We consider a situation where two traits have been measured in two distinct datasets of unrelated individuals. We assume that samples are drawn from the same ethnic group, i.e. allele frequencies and pattern of linkage disequilibrium (LD) are identical in both populations. For each of the two samples, we consider for each variant a linear trend model between the outcome phenotypes Y and the genotypes X (or a log-odds generalised linear model if one of the two outcome phenotypes Y is binary):We are interested in a situation where single variant association p-values and MAFs, or estimated regression coefficients and their estimated precisions , are available for both datasets at Q variants, typically SNPs but also indels. We make two additional assumptions and discuss later in this paper how these can be relaxed. Firstly, that the causal variant is included in the set of Q variants, either directly typed or well imputed [17]–[19]. Secondly, that at most one association is present for each trait in the genomic region of interest. We are interested in exploring whether the data support a shared
