correlations with the larger set of imputed SNPs and their characteristics change. We have observed this in another study testing differences in imputation performance among African Americans by reference panels and imputation software (Hancock et al. 2012). In that study, the same genotyped SNPs were used for imputation with both HapMap and 1000G reference panels, but the average R2 was reduced somewhat for 1000G due to the greater prevalence of low MAF SNPs in the 1000G panels. The counter point is that the coverage for 1000G was much better. Thus, investigators must balance imputation accuracy and coverage in choosing which approach to take.
