We identified enriched gene annotation categories using GeneTrail (http://genetrail.bioinf.uni-sb.de/; 77). We used all GO functional annotations (78) and the KEGG pathways (79,80). When testing for enrichments among the 33 and 200 most differentially expressed gene sets (see Supplementary Material, Tables S4 and S5), we used all 11 282 expressed autosomal genes as a reference. In order to determine enrichments for the 117 lymphocyte count candidate gene set (see Supplementary Material, Table S7), we used the 10 239 genes included in the integrated analysis as a reference. P-values were determined based on the hypergeometric distribution, and the FDR was estimated using the method of Benjamini and Hochberg (81).
