90% credible sets for SNP-, mark-, and path-mapping.

<p>We compare <i>pathfinder</i> to the technique of independently fine-mapping the two levels of data, with respect to (A) the calibration of their credible sets and (B) the size of their credible sets. In (A), we compare the proportion of causal variants that were captured in the 90% credible sets using <i>pathfinder</i> vs. independent fine-mapping against the expected proportion (represented by the dotted line). In (B), we display the corresponding sizes of these credible sets.