Title: Mating programs including genomic relationships and dominance effects Authors
|Sun, Chuanyu -|
|O'Connell, Jeff -|
|Weigel, Kent -|
|Gianola, Daniel -|
Submitted to: Journal of Dairy Science
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: August 16, 2013
Publication Date: December 1, 2013
Citation: Sun, C., Van Raden, P.M., O'Connell, J.R., Weigel, K.A., Gianola, D. 2013. Mating programs including genomic relationships and dominance effects. Journal of Dairy Science. 96(10):8014-8023. Interpretive Summary: Breed associations, artificial-insemination organizations, and on-farm software providers need new computerized mating programs for genomic selection so that genomic inbreeding could be minimized by comparing genotypes of potential mates. Efficient methods for transferring elements of the genomic relationship matrix from a central database to customers were developed. Mating programs including genomic relationships were more effective than those using pedigree relationships, and the expected decrease in inbreeding was worth >$2 million annually for US Holsteins.
Technical Abstract: Breed associations, artificial-insemination organizations, and on-farm software providers need new computerized mating programs for genomic selection so that genomic inbreeding could be minimized by comparing genotypes of potential mates. Efficient methods for transferring elements of the genomic relationship matrix from a central database to customers were developed. Because relationships among >230,000 genotyped Holsteins were difficult to store and transfer, relationships were computed between only requested bulls and cows via a web query or for all genotyped cows with only marketed bulls (e.g., >160,000 cows and >1,500 bulls for Holsteins). For both Jerseys and Holsteins, selection and mating programs were tested using the top 50 marketed bulls for genomic and traditional lifetime net merit as well as 50 randomly selected bulls. The 500 youngest genotyped cows in the largest herd were assigned mates of the same breed with limits of 10 cows per bull and 1 bull per cow (only 79 cows and 8 bulls for Brown Swiss). Linear programming to maximize expected progeny values, a method based on sequential selection of least-related mates, and random mating were compared using a genomic or pedigree relationship matrix for mate inbreeding and additive only or additive plus dominance effects for mate genetic merit. Dominance effects for 45,187 markers and variance were estimated for management group deviation for milk yield. Use of genomic instead of pedigree relationships greatly improved expected progeny value and decreased expected progeny inbreeding. Dominance variance was 4.1 and 3.7% of phenotypic variance for Holsteins and Jerseys, respectively. Including dominance in a mating program increased expected progeny value for milk yield by 86 kg for Holsteins and 52 kg for Jerseys with linear programming, genomic relationships, and the top 50 bulls for genomic PTA milk. Linear programming increased expected progeny value and decreased expected progeny inbreeding more than with the sequential method and much more than with random mating for Brown Swiss, Holsteins, and Jerseys. Based on lifetime net merit, the best combination was selection on genomic merit and mate assignment by genomic relationship. The economic value of using genomic instead of pedigree relationships was >$2 million per year for Holsteins when applied to all genotyped cows and assuming that each cow provided 1 replacement every 2.5 yr. That economic value will grow as more cows are genotyped.