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

Abstract

Background: Commonly used genetic models of phenotype assume that alleles have similar effects in all individuals. Mounting evidence suggests that higher-order interactions, which may vary between populations, profoundly affect phenotypic variation. The effect of genetic and environmental perturbations may therefore differ across genetic backgrounds. In this manuscript, we develop a statistical test that determines whether the effect of a mutation on a complex phenotype changes as a function of ancestral background. Results:We apply our test to two large cohorts of mice and observe 49 significant gene by ancestry interaction associations across 14 phenotypes as well as over 1,400 Bonferroni-corrected gene by ancestry interaction associations in gene expression data. We also observe evidence of rapid selection pressure on individual polymorphisms within one of the cohorts. Conclusions:Unlike our prior work in human populations, we observe widespread evidence of ancestryspecific SNP effects, perhaps reflecting the greater divergence present in crosses among laboratory mice.

Factorizing polygenic epistasis improves prediction and uncovers biological pathways in complex traits

“…More recent polygenic epistasis models make simplifying assumptions to improve power and interpretation. For example, autosome-sex interactions represent a particular form of epistasis that strongly affects many complex traits [53][54][55][56], and interactions between a locus (or region) with a polygenic background can identify epistasis hubs [57][58][59][60][61][62]. EFA instead assumes that polygenic epistasis is driven by interactions among a few latent polygenic pathways.…”

A model and test for coordinated polygenic epistasis in complex traits

“…Genome-wide searches for interacting single-nucleotide polymorphism (SNP) pairs are computationally and statistically onerous, despite some success (21). Additionally, recent and classical methods for epistasis increase power by aggregating interactions across the whole genome (3,(22)(23)(24)(25)(26)(27)(28), and their results further support the potential importance of epistasis in complex traits. Interestingly, all of these approaches assume that interaction effect sizes and directions are independent of main effects or, in other words, that epistasis is uncoordinated.…”

Coordinated Interaction: A model and test for globally signed epistasis in complex traits

“…For example, epistasis significantly impacts evolutionary models, including response to selection or changing environment (Barton, 2017;Corbett-Detig et al, 2013). Epistasis also fundamentally shapes genetic architecture, as the direction of an allele's effect can change based on genetic background Rau et al, 2019;Sittig et al, 2016). Epistatic interactions are pervasive in model systems, including in model organisms (Bloom et al, 2015;Brem et al, 2005;Corbett-Detig et al, 2013;Forsberg et al, 2017;Huang et al, 2012;Mackay, 2014) and in recent mammalian gene-level interaction screens (Dixit et al, 2016;Horlbeck et al, 2018;Norman et al, 2019).…”

“…Genome-wide searches for interacting SNP pairs are computationally and statistically onerous, despite some success (Marchini et al, 2005). Additionally, recent methods for epistasis increase power by aggregating interactions across the whole genome (Crawford et al, 2017;Jannink, 2007;Rau et al, 2019), and their results further support the potential importance of epistasis in complex traits. Interestingly, all of these approaches assume that epistasis is unstructured in that interaction effect sizes and directions are independent of main effects.…”