The second scenario to consider for use of public controls is in making decisions about GWAS design: given a budget sufficient to ascertain and genotype 4,000 individuals is the most advantageous power achieved by following this baseline study design (2,000 cases and 2,000 controls) or by reducing the number of study controls ascertained and genotyped, relying on public controls instead? Figure 5 presents power by imputation accuracy for the baseline study design (blue diamond and blue dashed line) and several alternatives. To a much greater extent than adding public controls to an existing study (Fig. 4), redirecting resources to increase cases and relying on public controls appears to substantially increase power of a study (Fig. 5). Choosing a study design that targets 3,000 cases, 1,000 study controls, and 2,000 public controls increases power to between 85 and 97 % for average R2 of 0.7–0.9. Pushing this approach further to targeting 4,000 cases and using all public controls makes a more substantial improvement in power [e.g., obtaining 4,000 cases and 8,000 public controls generates greater than 95 % power for average R2 of 0.5 or greater (purple line)].
