We estimated the predicted cis-genetic population differences in expression by first computing predicted expression from genotype dosage (0, 1 or 2; see below). With these predicted expression values, we performed differential expression for genetic ancestry using a model analogous to equation (1) (see ‘Global ancestry-associated differential expression analysis’) to obtain predicted genetic ancestry effects. We extracted the observed population differences in expression from the effect sizes estimated after applying mash as described in ‘Global ancestry-associated differential expression analysis’.
