Tuesday, August 7, 2012

Gene Tests vs Genomic Selection

Two different paradigms have existed regarding the use of DNA markers in animal breeding. The first strategy is gene tests, also referred to as marker assisted selection (MAS).  The second is genomic selection.

Gene Test

Gene tests attempt to predict a trait or breeding value based on the results (genotypes) of a small number of DNA markers. These tests are either developed using a candidate gene approach or from genome-wide analyses.  In a candidate gene approach, a scientist assumes which genes influence a trait and investigates variants within those genes for an association with the trait of interest.  These assumptions can be wrong and a scientist may identify an association by random chance.  In a genome-wide approach, a scientist makes no assumptions about which genes influence the trait, but analyzes markers evenly spaced throughout the genome.

Genomic Selection

In genomic selection, thousands of evenly spaced DNA markers are genotyped in a large population of animals (more than 1,000).  A statistical model is then created which predicts estimated breeding values of an animal based on that animal's genotypes.  Often, these genomic predictions are combined with traditional genetic evaluations to create genomic-enhanced EPDs.   

Gene Test Fall Flat

The first gene test, GeneSTAR, was marketed in 2000.  It tested a single variant in a single gene that was thought to influence marbling.  In the years that followed more gene test were developed.  For simple traits, which are not influenced by the environment and are controlled by a small number of genes, gene tests are highly accurate.  Genetic abnormalities, coat color, and the presence or absence of horns are examples of simple traits.  But, the majority of economically relevant traits are complex.  Complex traits (growth, marbling, calving ease, etc.) are influenced by both the environment and hundreds or thousands of genes.  Typically, these hundreds of genes individually have a small effect on the trait.  Thus, testing a handful of DNA markers will not lead to an accurate prediction of an animal's genetic merit.  The NBCEC reports the results of attempts to validate many of these tests.  For example, the previously mentioned gene test for marbling was not significantly associated with marbling score. The gene tests for meat tenderness are accurate predictors, although the commercialized tests do not appear to be the causal variants.  Except for gene tests for meat tenderness, most of these tests were a poor investment for cattle producers.

Genomic Selection Delivers the Knockout Punch

In 2001 three scientists, Theo Meuwissen, Ben Hayes, and Mike Goddard, proposed a new way to utilized DNA markers in animal breeding (article is here for you scholarly types).  They showed that using genotypes from tens of thousands of DNA markers spread throughout the genome, accurate estimates of genetic merit could be predicted.  Rather than looking at a small number of genes, every single gene in the genome is taken into account by a nearby DNA marker.  Unfortunately, the DNA technology to implement this method in cattle was not available until 2008.  But, since 2008, genomic selection has been implemented by the dairy industry (see popular press articles in Forbes and The Atlantic), and it is starting to gain traction in the beef industry.

Conclusion

So, why was the Angus breeder in my first post skeptical about the use of genomics in animal breeding?  I'm not sure, but I would venture to guess that livestock producers don't realize that there is a major difference between gene tests and genomic selection.  Of course, we will continue to use gene tests for simple traits, especially genetic abnormalities.  But, when predicting quantitative and complex traits, gene tests are out and genomic selection is here to stay.  
Post a Comment