Sponsored by:
Estimating genetic correlation of direct and indirect effect using UK Biobank sibling dataJie Song, Clara Zou, Yiliang Zhang, Qiongshi LuUniversity of Wisconsin-MadisonA person's phenotype is affected by its genotype via both direct and indirect pathway, that is, the direct path via its own genotype, and indirect path via the environment fostered by its parents or relatives. Thus the traditional linear mixed model that only contains person's own genotype will produce effect sizes that is mixture of direct and indirect effect, then genetic correlation estimates for a trait pairs will also be a mixture of direct effect correlation and indirect effect correlation. Accurate dissection of genetic correlation is helpful for us to better interpreting genetic relatedness among different phenotypes. We propose a framework for sibling data structure of two traits with direct/indirect effect partition, and apply the method of moment method to UK Biobank (UKB) sibling data for detecting genetic correlation of direct/indirect effects of several traits including education attainment(EA), height, body mass index(BMI), income, and overall health. We found EA and Income has moderate indirect effect correlation, whose effect size is comparable with direct effect correlation. Jackknife is applied to produce standard error estimation, but the result is relatively conservative. To improve power we perform meta-analysis with WLS and Add Health cohort. This model can also be utilized for single trait analysis. We observed significant sibling environment correlation for all traits. Finally, instead of using individual level data, we try to only use summary statistics for genetic correlation dissection estimation.