Our simulations demonstrate that exome-sequencing studies can be affected by population stratification, which may produce spurious associations. We have shown that a simple permutation scheme is sufficient to correct for population stratification when discrete clusters corresponding to genome-wide ancestry are known or can be inferred by applying standard methods to GWAS chip data88, 89, 93. The permutation scheme is appealing in that it generalizes most burden of multiple rare variants tests, however, some tests may also be amenable to the use of PCA covariates in instances in which population structure is best described by continuous clines rather than discrete clusters89.
