Recently, the DNA testing industry matured from marker tests involving a handful of markers explaining a relatively modest amount (0-10%) of the genetic variation in the target trait, to panels involving hundreds or thousands of markers. This is an exciting development because many of the production traits of interest to beef cattle producers are likely to be controlled by a large number of genes.

The proportion of genetic variation explained by these high- density panels provides producers with a way to quantitatively evaluate the merit of commercial products. Accurate estimates of this proportion should now be the focus of test-panel evaluations. Such estimates will also enable breed associations to incorporate DNA data into expected progeny differences (EPDs).

BIF recommendation

DNA tests can be used much more effectively when incorporated into and presented as EPDs as recommended by a recent Beef Improvement Federation (BIF) task force. The recent incorporation of DNA tests into EPDs for carcass traits by the American Angus Association is an important milestone in the application of DNA testing in beef cattle. This greatly simplifies selection decisions by eliminating the dilemma of how much relative emphasis to place on DNA tests vs. conventional EPDs, when both are available.

Under this scenario, the value of a DNA test will be dependent upon how much the DNA test improves the accuracy of EPDs at the time of selection and the economic importance of the trait. DNA tests are most valuable for traits that aren't routinely recorded before selection decisions are made. For example, a DNA test accounting for 50% of the genetic variation in an economically important trait like feed efficiency for which EPDs do not exist is likely to be more valuable than a DNA test accounting for 50% of the variation in an easily measured trait like weaning weight.

For traits where EPDs aren't available, the accuracy of the test result will depend solely on the proportion of genetic variation explained by the DNA marker test. Table 1 reports the BIF accuracy associated with a DNA test explaining varying proportions of the genetic variation in a trait. Also reported is the number of progeny performance records for two levels of heritability required for a sire's EPD to achieve an equivalent BIF accuracy, in the absence of any other information on the sire.

In order to make informed purchasing decisions, producers want and need accurate information regarding how well DNA tests work. The U.S. National Beef Cattle Evaluation Consortium (NBCEC) has provided this information by independently “validating” DNA tests for the U.S. beef industry since the early 2000s.

From the beginning, this has been a voluntary process conducted in cooperation with, and initiated by, the DNA testing companies. The original emphasis of validation was on whether the test “worked” as claimed by the company marketing the test.

NBCEC posts the results for all tests that have undergone the validation process at www.NBCEC.org, regardless of the association of the test with the target trait, unless the test is withdrawn from the market. Companies have sometimes used the term “validated by NBCEC” to indicate that the test went through the “validation process.”

This has caused some confusion in cases where the results were mixed in that the test worked well in some populations and had no effect or even a negative association with the trait in other populations. The table on page 16 of this issue documents the currently marketed DNA tests for production traits, and includes the summary statement from the NBCEC website for tests that have been through the validation process.

While statistical significance is an important measure of whether a DNA test works, the amount of genetic variation explained by the test is of much more practical importance to seedstock and commercial beef producers. As a result, NBCEC's validation process has moved away from reporting only whether a DNA test works, toward an estimation of the proportion of genetic variation explained by DNA marker panels.

Estimating the value

NBCEC developed a robust procedure to estimate this value, and the consortium stands ready to estimate these values and post them to the NBCEC website as a part of its ongoing efforts to provide unbiased information on DNA tests to the beef industry. Since this procedure became available, there have been no requests by DNA testing companies for an NBCEC validation.

Commercial companies marketing high-density tests such as the Pfizer HD 50K for Angus, and panels that contain a subset of the most informative markers derived from 50K experiments like the Igenity profile for Angus, have chosen to estimate the proportion of genetic variation explained by their tests themselves, or via other entities. Some potential customers of the technology are skeptical of using DNA tests based solely on the claims of the companies selling the tests.

The process of independent validation isn't without challenges. The availability of appropriate populations has been a continual problem. The shift from validation to estimation of the proportion of variation explained places additional demands on the population structure. Ideally, validation should be undertaken using a population unrelated to the discovery population used to identify the markers believed to be predictive of any particular trait.

However, it's now commonplace to use discovery populations that include many, if not all, of the widely used AI sires in a particular breed. In that case, the validation population is likely to include close relatives of the discovery population. Inclusion of animals that are very influential in a particular pedigree in the discovery process makes it essentially impossible to have a sample of animals that are “independent” from the intended population of inference due to genetic relationships.

These relationships can result in inflated estimates of the proportion of genetic variation explained by DNA marker panels. Therefore, validation results in populations related to the discovery population are likely to represent the upper limit or “best-case scenario” of the performance of the test.

Clearly, the validation process and underlying analyses have evolved as the DNA testing industry has matured from single-gene tests to panels of an ever-increasing number of markers. Focusing the validation process on providing the data required to incorporate the DNA marker information into national genetic evaluation systems comes at an opportune time, given the emergence of products derived from high-density assays.

NBCEC is excited about the opportunities that will become available using high-density markers and corresponding reduced panels. However, NBCEC has been told by leading cattle producers of the importance of presenting DNA information in a way that allows producers to determine their own value proposition.

Many of these producers want the information delivered as EPDs with corresponding BIF accuracies, which are the existing currency for communicating genetic merit. It's difficult to envision how producers can evaluate investment in DNA testing for genetic improvement in the absence of such information.

Ultimately, it will be the demand of producers for this information as a prerequisite for purchase that will dictate the rate and value of adoption of this potentially game-changing technology.

Alison Van Eenennaam is a University of California-Davis Extension animal biotechnology specialist.

Table 1. Relationship between the percentage of genetic variation explained by a DNA test (assuming the animal and its relatives have no records for the trait) and Beef Improvement Federation (BIF) accuracy, and the number of progeny test records required to obtain equivalent accuracy values for traits of low (0.1) and moderate (0.3) heritability
Number of progeny records required
Percentage of genetic variation explained by a DNA test BIF accuracy Low heritability (0.1) Moderate heritability (0.3)
1% .01 1 1
4% .02 2 1
9% .05 4 2
16% .08 8 3
25% .13 13 5
36% .20 22 7
49% .29 38 12
64% .40 70 22
81% .56 167 53
98% .93 1,921 608
99% .99 3,800 1,225