There was some reason to expect that these endophenotypes would provide sufficient power to detect at least a handful of variants in our sample of between 3,088 and 4,469 individuals. More direct measures of biological function, for example, have shown greater power in detecting associated loci in prior work. As we noted in the accompanying method paper (Iacono et al., 2014), genetic association studies of bone mineral density, cholesterol levels, and QT interval all identified genetic associations in samples of fewer than 5,000 individuals (e.g., see Figure 2b in Visscher, Brown, McCarthy, & Yang, 2012). In contrast, investigations of more distal phenotypes, such as height and body mass index, required closer to 20,000 individuals before any significant and replicable loci were discovered.
