Single marker analyses were performed in SAS42 and followed the original study in assessing the Cochran-Armitage trend test. For GWAS data, the trend test is widely thought to be the most robust and broadly appropriate primary hypothesis test currently available and here it also provides the closest possible test of replication. Previous work demonstrated 1) validity of the trend statistic in the presence of Hardy Weinberg disequilibrium, and 2) asymptotic equivalence of allele and genotype-based trend statistics when Hardy Weinberg equilibrium holds43. We present uncorrected results from one-tailed tests of the three direct replication SNPs (rs1344706 in ZNF804A, rs6490121 in NOS1 and rs9922369 in RGRIP1L). We treat the remaining analyzed SNPs in ZNF804A (10 after one SNP dropped during QC) as independent (based on their selection as tags) and apply Bonferroni correction for 10 tests.
