First, we performed single-GWAS data analysis using genes preferentially expressed in the central nervous system (CNS) [13], [34] as the annotation data. Specifically, we generated the annotation vector as follows: The entries in corresponding to SNPs within 50-kb of the genes from the CNS set were set to be 1. Among all the SNPs, 21.9% were thus annotated to be CNS related. The analysis results of these five psychiatric disorders are given in Table 1. The estimated fold enrichment of the CNS set was 1.749 (s.e. 0.447), 1.261 (s.e. 0.055), 1.467 (s.e. 0.033), 1.177 (s.e. 0.058) and 1.391 (s.e. 0.022) for ADHD, ASD, BPD, MDD and SCZ, respectively. PGC investigators also evaluated enrichment of the CNS gene set by variance component estimation using linear mixed models (LMM) [12], suggesting about 1.6 and 1.5 fold enrichments in BPD and SCZ, respectively. Two explanations may contribute to these minor differences in fold enrichment estimation between GPA and LMM: First, GPA only used summary statistics while LMM used both phenotype and genotype data. Second, GPA and LMM employed different mathematical definitions of fold
