Strategies and utility of imputed SNP genotypes for genomic analysis in dairy cattle

BMC Genomics

Table 4 Accuracy of prediction of direct genomic value (DGV) for 5 dairy traits based on Bovine3K, BovineLD 7K, 50K, imputed up to 50K, imputed up to 800K and imputed 800K-dosage

Genotypes used	Mean allelic error rate (%) of imputation	Milk volume	Fat yield	Protein yield	Survival	Daughter fertility
50K	-	0.540	0.527	0.499	0.224	0.251
Subset Bovine3K	-	0.444	0.464	0.429	0.187	0.200
Subset Bovine LD 7K	-	0.481	0.516	0.443	0.186	0.232
50K-imputed (Test imputedA^A using Bovine3K)	3.86	0.533	0.523	0.496	0.200	0.244
50K-imputed (Test imputed^A with BovineLD)	2.30	0.546	0.531	0.507	0.214	0.246
50K-imputed (Train & Test imputed^B using Bovine3K)	5.52	0.505	0.515	0.481	0.207	0.245
50K-imputed (Train & Test imputed^B using BovineLD)	3.06	0.530	0.524	0.492	0.209	0.248
800K-imputed^C	-	0.558	0.530	0.526	0.232	0.256
800K-dosage^C	-	0.554	0.525	0.520	0.229	0.253

^AGenotypes of 452 young bulls with subset of original SNPs were imputed (using IMPUTE2) up to 50K using 1753 bulls as reference set. Hence for DGV prediction entire test set (452 young bulls) had imputed genotypes and all the training bulls (1753) had actual 50K genotypes.
^BGenotypes of 2055 bulls with subset of original SNPs were imputed (using IMPUTE2) up to 50K using 136 bulls as reference set. Hence for DGV prediction the entire test set (452 young bulls) and 1617 bulls out of the training set of 1753 bulls had imputed genotypes.
^CData on 2205 bulls genotyped for 50K were imputed using IMPUTE2 up to 800K using 845 cows genotyped on 800K as reference.

ISSN: 1471-2164