Effect of Reference Population Size and Imputation Methods on the Accuracy of Imputation in Pure and Mixed Populations

    Imputation as a method of creating low-density chips to high-density chips has been introduced to increase the accuracy of genomic selection in animals. In the current study, to investing imputation accuracy, three populations of mixed (scenario 1), pure (scenario 2) and mixed + pure (scenario 3) were simulated using QMSim. Two methods of imputation including Beagle and Flmpute were used for two types of low and high density chips. Selected reference population sizes for each scenario were 250, 500 and 1000 animals. The results showed that in all considered scenarios, the accuracy of imputation raised by increasing the reference population size from 250 to 500 animals, but decreased by increasing the reference population size from 500 to 1000 animals. The accuracy of imputation using Flmpute method was greater than that of Beagle for the small reference population (250 animals). In all scenarios and reference population sizes of 500 and 1000 animals, increased accuracy in Flmpute method was not significant in compared to the Beagle method. The accuracy of the imputation was higher for scenario 1than for scenario 2. Also the increase in the accuracy of the imputation in Scenario 3 was not significant in compared to Scenario 1. Generally, the results of the current study showed that in developing countries where small genotyped animal populations are available, to increase the accuracy of genomic selection, using Flmpute method and mixed population and increasing the relationship between the reference and the target population could be a suitable approach.