Fig. 4: Imputation accuracy in each allele frequency bin.
From: Accurate imputation of human leukocyte antigens with CookHLA
We used cross-validation by splitting the T1DGC panel into reference (N = 1000) and the target sample (N = 4225). For HIBAG-prefit, we used the European prefit model. The accuracy refers to sensitivity to correctly impute an allele given a specific allele.
