The COGA and SAGE samples used the same genotyping platform: Illumina Human 1M BeadChips (both datasets have 1,069,796 SNPs). Results from the two GWA analyses were meta-analyzed by combining the separate results of COGA and SAGE samples (regression coefficient and standard error) into one meta-analysis of overall effects. For meta-analysis of two datasets, the basic meta-analysis function in PLINK was applied. The between-study heterogeneity was tested by the χ2-based Cochrane’s Q statistic. Fixed-effect model was used due to the homogeneity Q statistic and its associated p-value was reported to provide support for this decision. Fixed-effect meta-analysis p-values and fixed-effect ORs were estimated.
