For each individual, imputation accuracy was calculated using the same two measures: (i) the proportion of all of the horses' genotypes that were correctly imputed; and (ii) the correlation between the true and imputed genotypes across all the horses' SNPs when coded as above.
a a genotype is correctly imputed if both alleles are correctly imputed.
More than 99% of the alleles were correctly imputed if more than 300 animals were genotyped at high-density.
Specifically, when 87.8% of missing genotypes were correctly imputed, the accuracy of GEBV was reduced by only 5%.
Using Minimac, up to 97.1% of the alleles were correctly imputed based on 50 pre-selected reference animals.
Around 50% of the SNPs were accurately imputed (meaning an accuracy of 100% over all horses) for each animal and the vast majority of the SNPs were correctly imputed in at least 80% of the horses.
Assessing whether the sequence of offspring is correctly imputed can only be done if a sample of such offspring is actually sequenced, and such data are presently not available in sufficient quantities.
In a scenario with low-density genotypes (14 SNPs/Morgan), these authors observed that 87.8% of missing genotypes were correctly imputed but 95% of the accuracy of GEBV obtained with high-density SNP genotypes (1500 SNPs/Morgan) was achieved.
For each imputed SNP, imputation accuracy was assessed using two measures: (i) the proportion of genotypes for the SNP that were correctly imputed among samples; and (ii) the correlation between the true and imputed genotypic allele counts across all samples (homozygote for allele 1, coded 0; heterozygous, coded 1; homozygous for allele 2, coded 2).
Whereas a 1K SNP panel was generally sufficient to ensure that more than 80% of genotypes were correctly imputed, inference from other studies suggests that a 2 to 3K SNP panel would ensure that the subsequent loss in accuracy for, for example, genomic prediction was minimal [ 10, 44].
