An imputation strategy which results in an alternative parameterization of the single stage genomic evaluation

  • J M Hickey
  • G Gorjanc
  • B P Kinghorn
  • B Tier
  • J H.J van der Werf
  • M A Cleveland

Abstract

A method is presented for imputing genotypes in pedigreed populations based on long-range phasing, haplotype libraries, recombination modelling and segregation analysis. In two very different data sets, one a pig data set comprising animals from a single line and the other a multiple breed cattle data set, imputation accuracy was high and was always higher than that of Impute2 a widely used alternative. Accuracy was highest for animals which had both parents genotyped at high-density, however some animals with neither parent genotyped at high-density also had high imputation accuracy. The method imputes genotypes or sum of the allele probabilities for all animals in the pedigree and thus facilitates single stage genomic evaluations combining all available pedigree, genomic, and phenotypic information in a single step. This was explored using both simulated and real data with favourable results