Volume 20, Issue 3 , November 2018, , Pages 375-387
Abstract
The objective of this study was evaluating single-trait and multiple-trait animal models with considering imputation in simulated genomic data to estimate the accuracies of genomic prediction across various genomic scenarios and to detect genotype × environment (G × E) interaction. Genomic ...
Read More
The objective of this study was evaluating single-trait and multiple-trait animal models with considering imputation in simulated genomic data to estimate the accuracies of genomic prediction across various genomic scenarios and to detect genotype × environment (G × E) interaction. Genomic data were simulated to reflect variations in number of QTL (90 and 900) and linkage disequilibrium (LD = low and high) using 50K SNP panel. Afterwards, 90 percent of the markers randomly removed and imputation was performed using FImpute software (version 2.2). The average accuracy of imputation for scenarios with high and low LD was 0.976 and 0.943, respectively. In all scenarios, negligible difference on the genomic accuracies was evident, when original genotypes and imputed genotypes were compared. The genomic accuracy reduced with decreasing the LD, heritability and the genetic correlation among the traits. Comparing to single-trait animal model, using multiple-trait animal model increased genomic accuracy. The level of LD and genetic correlation across environments play important roles providing genotype × environment interaction exists. On the one hand, considering genotype × environment interaction and its effect on increasing of genomic accuracy and imputation of low to high density marker panels (especially high LD scenarios) to reduce of the cost of genomic evaluation on the other hand could be a suitable and practical approach to improve genomic selection.