The regression-based method was compared to Sidak's combination test, also known as Sidak's correction [26], and to a modified version of Sidak's correction [40]. The corrected gene p-value, based on Sidak's correction is defined as follows for gene g:(4)where , defined in eq. 1, is the most significant SNP p-value for gene g, and N is the total number of SNPs with available association statistics for gene g. A modification of eq. 4 proposed in [40] uses (N+1)/2 as the exponent to adjust for linkage disequilibrium between regional SNPs, assuming ∼50% of SNPs in a given genomic region are in tight linkage disequilibrium:(5)
