were consistent for COGA and SAGE sample separately (p-sa < 0.05 and p-co < 0.05). Applying a permutation procedure for multiple test correction also yielded significant p-values (corrected empirical p-values). 19 of the top 25 SNPs for the SAGE sample had corrected p <0.05 while 10 of the top 25 SNPs for the COGA sample had corrected p <0.05 (Table 2). Among these 25 top risk SNPs, we found two SNPs, i.e., rs1041264 and rs17148121 at RAPGEF1, that had significant cis-acting regulatory effects on the transcript expression of RAPGEF1 in the brain (p=0.029 and 0.018, respectively) (Table 5).
