Many statistical approaches have been introduced for fine-mapping ranging from a simple ranking of marginal association statistics to Bayesian approaches that integrate elaborate priors [5], [8], [9], [10], [11], . Due to the fact that fine-mapping can be casted as a variable selection problem, both LASSO-like procedures that estimate empirical probabilities of inclusion for SNPs based on sub-sampling [13], as well as Bayesian approaches that perform joint multipoint inference to compute posterior inclusion probabilities [14] have been proposed. The inclusion probabilities provided by these methods offer a natural way to prioritize variants in fine-mapping. However, although neither of the two variable selection approaches assume a fixed number of causal variants, they both require individual level data which is often not readily available. Ranking of SNPs for follow-up analysis can also be performed based on correlation-adjusted t-scores that explicitly take into account the correlation structure among variants, thus requiring individual level data [12] as well. Recent works [5], [8], [9] have proposed to estimate posterior probabilities and credible sets for variants to be causal under the simplifying assumption of single causal
