Imputation efficacy has been defined as the proportion of imputable SNPs for which imputation was deemed successful [Nothnagel et al., 2009]. In keeping with this definition, we defined imputation yield as the absolute number of reference SNPs (i.e., SNPs in the reference panel but not in the study sample) for which imputation was deemed successful. Using the software Haploview [Barrett et al., 2005], available at http://www.broad.mit.edu/mpg/haploview/, we estimated coverage of HapMap variation [Barrett and Cardon, 2006] for a given reference panel based on tag sets consisting of the HUFS study sample with and without successfully imputed SNPs. Briefly, coverage was measured by pairwise correlation between a tag SNP and a potentially captured SNP. A potentially captured SNP was considered covered if r2 ≥ 0.8 between itself and any tag SNP. Coverage was reported as the proportion of the set of potentially captured SNPS covered by the set of tag SNPs.
