11th Annual IGSS Conference • September 24, 2020

Integrating Genetics and the Social Sciences 2020

Estimating genetic nurture using genome-wide association study summary statistics

Yuchang Wu, Department of Biostatistics and Medical Informatics, University of Wisconsin-Madison

A person's genotype affects her phenotype not only directly but also indirectly, via the rearing environment fostered by the biological parents and other relatives. In other words, parents' genetics could affect their children's phenotype through the environment, a phenomenon known as genetic nurture. Marginal effect estimates in genome-wide association studies (GWAS) are mixtures of the direct (i.e., nature) and indirect genetic effects (i.e., genetic nurture) due to the correlations in genotypes between parents and their offspring. Accurate dissection of direct and indirect effects is critical for the application and interpretation of association mapping and genetic prediction results. However, existing methods to estimate these effects require individual-level genetic and phenotypic data within families, which is difficult to obtain in practice. Here, we propose a novel statistical framework to dissect the direct and indirect genetic effects on complex traits using only GWAS summary statistics as input. We validate our framework with an application to GWAS of birth weight and obtained nearly-identical association results compared to previous studies based on individual-level genetic data. We further showcase the performance of our method through an in-depth application to GWAS of educational attainment (EA). We demonstrate that while polygenic scores (PGSs) based on direct and indirect effects of EA are both predictive of EA phenotype in the population, only direct effect PGS is predictive within families. In addition, direct effect PGS showed substantially more robust predictive performance across population groups in the UK Biobank. Direct and indirect effects of EA showed distinct patterns of genetic correlations with other complex traits. In particular, we identified a significant positive genetic correlation of autism spectrum disorder (ASD) with EA direct effects but not with indirect effects. We further validated this finding on 7,804 proband-parent trios of ASD. PGS of EA direct effects was significantly over-transmitted from parents to ASD probands while PGS based on indirect effects was slightly under-transmitted. Taken together, our study demonstrates that a traditional GWAS approach, in conjunction with phenotypic data on study participants' children, could greatly benefit studies on genetic nurture.

Presenter's website

Poster