However, the largest of these effect sizes is very small. For example, the significant WTCCC association for bipolar disorder was based on an allele frequency of 25% in cases and 28% in controls, although effect sizes for some of the other disorders were greater. The most important implication of the many GWA studies to date is that there are few ‘low-hanging fruit’ for common disorders other than age-related macular degeneration. The largest effect sizes generally involve replicated relative risks of about 1.2 comparing cases and controls; for dimensional analyses, the largest associations explain about 1% of the variance in the population. For example, these are the effect sizes of a common variant in the FTO gene which has been significantly associated with obesity and body mass index in 14 studies (Frayling, Timpson, Weedon, Zeggini, Freathy et al., 2007). In our own GWA studies of reading and general cognitive ability across the normal distribution, our largest effect sizes are less than .5% of the variance (Butcher, Davis, Craig, & Plomin, 2008; Meaburn, Harlaar, Craig, Schalkwyk, & Plomin, 2007).
