Genome-wide association studies (GWAS) represent a powerful approach to the identification of genetic variants involved in common human diseases[1]. GWAS use commercial SNP microarrays to genotype large numbers of genetic markers. However, SNP microarrays currently can only genotype up to one million of the 9–10 million common SNPs in the assembled human genome [2]. In addition, for a typical case-control design, several thousand cases and several thousand controls may be needed for adequate power to detect associations[3]. With little cost, imputation can boost power both by increasing SNP coverage and by combining samples from similar studies. Based on haplotypes from the International HapMap project[4], imputation infers untyped variants from known genotypes. The inference uses one of several model-based methods, and the resulting imputed SNPs can be tested for association with a phenotype [5]. The power of this method has been demonstrated in the literature where several groups have found novel causal genes [6], [7], [8], [9].
