Abstract
Social genetic effects (SGE, also called indirect genetic effects) are associations between genotypes of one individual and phenotype of another. SGE can arise when two individuals interact and heritable traits of one individual influence the phenotype of the other. To better understand the architecture of SGE, we re-analysed an existing dataset comprising 170 behavioural, physiological and morphological phenotypes measured in outbred laboratory mice. For all phenotypes and in order to compare SGE with better-known direct genetic effects (DGE, associations between an individual’s genotypes and their own phenotype), we analysed polygenic models with random terms for SGE and DGE and performed the genome-wide association study of both SGE (sgeGWAS) and DGE (dgeGWAS). Our analyses yielded two main insights: first, SGE and DGE acting on the same phenotype generally arise from partially different loci and/or loci with different effect sizes; secondly, individual SGE associations typically explain less phenotypic variance than DGE associations. Our results shed light on the architecture of SGE and have important implications for the design of future studies. Importantly, we detail and validate methods that can be used for sgeGWAS in outbred populations with any levels of relatedness and group sizes, and provide software to perform these analyses.