Finally, Table 6 gives similar calculations for the correlation between predicted and observed quantitative traits with high (h 2 = 0.8) and moderate (h 2 = 0.4) heritability. The prospects here appear more challenging in terms of the sample sizes needed to approach the achievable correlation. For example, height has heritability of about 0.8, and the number of associated variants is known to be at least in the hundreds [7]. In the most optimistic scenario shown, 31,000 subjects would be required to derive a predictor with correlation 0.8 with the true height. In fact these sample sizes are now being approached by collaborative studies, and this result confirms that this is necessary for accurate prediction of quantitative traits in addition to the primary goal of identifying individually associated markers.
