Our analysis and meta-analysis represent the largest and most powerful investigation into the genetic architecture of MDD to date. The lack of clear-cut evidence for association allows us to exclude genetic main effects that are common (MAF>0.1) with moderate to strong effect sizes (GRR>1.4). As noted earlier, the MDD2000+ sample alone possessed >90% power to detect the median MAF (0.36) and GRR (1.33) of GWAS associations published to date with P<5 × 10−8. Moreover, as we would expect to detect >20% of variants with OR 1.2 and MAF 0.2–0.7 (Supplementary File 1), and having detected none that are genome-wide significant, we can probably draw an even more stringent boundary on the risk allele architecture under a main effect model. These results are informative for researchers planning association studies of MDD.
