Selective breeding in the genomic age

PROFESSOR DORIAN GARRICK - School of Agriculture, College of Sciences

How do farmers and other breeders know which animals to breed from? The decision is complicated, given that not only are there many individuals to choose from, but there is also a wide variety of characteristics on which to base selection decisions.

‘We base that decision on predictions about how those offspring are likely to perform,’ says Professor Dorian Garrick of Massey’s School of Agriculture. ‘If we’re talking about dairy cattle, there is, for example, the volume of milk they produce, the amount and nature of the protein and the fat in that milk, the likelihood of them getting disease, whether or not they are polled or horned, and how long the daughters might last in a herd.

We have to make predictions simultaneously for a large portfolio of traits and, on the basis of those, decide which candidates we think are the most appropriate to generate the kinds of individuals we would like to have in the next generation.’

When he was in the early stages of his career breeders relied solely on predictions based on pedigree information and performance measures on the selection candidate itself, as well as its ancestors and any progeny. ‘We’ve always known there were limitations with the pedigree-based approach, because if you breed together an elite sire and an elite dam not all of your offspring are elite,’ Professor Garrick explains. ‘Full brothers and sisters vary quite a lot in their performance but we can’t tell from the pedigree which full sibs are going to be better than the others without somehow trying to observe their genes.’

Traditionally, the only way to observe an animal’s genes or genotype for complex traits was to observe their visible characteristics, or phenotype. ‘This meant measuring them for particular performance traits, such as seeing how well they grew. But over the last ten to fifteen years we have increasingly been using genomic information to make that decision as well.’

We have to make predictions simultaneously for a large portfolio of traits and, on the basis of those, decide which candidates we think are the most appropriate.


Today, genetic technology makes it possible to directly characterise the genome, allowing breeders more power to distinguish between brothers and sisters in terms of which ones have inherited more favourable characteristics. But the sheer amount of information now available comes with its own challenges. ‘The most common assays have had 50,000 genetic markers from each individual, which creates statistical challenges in itself,’ says Professor Garrick. ‘But more recently, the numbers of individuals that are genotyped in these populations has gone up and we have also seen an increase in sequencing, so we might have ten million or more variants we are aware of in each of these candidates.

‘Probably more than half of my work is theoretical, working on how to handle this data, including how to make it computationally workable on a normal computer. Then, I work in the application of the algorithms that we develop in a number of different species and some of the practical problems associated with those species. I work a lot in beef cattle and dairy cattle in particular.’

A major puzzle that Professor Garrick has helped to solve has involved finding a way to integrate both traditional pedigree and genomic information into a single useable result. ‘Initially, we would come up with a prediction based only on the DNA and then we’d also have a prediction based only on the traditional pedigree methods,’ he says. ‘A number of researchers around the world thought it would be much better if we could develop what is known as a single-step approach, where we take the genotypes and the pedigree and performance information and in one analysis tell who are the best candidates. With colleagues, I developed some innovative algorithms and software to solve that particular problem, and it is now used by some of the leading food-producing companies around the world.’ To market and develop the software, Professor Garrick formed a company, Theta Solutions, LLC.

As well as the improvement of complex traits, the approaches Professor Garrick has developed are also used to investigate the genetic basis of animal diseases. And he is always keen to give help and advice. ‘The Vet School has had cases coming through pathology where there are suspicions they might have a genetic basis. We have now found the causal mutation for four different diseases in sheep. For anybody who picks up the phone or emails with a problem we might be able to make some traction with, we will do our best to try to find some approaches to move forward and help them out.’

Project details

Websites College of Sciences, Professor Garrick on Google Scholar