Bias of genetic trend of genomic predictions based on both real and simulated dairy cattle data

Abstract Text:

This study investigated the phenomenon of bias in the trend of genomic predictions and attempted to find the reason and solution for this bias. The data used in this study include Danish Jersey data and simulation data. In Jersey data, the bias was reduced when cows were included in the reference population. In simulated data, there was no bias when the test animals were unselected cows. When the G matrix was derived from genotypes of causal genes, the bias was reduced. The results suggest that the main reasons for causing the bias of the prediction trends are the selection of bulls and bull dams as well as the inaccurate relationship matrix. The possible strategies to eliminate the bias could be to use cow reference and improve genomic relationship matrix.

Keywords:

prediction bias

reference population

genomic relationship matrix