Sample Size for Successful Genome-Wide Association Study of Major Depressive Disorder

Major depressive disorder (MDD) is a complex, heritable psychiatric disorder. Advanced statistical genetics for genome-wide association studies (GWASs) have suggested that the heritability of MDD is largely explained by common single nucleotide polymorphisms (SNPs). However, until recently, there has been little success in identifying MDD-associated SNPs. Here, based on an empirical Bayes estimation of a semi-parametric hierarchical mixture model using summary statistics from GWASs, we show that MDD has a distinctive polygenic architecture consisting of a relatively small number of risk variants (~17%), e.g., compared to schizophrenia (~42%). In addition, these risk variants were estimated to have very small effects (genotypic odds ratio ≤ 1.04 under the additive model). Based on the estimated architecture, the required sample size for detecting significant SNPs in a future GWAS was predicted to be exceptionally large. It is noteworthy that the number of genome-wide significant MDD-associated SNPs would rapidly increase when collecting 50,000 or more MDD-cases (and the same number of controls); it can reach as much as 100 SNPs out of nearly independent (linkage disequilibrium pruned) 100,000 SNPs for ~120,000 MDD-cases.