Fusing EPD information with DNA test results to improve prediction accuracy is now a reality.
The promise of using DNA information to improve the accuracy of expected progeny differences (EPDs) on young animals is starting to be realized, at least for Angus cattle.
First, there was an agreement between Angus Genetics Inc.® (AGI) and IGENITY® to calculate genomic-enhanced expected progeny differences (GEPDs) for multiple carcass traits using American Angus Association® carcass and ultrasound data. Then, in November 2010, AGI announced an agreement to similarly accept Pfizer Animal Genetics’ High-Density 50K test results for incorporation into GEPDs.
Genomic-enhanced EPDs are produced using DNA and traditional (performance records, pedigree) information sources. The inclusion of DNA information should improve the accuracy of EPDs, especially for young animals with little performance data. Producers often ask how much a DNA test improves accuracy. The answer depends on the amount of genetic variation accounted for by the test, and the accuracy of the EPD for that trait in the absence of DNA information.
Figure 1 shows the Beef Improvement Federation (BIF) accuracy of a DNA test that accounts for 25% of the genetic variation in a trait. If no pedigree, individual or parent records exist for that trait, then the BIF accuracy of an EPD based on the DNA data alone would be 0.13 (red line). If an animal and its parents have records for the trait (blue line), then the BIF accuracy would range between 0.07 and 0.35 depending upon the heritability (h2) of the trait (see Table 1 for h2 estimates of common traits).
In that situation, the addition of DNA test information would boost the BIF accuracy most (0.11) for the low heritability trait, and least (0.05) for the most highly heritable trait. In practice, this means that DNA testing is most useful for traits for which no other record exists (e.g., meat tenderness), and/or low heritability traits (e.g., calving ease, fertility).
To incorporate DNA test information into EPDs, it’s necessary to determine the amount of genetic variation associated with the test (25% in the example in Figure 1). Ideally this value would be high, such that DNA tests results would be very predictive of genetic merit. Independent estimates of this proportion aren’t available for all breeds and tests on the market.
Some estimates are available for Angus cattle (Table 1). The U.S. IGENITY estimates were developed in the U.S. Angus population, whereas the U.S. Pfizer estimates were developed in company-derived populations, and the Australian estimates were developed in the Australian Angus population. To incorporate Pfizer HD 50K data into the American Angus EPDs, these estimates will need to be recalculated for the U.S. Angus population.
Breeds other than Angus are likewise going to have to derive these genetic variations associated with any DNA test. To date, data suggest that tests developed for one breed are unlikely to work very well in another.
Even within Angus, tests trained in North American Angus were associated with less genetic variation when used in the Australian/New Zealand Angus population, and required regional recalibration for that population and production system. One possible reason for this is that a portion of the accuracy associated with genetic testing is the result of linkage – relationships between the animals in the training population (the population used to develop the test) and the test population. Tests are likely to be less accurate (explain less of the genetic variation) when used to evaluate more-distantly related animals, such as those in another country where animals may be several generations removed from U.S. genetics.
The National Beef Cattle Evaluation Consortium (NBCEC) has encouraged discussion about how to develop DNA tests to improve the accuracies of EPDs for breeds other than Angus. One of the requirements in DNA test development is access to a large training population of genotyped animals from the target breed or its crosses. Some breed associations (Hereford, for example) are starting to develop such populations with an eye toward developing breed-specific DNA tests. A similar approach is being implemented in Australia where breed associations are establishing reference populations to provide training and validation populations for breed-specific DNA test development.
It’s becoming increasingly clear that to obtain accurate DNA tests, it’s necessary to train on large numbers of records. The dairy industry is currently training on records from tens of thousands of animals including well-proven Holstein sires to obtain their genomically enhanced predictive values. Obtaining similarly large numbers of records from breeds other than Angus poses a challenge for the beef industry.
The recent development of a high-density (~700K SNP) bovine marker panels provides a potential solution.
With the 50K marker panel, as shown in Figure 2, a marker associated with a trait in one breed may not have the same association in another breed. The reason is that the marker is often located a “long way” from the gene and so it’s not found to be associated in all breeds with the variant of the gene that is causing the effect. By increasing the number of SNP markers to 700 K, markers are more closely spaced and there is a greater likelihood of finding SNPs that are close to the gene (red markers in the diagram), and hence the marker will “work” in both breeds.
These high-density bovine marker panels provide an opportunity for breeds to pool information. Developing large multi-breed training data sets may collectively improve the accuracy of tests for all breeds, more than any single breed can do on its own due to the larger number of combined records.
These high-density panels come with increased cost ($200-$300 each), but this price will likely drop in the future. Ultimately the availability of high-density arrays may enable the use of DNA test information to improve EPD accuracy for many breeds of cattle, including those with a relatively small number of performance-recorded animals.
Although this field is evolving quickly, the ultimate goal for using genomics remains the same, and that is to cost-effectively improve accuracy of EPD produced in National Cattle Evaluation (NCE). As outlined by BIF, “information from DNA tests only has value in selection when incorporated with all other available forms of performance information for economically important traits in NCE, and when communicated in the form of an EPD with a corresponding BIF accuracy. For some economically important traits, information other than DNA tests may not be available. Selection tools based on these tests should still be expressed as EPD within the normal parameters of NCE.”
a. Heritability from http://www.angus.org/Nce/Heritabilities.aspx; or literature estimates.
b. MacNeil, M. D., S. L. Northcutt, R. D. Schnabel, D. J. Garrick, B.W. Woodward, and J. F. Taylor. 2010. Genetic correlations between carcass traits and molecular breeding values in Angus cattle. 9th World Cong. Genet. Appl. Livest. Prod. Leipzig, Germany. August 2010.
c. Pfizer Animal Genetics. 2010. Technical Summary. http://www.pfizeranimalgenetics.com/sites/PAG/Documents/50K Tech Summary.pdf
d. Animal Genetics and Breeding Unit (AGBU). 2010. Evaluation of Pfizer Animal Genetics HD 50K MVP Calibration. http://agbu.une.edu.au/pdf/Pfizer_50K_September 2010.pdf