The Accuracy of Genotype Imputation in Selected South African Sheep Breeds from Australian Reference Panels

Abstract

The cost of genotyping is becoming increasingly affordable but remains an influential factor for determining the SNP-density at which genotyping can proceed. Compared to Australian breeding programs, the South African wool sheep industry represents parallel objectives within similar environments but presently lacks the necessary infrastructure to exploit modern technologies such as genomic selection. The aim of the study was to determine the feasibility of across country imputation as an alternative to high density genotyping on a local basis. Following imputation from a 15k to 50k density, mean accuracy levels of 0.87 and 0.85 were observed in the Merino and Dohne Merino breeds, while the highest levels of accuracy of 0.88 and 0.90 was observed in the Dorper and White Dorper breeds, respectively. The extent of genetic relationships was considered amongst the key factors that limit the ability to impute at an accuracy above 90%, but the observed results suggest that across country imputation could remain useful. Imputation from reference panels genotyped at densities higher than 50k and research into across country prediction is recommended.