The typical methods for filtering poorly-imputed SNPs are using either the variance ratio or the imputed information score combined with minor allele frequency. Imputation quality is especially important in a study that combines genotypes from different platforms. Therefore, we increased our thresholds for variance ratio and the imputed information score in Table 3, and in Supplementary Tables S1 and S2. But these measures were ineffective in this extreme situation. However, IQS may be used as an effective filter to combine data genotyped on different platforms.
