Advancements in Genetic Prediction
Whole Genome Selection Project Involving 2,000 Industry AI Sires
Abstract:
Whole genome selection (WGS) uses markers spanning the genome to predict genetic merit for economically important traits. WGS may increase the rate of genetic progress through improved accuracy and reduced generation interval, especially for traits that cannot be measured on breeding animals. In contrast to single gene or marker approaches, WGS makes the more realistic assumption that performance is affected by many chromosomal loci (the polygenic model).
WGS research in cattle has just recently been made feasible by the BovineSNP50 BeadChip (50K), a technology that provides genotypes on about 50,000 single nucleotide polymorphism (SNP) markers that are distributed evenly throughout the bovine genome. Genetic predictions from WGS are being used for selection of AI sires in the Holstein breed. However, the structure of the beef seedstock industry is quite different from that of the dairy industry, and it is unknown how well WGS will work in beef cattle.
We believe successful implementation of WGS will require a high degree of organization and broad participation by the beef cattle industry. To be most effective, WGS should be incorporated into the national cattle evaluation (NCE) system.
In order to provide a critical mass of data, genotypes for the 50K chip were obtained on 2,026 influential artificial insemination (AI) sires of 16 breeds: 402 Angus, 317 Hereford, 253 Simmental, 173 Red Angus, 136 Gelbvieh, 131 Limousin, 125 Charolais, 86 Shorthorn, 68 Brangus, 64 Beefmaster, 59 Maine-Anjou, 53 Brahman, 47 Chi-Angus, 43 Santa Gertrudis, 42 Salers and 27 Braunvieh. Of these sires, 551 had been sampled for the Germplasm Evaluation (GPE) project at USMARC, and the remaining 1,475 were selected by the respective breed associations, which provided semen for DNA. These sires were generally influential in their breeds and still viable candidates for selection. Many of them have high-accuracy EPDs. The process of WGS requires a training data set, a population of animals with 50K genotypes and the appropriate phenotypes.
At USMARC, the primary training data is in Cycle VII of the GPE project, representing the Angus, Charolais, Gelbvieh, Hereford, Limousin, Red Angus and Simmental breeds as either progeny (F1; 590 steers) or grandprogeny (F1 x F1 = F12; 1,306 steers and 707 females) of about 22 sires per breed. Phenotypes are available for growth traits, carcass and meat quality traits on the steers, individual feed intake and various measures of feed efficiency on the F12 steers and females, several years of cow productivity data on the F12 females, and lifetime productivity and mature cow maintenance on the F1 females (for which 50K genotypes can either be inferred or were collected).
Single SNP associations were conducted and generated many more significant associations than would be expected due to chance. Various statistical methods for developing effective prediction equations for economically important traits from 50K genotypes are being explored. It is not clear whether prediction equations will need to be breed-specific or whether common prediction equations across all breeds will be sufficient. It will be necessary to validate prediction equations on independent data. When prediction equations have been validated sufficiently, molecular breeding values (MBVs) derived from the 50K genotypes of the 2,000 AI bulls will be provided to the breed associations.
Editor’s Note: The above material is provided by and posted with permission of the Beef Improvement Federation. Please direct reprint requests to BIF via the “Contact BIF” page at www.beefimprovement.org.
