Investigation of the genomic accuracy of single-trait and multiple-trait animal models in the presence of interaction between genotype and environment

The objective of this study was to evaluate different animal models in different genomic scenarios to estimate the breeding values and to detect genotype × environment (G × E) interaction. Genomic data were simulated to reflect variations in number of QTL (100 and 500) and linkage disequilibrium (low and high) using 10K SNP panel for 30 chromosomes. On this genome, the trait simulated in three different environments with heritabilities 0.10, 0.25 and 0.50, respectively. In the next phase, low (0.47) and high (0.83) genetic correlations were assigned between third environment (h2=0.50) with first (h2=0.10) and second (h2=0.25) environments. The results indicated that the accuracy of genomic prediction increased with increasing the heritability, linkage disequilibrium and the genetic correlation between the traits. Comparing to single trait animal model, multiple trait animal model increased accuracy of genomic prediction. Accuracies of genomic predictions were generally high when the genomic breeding values of third environment were estimated using information of their relatives in the second environment, and the highest accuracy (0.422) obtained from the scenario with high linkage disequilibrium and low QTL. Also, accuracies of genomic prediction increased with increasing percentage of animals from 25 to 75. Generally, the level of LD, type of animals in training set, number of phenotypic records in validation set and genetic correlation across environments play important roles if G × E interaction exists. In conclusion, considering the G × E interaction contributes to understanding variations of quantitative trait and increasing accuracy of genomic prediction.