However, before concluding that molecular genetic analysis trumps the phenotypic separation of unipolar from bipolar, two points should be born in mind. GWAS results show that the majority of heritability can be assigned to many loci of small effects. How many that might be depends on the unknown contribution of rarer variants of large effect, but we can provide a rough estimate by assuming that depression is a quantitative trait, in which MD is one extreme (following the same reasoning for the power estimates for a successful MD GWAS [Yang et al., 2010b]). From the distribution of effect sizes of other quantitative traits, we can estimate the number of loci required to explain the heritability of MD. Assuming an exponential distribution (Goldstein, 2009), about 2,500 loci are required to explain half the heritability. This estimate is conservative, since the distribution of variants more closely follows a Weibull distribution than an exponential (Park et al., 2010). In short, the number of variants required to explain MD heritability implies that about one in five genes expressed in the brain are likely to be involved.
