS) from a variety of genome-wide association studies (GWAS) in a dataset of 28 Chinese provinces, with a focus on educational attainment (EA) and height. European-derived EA PGSs showed stronger correlations with average IQs of Chinese provinces (r = .52) than did East Asian-derived PGSs (r = .21), likely reflecting larger sample sizes. Both height PGSs derived from mixed ancestry and from European samples were positively correlated with average height (r = .71 and .68). Additionally, both genetic and phenotypic height showed positive correlations with latitude (r = .72 and .77, respectively), corroborating Bergmann’s rule and supporting the observation that northern Chinese tend to be taller. The PGS derived from within-family GWAS of height showed stronger correlation (r = .82) with phenotypic height and latitude than the between-family derived PGS. Whereas IQ PGS was positively correlated to latitude (r = .42), this was not the case for EA PGS. Negative correlations were also observed between schizophrenia PGS and both EA and IQ PGS (r = −.44 and −.50). Results from multiple regression analyses indicated that both genetic factors and environmental conditions (measured by HDI and infant mortality) influenced stature, with genetic factors having a stronger effect (∼ 0.8) compared to environmental conditions (∼ 0.22 to 0.45). Mediation analysis showed that the genetic effects of EA and IQ PGS on IQ are partially mediated through their effects on HDI and infant mortality. Applying Jensen’s method, we found that polygenic scores with a stronger genetic signal of selection exhibited slightly higher predictive accuracy (r = .25−0.27, p < .01). Keywords: Polygenic score, Education, Height, Intelligence, China
Download PDF