For simulations, each replicate of data is limited to 100 cases, 100 controls, and 100 samples from each of the external cohorts. This is to minimize the calculation time for IBS relationships, which is computationally intensive when performed in thousands of replicates. Although we recognize that this sample size is unrealistic for GWA studies, it is sufficient to assess false-positive error rates and compare the relative power of the three association tests.
