A more practical way of evaluating this approach is to look at the false positive rate. Specifically, although no SNPs are associated with case/control status based on the true genotypes, there were 4016 imputed SNPs that reach genome-wide significance (p<5×10−8). The IQS filter >0.9 eliminated all the false positive SNPs, but the imputation accuracy filter >0.99 still retained 759 false-positive SNPs. Based on these results, IQS is better for discriminating between well-imputed SNPs and poorly-imputed SNPs.
