We have developed a novel Bayesian statistical procedure to assess whether two association signals are colocalised. Our method is best suited for associations detected by GWAS, which are likely to reflect common, imputable, variations with small effects, or a rare variants with large effect sizes. Our aim differs from a typical fine-mapping exercise in the sense that we are not interested in knowing which variant is likely to be causal but only whether a shared causal variant is plausible. The strength of this approach lies in its speed and analytical forms, combined with the fact that it can use single variant p-values when only these are available.
