Using Single-Step Genomic BLUP to Compute Genomic Enhanced Breeding values for Self-Modulation in Working Dogs presented by Molly Riser

The objective of this study was to evaluate the potential of genomic selection to improve behavioral traits in working dogs. Phenotypes of self-modulation on approximately 5000 Labrador Retrievers, a pedigree file containing more than 20,000 records, and the whole genome sequence were available for 463 animals. After selecting variants and removing low call and monomorphic SNPs, a total of 137k SNPs on 457 individuals were available for analysis. The model included the fixed effects of sex, year of birth, contemporary group and breeder ID, and the random additive genetic and residual effects. Heritability was estimated at 0.16, and no differences were observed between variance components with traditional and genomic AIREML. Pedigree and genomic enhanced breeding values were calculated using the complete population, and accuracy was assessed by cross-validation, simulating the selection of puppies. A random sample of 100 individuals with their own phenotypes and progeny had their records removed from the dataset, as well as their progeny’s phenotypes. Breeding values were recalculated in the reduced dataset, and the correlation between the reduced and complete data breeding values was the accuracy estimate. Genomic information provided gains of 10% in accuracy, and reranking of the top animals was observed. Gains in accuracy show that genomic selection can help improve working dogs by more accurately identifying superior animals at younger ages before phenotypes can be collected. Gains are expected to increase as more animals are genotyped, and more phenotypes are collected. Future steps include evaluation of other behavior and health traits.