Association analysis was conducted on 2431 MDD cases and 3673 screened controls using allelic dosages for imputed SNPs passing QC. The test of association was logistic regression, including the first three ancestry principal components (PC) and analysis set as covariates. The INFO score (ratio of observed to expected dosage variance), a measure of the quality of imputation of SNPs was used to interpret results. A second association analysis was conducted restricting cases to those with recurrent early onset MDD (REO, age of MDD onset <31 years with multiple episodes of MDD). Separate analyses were conducted by sex, as different prevalences of MDD between males and females could imply existence of sex-specific genetic risk variants.
