Two separate set-based analyses were conducted to obtain empirical values of association: (1) a selected list derived from the top 5% of the genes in the MGS GWAS dataset and, (2) a more focused examination of candidate genes for AD. For assessing the top genes in the MGS dataset, we listed the top performing SNPs located either within genes or within 100kb of genes. The top 5% of genes were thus selected, however due to the considerable computing effort required to perform the permutations, a further cutoff was applied to this list; set-based gene-wise permutations were limited to genes where the top SNP showed p≤0.0005. Further, to avoid over-inflation of the permuted p-value, the SNP with the smallest p-value was removed from the set-based analysis.
