Background: Linear type traits, which reflect the muscular characteristics of an animal, could provide insight into how, in some cases, morphologically very different animals can yield the same carcass weight. Such variability may contribute to differences in the overall value of the carcass since primal cuts vary greatly in price; such variability may also hinder successful genome-based association studies. Therefore, the objective of our study was to identify genomic regions that are associated with five muscularity linear type traits and to determine if these significant regions are common across five different breeds. Analyses were carried out using linear mixed models on imputed whole-genome sequence data in each of the five breeds, separately. Then, the results of the within-breed analyses were used to conduct an across-breed meta-analysis per trait. Results: We identified many quantitative trait loci (QTL) that are located across the whole genome and associated with each trait in each breed. The only commonality among the breeds and traits was a large-effect pleiotropic QTL on BTA2 that contained the MSTN gene, which was associated with all traits in the Charolais and Limousin breeds. Other plausible candidate genes were identified for muscularity traits including PDE1A , PPP1R1C and multiple collagen and HOXD genes. In addition, associated (gene ontology) GO terms and KEGG pathways tended to differ between breeds and between traits especially in the numerically smaller populations of Angus, Hereford, and Simmental breeds. Most of the SNPs that were associated with any of the traits were intergenic or intronic SNPs located within regulatory regions of the genome. Conclusions: The commonality between the Charolais and Limousin breeds indicates that the genetic architecture of the muscularity traits may be similar in these breeds due to their similar origins. Conversely, there were vast differences in the QTL associated with muscularity in Angus, Hereford, and Simmental. Knowledge of these differences in genetic architecture between breeds is useful to develop accurate genomic prediction equations that can operate effectively across breeds. Overall, the associated QTL differed according to trait, which suggests that breeding for a morphologically different (e.g. longer and wider versus shorter and smaller) more efficient animal may become possible in the future.

This study investigated the effects of alternative mating programs that incorporate genomic information on expected progeny herd performance and inbreeding, as well as methods to include un-genotyped animals in such mating programs. A total of 54,535 Holstein-Friesian cattle with imputed high-density genotypes (547,650 SNP after edits) were available. First, to quantify the accuracy of imputing un-genotyped animals (often an issue in populations), a sub-population of 729 genotyped animals had their genotypes masked, and their allele dosages were imputed, using linear regression exploiting information on genotyped relatives. The reference population for imputation included all genotyped animals, excluding the 729 selected animals and their sires, dams, and grandsires, and had either (1) their sires' genotypes, (2) their dams' genotypes (3) both their sires' and their dams' genotypes, or (4) both their sires' and maternal grandsires' genotypes introduced into the reference population. The correlations between true genotypes and the imputed allele dosages ranged from 0.58 (sire only) to 0.68 (both sire and dam). A herd of 100 cows was then simulated (1,000 replicates) from the sub-population of 729 imputed animals. The top 10 bulls from the genotyped population, based on their total genetic merit index (TMI) were selected to be used as sires. Three mating allotment methods were investigated: (1) random mating, (2) sequential mating based on maximizing only the expected TMI of the progeny, and (3) linear programming to maximize a generated index constructed to maximize genetic merit and minimize expected progeny inbreeding as well as intra- and inter-progeny variability in genetic merit. Relationships among candidate parents were calculated using either the pedigree relationship matrix or the genomic relationship matrix; the latter was constructed using either the true genotypes of both parents or the true genotypes of the sire plus the imputed allele dosages of the dam. Using the genomic co-ancestry estimates resulted in lower average herd expected genomic inbreeding levels compared with using the pedigree-based co-ancestry estimates. Additionally, if the dams were not genotyped, using their imputed allele dosages also resulted in lower average herd expected inbreeding levels compared with using the pedigree co-ancestry estimates. The inter-progeny coefficient of variation for selected traits, milk and fertility, estimated breeding values were reduced by 12 to 65% using the linear programing method compared with sequential mating.

The proceedings of the 12th World Congress on Genetics Applied to Livestock Production provide you with 816 papers representing the leading research in livestock genetics around the globe. This boo...