FIGURE 5.

Comparison of imputation accuracy with regard to differently sized and composed reference panels for the DSN target population. The imputation was performed from 50 k to sequence level with Beagle using the 30 DSN reference panel and successively adding 100 HF animals or animals from various breeds (VAR) to the initial 30 DSN reference panel. The imputation accuracy was calculated as relative Manhattan distance. We observed a decrease in imputation accuracy for DSN when animals distant from DSN are added to the reference panel even when the size of the reference panel increased. Interestingly, the drop in imputation accuracy was bigger when adding HF animals than when adding animals from various breeds. The intensity of the color refers to the total number of animals in the reference panel.