Genomic selection combines statistical methods with genomic data to predict genetic values for complex traits. The accuracy of prediction of genetic values in selected population has a great effect on the success of this selection method. Accuracy of genomic prediction is highly dependent on the statistical model used to estimate marker effects in reference population. Various factors such as density markers, reference population and genetic architecture affect choosing the best model for analysis. The aim of this study was to evaluate methods for genomic selection in the Iranian Holstein population. Four Bayesian linear regression models, Bayes-A, Bayes-B, Bayes- , Bayesian-LASSO and a genomic linear method (GBLUP) were compared using stochastic simulation across three types of reference population and a range of numbers of quantitative trait loci (QTL) in two classes of marker density. Bayesian methods had a higher accuracy than GBLUP especially when the number of loci was low and data from other population were used. Construction of reference population by bulls and cows together with data from other populations and using Bayes B is suggested for genomic selection in Iranian Holstein.
Rights and permissions | |
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License. |