This method implements a gene set enrichment analysis [29] (GSEA) without requiring the empirical phenotype-based test procedure to estimate the significance of the categories of genes, enabling its application to family based studies as well as population-based genome-wide association study meta-analyses. In our work. each gene is scored by the most significant p-value among all of the SNPs located within the gene or up to 20 kb from the 5′ and 3′ ends of the genic sequences (Text S2). This value is corrected for confounding effects: the gene size, number of SNPs per kb, number of independent SNPs per kb, number of recombination hotspots per kb, linkage disequilibrium units per kb, and genetic distance measured in centiMorgans per kb. This is done by applying a step-wise multiple linear regression analysis to the normalized (Z-score) p-value of each gene [27]. MAGENTA also corrects for physical proximity along the chromosome retaining only one gene per cluster of genes in a category of genes. The nominal GSEA p-value is calculated by comparing the extent of the “leading edge fraction” (i.e., the subset of
