The set-based method in PLINK (pngu.mgh.harvard.edu/purcell/plink/) (Purcell et al. 2007) has the advantage of performing association as well as permutation tests. For each set, the program selects independent SNPs according to a user-defined threshold value above a particular r2 (in this case 0.8). For each set of independent SNPs, a permutation procedure can be applied by keeping the linkage disequilibrium (LD) constant and permuting the phenotype a large number of times (10,000 times in this case). The empirical p-value is defined as the number of times the permuted set-statistic (i.e. the mean single-SNP association statistic over the independent SNPs) exceeds an original set statistic +1 (R+1) divided by the number of permutations +1 (N+1=10,001).
