Reduction in the number of SNPs in the 90% Credible Set after incorporating functional annotations.

Shown here are the number of SNPs in the 90% Confidence Set for each of the lipid phenotypes as estimated using PAINTOR. After marginally running PAINTOR on the entire pool of annotations, we selected the top five annotations for each trait and fit full trait-specific models on each of the densely imputed data sets. We compared PAINTOR with or without integration of functional annotation data. The magnitude in the reduction in the size of the confidence set approximately mirrors what we observe in simulations.Reduction in the number of SNPs in the 90% Credible Set after incorporating functional annotations.