Key points are not available for this paper at this time.
Abstract Genomic selection has the potential to accelerate the genetic improvement rate for difficult/expensive to measure traits such as feed efficiency and fertility traits in beef cattle. To develop genomic selection tools for Canadian beef cattle, we collected or sourced phenotypic data of feed efficiency along with other traits including carcass merit, female feed intake and fertility related traits, and their genotypes from both research and commercial herds. The phenotypic data were consolidated and recorded for contemporary groups, leading to the development of data sets including 11,292 beef cattle with residual feed intake, dry matter intake, average daily gain, metabolic body weight, 7,299 to 8,081 cattle with carcass merit traits, and 1,802 to 2,792 cows with feed intake and fertility traits after quality controls. Genotype data of the animals from various single nucleotide polymorphisms (SNP) panels were merged to the same allele format of 50K, and then imputed to 770K SNPs, and eventually to whole genome sequence variants. The refined Canadian beef cattle data sets have not only allowed us to investigate genetic architectures of the beef performance traits but also enabled the development of genomic selection tools including molecular breeding values and multiple trait selection indexes with a moderate to moderately high accuracy for the traits through a cross-validation of within reference data set. The genomic selection tools have been successfully deployed to over 10,415 breeding candidates submitted by Canadian beef producers through multiple demonstration projects. The average accuracy of the molecular breeding values and multiple trait selection index of the commercial beef cattle ranged from 0.38 to 0.49 for feed efficiency and carcass merit traits, and from 0.26 to 0.38 for female feed intake and fertility traits. Current research is focusing on calibrating the genomic selection tools with larger data sets and with more advanced genomic prediction methods.
Li et al. (Sun,) studied this question.