Skip to main content

Table 5 Accuracy comparisons between the multi-step piecemeal and the usual two/three-step imputation

From: Whole genome SNP genotype piecemeal imputation

BaseProgram Imputation a c c 1 a c c 2 a c c 3 a c c π +
Beagle 8 Holstein BTA 27 71.29 % 74.25 %   74.43 % 0.18 %
FImpute 6 K →50 K →777 K 82.80 % 82.74 %   82.92 % 0.18 %
Beagle 367 Simmental BTA 14 66.00 % 65.51 %   66.59 % 1.08 %
FImpute 6 K →50 K →660 K 68.43 % 68.54 %   68.56 % 0.02 %
Beagle 23 Simmental BTA 14 84.91 % 89.88 %   90.17 % 0.29 %
FImpute 50 K →660 K →Sequence 87.95 % 90.47 %   90.50 % 0.03 %
Beagle 23 Simmental BTA 14 81.19 %   83.94 % 86.26 % 2.32 %
FImpute 6 K →50 K →660K →Sequence 82.23 %   84.58 % 84.67 % 0.09 %
  1. Results are on the Holstein datasets for markers on chromosome 27 and for the Simmental datasets for markers on chromosome 14, respectively. 8 Holstein and 367 Simmental genotyped animals are used in the two-step independent testing (6 K →50 K →HD), with results in columns 4, 6 and 7. The piecemeal imputation uses the selected markers and their associated target marker clusters from the training step. Additional 23 Simmental sequenced and genotyped animals are used in the two/three-step imputation to Sequence (50K →660K →Sequence, 6K →50K →660K →Sequence). All one-step imputation accuracies are included in column 3. The last column labelled with + show the improvements, in bold, of the piecemeal imputation over the two- or three-step imputation