Imputation of untyped SNPs has become an important tool for discovery of new genotype-phenotype associations, generally improving density of coverage and statistical power (Spencer et al. 2009). Extending SNP imputation tools to the context of generating a common set of SNPs for analysis of samples genotyped on different arrays has proved challenging, with substantial biases observed here and in prior studies (Sinnott and Kraft 2012; Uh et al. 2012). However, the promise of accurately conducting this type of imputation is to substantially extend the benefit of publicly sharing GWAS data through repositories like dbGaP. Combining the original phenotypic and genotypic data from several studies into a single population control group and pairing these combined data with cases of the phenotype of interest allow for powerful opportunities to identify new genetic associations. A composite set of public controls can also be used to augment study controls to increase sample size and boost statistical power (Ho and Lange 2010). These study designs extend the scientific and societal benefits from the financial and time investments made by the original studies’ funding agencies and
