Friday General Session
Prospects for Genomic Selection in Beef Cattle
Abstract:
Genomic selection requires the construction of a prediction model in which a large number of animals that possess phenotypes, or expected progeny differences (EPDs), are genotyped with a high-density single nucleotide polymorphism (SNP) assay. The genotypes and EPDs (or phenotypes) are used to construct a model to predict genetic merit (molecular EPDs) based solely on an animal’s genotypes.
The accuracy of the produced molecular EPDs is largely determined by the extent of linkage disequilibrium (LD; the correlation between alleles at two loci on a chromosome), the distribution of polymorphisms with respect to allele frequencies and the number of individuals with EPDs that are available for genotyping. Simulation studies have suggested that from 30,000 to 300,000 SNPs are required to implement genomic selection in cattle; however, these studies have not appropriately modeled the extent of LD within the genome or the distribution of variation within the genome.
While it is now known that LD extends for at least 50 kb in Bos taurus cattle (r2 ≈ 0.3), we do not yet have a clear understanding about the number of genes that affect individual traits or the distribution of these genes according to their gene frequency.
This is of great importance because SNP chips such as the Illumina BovineSNP50 assay are designed to include evenly spaced SNPs that have high allele frequencies, and these SNP cannot be in strong LD with trait-influencing genes that have low allele frequencies. If most of the variation within the cattle genome that influenced a trait was at low frequency, genomic selection simply would not be able to achieve high accuracies for molecular EPDs, regardless of how many high-allele-frequency SNPs were scored or the number of animals that were genotyped to build the molecular EPD prediction model.
This appears to be the case in humans for highly heritable traits such as height, where genotyping 30,000 individuals explained only about 5%-10% of the variation. This suggests that there are many genes that cause variation in human height and that most have low allele frequencies so that different individuals have different combinations of genes that determine their height and these genes are generally not detected by SNP chips, which use high-frequency markers.
Fortunately, we have demonstrated that this is not the case for milk production traits in Holstein cattle, where molecular EPDs (PTAs) have accuracies of 60%-70% using prediction models based on ~6,000 animals. Therefore, the problems that face the implementation of genomic selection in Bos taurus beef cattle are:
- 1. There is no DNA Repository for beef bull semen where samples of 6,000 bulls can be amassed for construction of the molecular EPD prediction model;
- 2. LD does not extend far across breeds, so pooling animals across breeds may not provide a solution — unless a very high-density SNP chip can soon be produced;
- 3. There are few animals with phenotypes/EPDs for fertility, feed efficiency and disease resistance that can be used to develop molecular EPD models; and
- 4. The cost of current assays are ~$200 per sample, which will limit adoption.
To tackle these issues, we have genotyped ~8,000 animals representing Angus, Limousin, Charolais, Hereford and Simmental with EPDs/phenotypes for growth, carcass, feed efficiency and meat quality and our initial focus is the development of affordable tests based upon low-density SNP genotyping assays. The current status of this work will be reported at the meeting.
About the speaker:
Jerry Taylor holds the Wurdack Chair in Animal Genomics in the Division of Animal Sciences at the University of Missouri–Columbia. Jerry received a bachelor’s degree in mathematics and a bachelor’s honors degree in mathematical statistics from the University of Adelaide, and a doctorate in quantitative genetics from the University of New England in Australia. He is an elected fellow of the American Association for the Advancement of Science and received the 2008 Celebration of Excellence Distinguished Researcher Award in the College of Agriculture, Food and Natural Resources at the University of Missouri.
He is a member of the iBMAC Consortium, which developed the Illumina BovineSNP50 assay. Along with his co-investigators and graduate students, he has authored 144 peer-reviewed research articles, four book chapters and four patents, and he has edited one book.
He has assembled DNA samples and phenotypes on more than 18,000 Angus, Limousin, Charolais, Hereford, Simmental and Holstein cattle, and his research focuses on the identification of mutations responsible for phenotypic variation in growth, carcass composition, feed efficiency and milk production.
Prior to joining the University of Missouri in 2002, he was director of Genomics at RTI International in the Research Triangle Park, North Carolina. Taylor was cofounder, board member and a senior executive of GenomicFX, an agricultural biotechnology company located in Austin, Texas, where he participated in activities including merger and acquisition, product and business development and intellectual property management. From 1986 to 2000, he was a professor in the Faculty of Genetics and Department of Animal Science at Texas A&M University. Previously he was an assistant professor in the Graduate School of Tropical Veterinary Science at James Cook University in Townsville, Australia.
BIF thanks the California Beef Council for sponsoring Taylor’s attendance at this meeting.
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.
