We further tested whether the set of hard-to-impute SNPs compiled from the first group can be used to filter the imputed data in the second group. We applied a similar procedure as in Figure 1. We randomly divided the second group into cases and controls. Cases were genotyped on the Illumina 550 K array and the remaining Illumina 1 M SNPs were imputed. Controls were genotyped on the Illumina 1 M array. Figure 4 shows that the QQ plot can be adjusted to normal by IQS calculated from the first group. This implies that the development of a database of IQS scores for standard imputations would allow researchers to use data genotyped on different platforms and filter out potential false positives.
