For each imputation scenario based on the different software programs and reference panels, we calculated three imputation performance metrics, which captured different features of imputation accuracy and quality. First, after masking 2% of the genotyped SNPs, we calculated the concordance rate as the percentage of genotype calls for which the true genotype matches the most likely discrete imputed genotype. This concordance rate calculation, based on discrete imputed SNP genotypes, has been used often as a measure of imputation accuracy [6], [15], [25]–[29]. Second, using the same masked SNPs, we calculated the imputation quality score (IQS) as previously described by Lin et al. to adjust the concordance rate for chance agreement between imputed and true genotypes [30]. More specifically, the IQS, which is partly motivated by Cohen’s kappa statistic to quantify interrater agreement [31], controls for allele frequencies by taking the observed agreement between imputed and true genotypes (i.e., concordance rate) and subtracting out chance agreement, based on the sum of products of marginal frequencies that would occur if genotypes were called at random [30]. Therefore, the IQS metric is particularly
