First Friday talk from Alex Young Disentangling nature and nurture using genomic family data

Genetic associations are often interpreted as the result of direct genetic effects, i.e. causal effects of alleles in an individual on that individual. However, indirect genetic effects (IGEs) from relatives can also lead to genetic association: for example, if alleles in parents affect offspring through the environment, a phenomenon termed ‘genetic nurture’. Previous work has suggested that IGEs from parents explain around 1/3^rd of the association between educational attainment (EA) and polygenic predictors (PGSs) of EA. Confounding due to population stratification and assortative mating can also contribute to genotype-phenotype associations, and recent evidence has shown that principal component adjustment and linear mixed models only imperfectly control for such confounding. A principled solution to the problem of confounding is to use random genotype variation within-families due to Mendelian segregation to estimate genetic effects. We show how to perform family-based heritability estimation and genetic association analyses and compare them to standard approaches using samples of unrelated individuals. A limitation of family-based approaches is that parental genotypes are often missing. We introduce a method for imputing missing parental genotypes based on Mendelian laws, and show how this can be used to increase power for family-based genetic association analyses. We illustrate this with an application to UK Biobank data, finding evidence for substantial confounding affecting summary statistics from unrelated individuals for several phenotypes, including educational attainment and cognitive ability. To investigate the phenomenon of indirect genetic effects from parents to offspring ('genetic nurture'), we extend two generational model family-based analysis (offspring and parents) to a three generational model including grandparents. Just as the two generation model can isolate the component of association due to direct genetic effects, the three generation model can also isolate the component due to IGEs from parents. To overcome the lack of probands with genotyped parents and grandparents, we extend the imputation method to impute PGSs of un-genotyped grandparents. We apply our method to an educational attainment PGS derived from a 3 million person genetic association study, isolating the component of association due to IGEs from parents using data from the Norwegian Mother, Father and Child Cohort Study (MoBa) and Generational Scotland (GS). We examine the degree to which estimates of indirect genetic effects from parents can be explained by assortative mating.

First Friday talk from Alex Young Disentangling nature and nurture using genomic family data

Genetic associations are often interpreted as the result of direct genetic effects, i.e. causal effects of alleles in an individual on that individual. However, indirect genetic effects (IGEs) from relatives can also lead to genetic association: for example, if alleles in parents affect offspring through the environment, a phenomenon termed ‘genetic nurture’. Previous work has suggested that IGEs from parents explain around 1/3^rd of the association between educational attainment (EA) and polygenic predictors (PGSs) of EA. Confounding due to population stratification and assortative mating can also contribute to genotype-phenotype associations, and recent evidence has shown that principal component adjustment and linear mixed models only imperfectly control for such confounding. A principled solution to the problem of confounding is to use random genotype variation within-families due to Mendelian segregation to estimate genetic effects. We show how to perform family-based heritability estimation and genetic association analyses and compare them to standard approaches using samples of unrelated individuals. A limitation of family-based approaches is that parental genotypes are often missing. We introduce a method for imputing missing parental genotypes based on Mendelian laws, and show how this can be used to increase power for family-based genetic association analyses. We illustrate this with an application to UK Biobank data, finding evidence for substantial confounding affecting summary statistics from unrelated individuals for several phenotypes, including educational attainment and cognitive ability. To investigate the phenomenon of indirect genetic effects from parents to offspring ('genetic nurture'), we extend two generational model family-based analysis (offspring and parents) to a three generational model including grandparents. Just as the two generation model can isolate the component of association due to direct genetic effects, the three generation model can also isolate the component due to IGEs from parents. To overcome the lack of probands with genotyped parents and grandparents, we extend the imputation method to impute PGSs of un-genotyped grandparents. We apply our method to an educational attainment PGS derived from a 3 million person genetic association study, isolating the component of association due to IGEs from parents using data from the Norwegian Mother, Father and Child Cohort Study (MoBa) and Generational Scotland (GS). We examine the degree to which estimates of indirect genetic effects from parents can be explained by assortative mating.