The genetics of major depression.

Power to Detect a Locus using GWASSimulated data are plotted to assess power under two different sets of conditions: varying the size of the locus effect (expressed as an odds ratio [OR]) and varying the number of loci used in the GWASs. Results are shown for an Affymetrix 500K array (which yields approximately 400,000 useful genotypes per individual) and simulations using all variants in HapMap (release 2). The simulations are taken from Spencer et al. (2009). The significance level was set at 5 × 10−8. Sample size is shown for the number of cases required; the simulations assume an equal number of controls.

GWAS Sample SizeSample sizes (horizontal axis) required for a GWAS to have 80% probability of detecting the number of loci shown on the vertical axis, at a significance level of 5 × 10−8. Results are shown for four complex traits: two disease and two quantitative phenotypes. The graph assumes that the number of loci detected increases linearly with increasing sample size (data are from Frayling et al., 2007, Lango Allen et al., 2010, Loos et al., 2008, Park et al., 2010, Scuteri et al., 2007, Speliotes et al., 2010, Thorleifsson et al., 2009, Wen et al., 2012, Willer et al., 2009).

The Effect of Disease Prevalence on Sample Size to Detect at Least One Locus for Major DepressionSample sizes (horizontal axis) required for a GWAS to have 80% probability of detecting at least one locus contributing to the risk of major depression, plotted against disease prevalence (vertical axis). In most surveys, major depression has a prevalence of about 10%. The genetic architecture of major depression is assumed to be either similar to height (black continuous line) or weight (red dotted line).