Table 3. Accuracy of importance sampling depending on the number of samples.

# Samples	Stdev./P-value
1,000	0.485
10,000	0.172
100,000	0.057
1,000,000	0.018

For each given number of samples used for the importance sampling, we measure the variance of the p-value estimate of BE by running the importance sampling 100 different times for the same dataset. We use the 69 associated loci from the Crohns disease data of Franke et al. [13]. The ratio between the standard deviation of the estimate and the target p-value is reported, which is averaged over 69 loci.