As anyone with an e-dating horror story will attest, relationships can be tricky. It’s not surprising, then, that estimating genetic correlations (relationships) among different traits in Canadian dairy selection is no small feat. For researchers on Genome Alberta’s Efficient Dairy Genome Project, it’s a tough task that’s well worth the trouble.
“If you go on farm and look at how two traits work with each other, you can calculate the phenotypic [observable] correlation,” said Pauline Martin, Postdoctoral Fellow at the University of Guelph.
Nature vs Nurture
Phenotype has a genetic component, yet is also highly dependent on the environment where the animal is raised. For example, with milk production, a cow has the genetic potential to produce a certain amount of milk. But the amount finally produced depends also on the feed of the cow (amount and quality) and plenty of other factors such as the health of the cow or the management of the breeder.
“My task is to separate all the environmental effects to estimate the relationship between the genetic governance of two traits. For that, I use mathematical modelling to estimate for each animal the effect of the main environmental factors. Within the same model, I also look at the performance of the animal’s relatives to extract the genetic part and help me separate the environmental effects.”
Making an impact
This work is critical to a key component of the research project, which is implementing a routine genetic evaluation service for feed efficiency (FE) and methane emission (ME). That involves estimating the real impact on other traits of selecting for feed efficiency and methane emission. To do so, researchers need to know two things: the genetic relationship between FE and ME and the other traits, and the genetic relationship among the other traits.
“In the last decade, more than 80 new traits have been evaluated routinely by the Canadian Dairy Network, and more are coming. However, there was no systematic estimation of genetic correlations between a new trait entering the evaluation and the previous ones. As this information will be needed for the project, it was the right time to overview the correlations previously calculated in other studies and estimate the missing ones.”
Correlation overload
It sounds easy if you say it quickly; in reality, though, there are more than 3,000 two-by-two correlations among the 80 traits. Researchers thus restricted themselves to traits that can be calculated from first lactation animals. With that, they wound up with around 30 traits that describe production, type, fertility, workability and health. That still corresponds to around 400 correlations, but it’s manageable.
Though it’s still a lot of work, the results are already paying dividends.
“Some of my colleagues already use my results to determine the best breeding strategy to incorporate feed efficiency in a selection index in Canada. These results will also be useful for the international research on dairy in general.”
The results could also improve the actual genetic selection.
“Now that we know which traits go together, we can calculate correlations among groups of traits directly – not just two-by-two – and incorporate this information into the selection equation. It may improve the accuracy of the selection, especially for low heritability traits that can benefit from their correlation with more heritable traits.”
The team’s success in correlating traits is drawing interest from abroad. France, which is also studying these traits, has asked about joining the project and recruited Martin to analyze their data, and somehow that’s fitting. Whether it’s genetic correlations or online dating, relationship challenges are the same in any language.
