Saturday, April 20, 2024

Predicting breeding merit

Avatar photo
From the basic to the complex: Russell Priest reports on two presentations dealing with the technology involved in making genetic predictions. The ultimate goal of being able to take some DNA from an animal and predict how that animal might perform is now not realistic. So says renowned New Zealand-born geneticist Dr Dorian Garrick. “The best we can expect now is to be able to predict breeding merit in close relatives within a breed and quantify the reliability of that prediction. Garrick, who holds the Jay Lush chair in Animal Breeding and Genetics at Iowa State University and is also the director of the US National Beef Cattle Evaluation Consortium, delivered two papers at the Beef + Lamb Ag-Innovations event in Feilding. His two presentations dealt with the technology involved in making genetic predictions about the performance of animals based on their DNA.
Reading Time: 2 minutes

Basic genetic principles: Garrick covered some basic background information on genetics before dealing with more complex issues. He stated that most of what we know about the cattle genome has come about because scientists believe it might hold some useful information for human genetics.

“Genes have beginnings and ends and are part of structures known as chromosomes and they come in pairs; one of each pair comes from the father and the other from the mother,” he said. “Humans have 23 chromosomes and cattle have 30, and there are approximately 30,000 genes in total. Genes are represented by base pairs and each chromosome has approximately 100 million of these.

Gene markers: As mentioned, gene markers are developed by scanning the DNA of groups of thousands of animals.

These groups are known as “training” or “discovery” groups. It is an expensive exercise developing gene markers for such large groups. Genotyping a group of 10,000 animals using a 20K chip would cost US$450,000, then the markers developed have to be tested to ensure they work.

The markers relate only to the population of animals from which the training group originated. For example, markers developed for Angus in the US will not necessarily work for Angus in this country unless there are strong pedigree linkages between the two populations. Most major breeds in the US have developed marker tests. The breeds that have developed these using large training populations are the breeds that have the most accurate genetic predictions.

Genomic verses pedigree-based predictions: Several different genotyping tests have been developed and used by different breeds in the US to develop their genetic marker tests. All have proved to give better genetic predictions than traditional pedigree and performance-based systems.

The accuracy of the gene marker tests depends heavily on the size of the training populations of genotyped and phenotyped (performance recorded) animals.

Blending: This is a technology where genomic information is combined with traditional pedigree and performance information to endeavour to improve the accuracy of the genetic prediction.

Garrick said four blending methods were used in the US and they all had serious problems. He believed none of them was going to work in the long term.

“Blending will only add to the accuracy of the prediction if the accuracy of the original information is very low. If an animal is already accurately evaluated by using a progeny test, for example, then a genomic prediction will not improve accuracy.”

Big genes: “There have been some major genes found that are causing large genetic variation and these are in similar chromosomal regions in several breeds,” Garrick said. “Some of these same regions describe a lot of the genetic variation for weaning weight, yearling weight, marbling, rib eye area and calving ease.

“However the 50K chip is not dense enough to be able to identify these gene effects.”

The current status of DNA technology:

  • Gene marker technology often achieves good predictions in close relatives but not in distant relatives
  • The reliability of the predictions varies according to the trait being assessed and how closely the animals being predicted are related to the animals in the training group
  • Imputation will make genomic prediction more affordable
  • Characterisation of major gene effects will improve genomic predictions.
People are also reading