The importance of pedigree information and the role of un-genotyped individuals on ‎the neccesity of repeating the genomic selection of binary threshold traits‎

Single-step genomic prediction models became the prevailing tool in genetic evaluations of livestock due to high genomic performance, high genotyping costs for some animals, and utilizing both genotyped and un-genotyped animals. The objective of the current study was to investigate the effect of genetic relationships between the training and validation populations on the performance of single and multi-step GBLUP and BayesB, which in this regard the importance of un-genotyped individuals in generations after genomic selection and their roles on the necessity of reusing genomic selection in binary threshold traits was investigated. For this purpose, genomic populations were simulated to reflect the different number of QTL (33 and 534) with two different distributions of QTL (normal and gamma) on 29 chromosomes. Four statistical models (GBLUP, SS-GBLUP, BayesB and SS-BayesB) were implemented to analyze the simulated data. According to the results, GEBV accuracy was influenced by the relationships between the training and validation populations more significantly for multi-step models than for single-step models. Multi-step and single-step BayesB methods showed an obvious advantage over GBLUP and SSGBLUP when the trait is controlled by fewer QTLs. For the normal QTL effects distribution, BayesB methods (especially multi-step BayesB) obtained the lower accuracy than that of if more QTLs affecting the trait. Generally, since the increasing generation distance is a major challenge in the genomic selection of threshold traits, therefore the use of single-step methods makes possible reuse of genomic selection during the early generations of selection unnecessary. However, using single-step BayesB methods and its advantages in genomic evaluation depends on the identification of trait genetic architecture.