An ancestry‐based approach for detecting interactions

Abstract

Background: Epistasis and gene-environment interactions are known to contribute

Modeling epistasis in mice and yeast using the proportion of two or more distinct genetic backgrounds: evidence for “polygenic epistasis”

“…The results of our study suggest that heterogenous SNP effects due to differing ancestries is pervasive in mouse populations, especially in diverse populations such as the HMDP. This observation matches prior observation of significant epistatic interactions in inbred strains of mice as well as examples in human studies 5 . Further analyses of these heterogenous-effect SNPs may reveal novel epistatic interactions which drive phenotypic expression, and suggests that careful attention to genetic ancestry should be considered when studying the role of an individual polymorphism on a phenotype.…”

“…A fundamental assumption of many of these studies is that an allele will have a similar effect in each member of the population, that is, that epistatic and other higher-order interactions across the genome can largely be ignored [1][2][3][4] . While we have previously observed only modest evidence of ancestry specific genetic effects in humans 5 , model organisms are often further diverged than human populations. For example, we observed radically different phenotypic consequences of null alleles of Tcf7l2 and Cacna1c when expressed on different inbred strain backgrounds 6 .…”

“…In this study, we approached epistasis in a different way by examining two distinct mouse cohorts and considering the effects of interactions between individual polymorphisms and global ancestry (Ɵ), which we defined as the percentage of the genome inherited from one of the parental lines of each cohort. We utilized a model we recently developed 5 which explored these interactions in human populations and extended this idea to model organisms by incorporating a linear mixed model capable of accounting for the highly structured relatedness of recombinant inbred (RI) lines 10 . Although our initial model was designed to identify both genetic background and environmental effects, the controlled environment in which mouse cohorts live is intended to minimize sources of environmental variance.…”

Uncovering the role of admixture in disease and drug response: Association of hepatocyte gene expression and DNA methylation with African Ancestry in African Americans

“…Several studies have shown that the first several principal components of methylation data can capture population structure in cohorts composed of European and African individuals (Barfield et al, 2014). A recent paper by Park DS et al found that genetic ancestry can be used as a proxy, not only for unknown covariates contributing to epistatic and gene-environment interactions from gene expression data, but also from DNA methylation data (D. S. Park et al, 2018). Indeed, approximately 75% of differential methylation between global populations were attributable to genetic ancestry in addition to ethnically distinct DMRs, which are known to correlate with environmental factors, such as maternal smoking during pregnancy (Galanter et al, 2017).…”

Mixed-model admixture mapping identifies smoking-dependent loci of lung function in African Americans

“…We performed a univariate Wald test with one degree of freedom to derive the p-value for interaction effect between genotype and exposure. By including a local ancestry term when testing for the interaction effect of genotype, we accounted for possible different LD patterns for European and African ancestral backgrounds 49 . Such conditional analysis can reduce power but assures that the interaction effect of genotype is driven by a biological mechanism rather than a better SNP tagging in a particular ancestral population.…”