Obtaining the genomic information
Genomic selection is a way to select animals based on genotyping or reading the DNA of the candidate.
The pig’s genome was first mapped in 2009 (Archibald et al.) and so-called DNA-chips (or SNP-chips) were developed. These chips are a collection of microscopic DNA spots attached to a solid surface that comprise areas of the genome which vary between animals. The type of variations used on the current SNP-chips are called Single Nucleotide Polymorphism (SNP). These SNPs are distributed through the whole genome. Each of them has 2 possible forms (called alleles) and showing variability in Choice populations. Choice has developed its own SNP chip to guaranty capturing best this variability.
Genotypes are obtained by reading these DNA chips. A DNA chips holds 60 000 probes, each able to hybridize with an allele form of a SNP. Hybridization is linked to a color which enables laser light to make them vibrate and provide different colors that discriminate the two alleles.
So, after collecting a tissue sample from the pig, DNA is extracted, purified, and amplified to be put in contact with the probes. After this the chip can be machine read and the genotype of the candidate is known for each SNP.
Using genomic information in genomic evaluation
Once the genotypes are obtained, the next part of genomic process is the construction of the evaluation model. By studying the genotypes and the phenotypes associated with them, a model assigns infinitesimal effect to each SNP on the different traits evaluated.
SNP are rarely the causal mutation. However, due to physical linkage on chromosomes or to previous selection, these SNPs can reflect for a part of the genetic variation due to the causal mutation. Because of this their correlation with changes in the phenotype can be calculated statistically. In addition, these SNPs effects are constantly updated by adding new animals with phenotypes and genotypes. That is why even in the genomic era, phenotypes are kings.
The goal of selection is to improve animals’ genetic potential for performance; the use of genomics gives geneticist new tools to do so faster and more efficiently.
The statistical methods used have been established for decades in animal genetics. Meuwissen et al. (2001) proposed three methods which were evaluated based on simulated data. Over time, these methods have been refined and many new ones have been proposed and implemented. A great step has been the easy calculation of the genomic relationship matrix (vanRaden, 2008) which is substituted for the traditional pedigree relationship matrix in gBLUP (Genomic Best Linear Unbiased Prediction).
What is the gain with genomic selection?
Gain in accuracy
Since DNA sequence is not linked to the environment nor changes during life of the animal, genotype is therefore a measure that can be captured early in an animal’s life. By adding information, genotypes lead to higher precision in the genetic evaluation.
Like classical EBVs (Estimated Breeding Values), gEBVs (Genomic Estimated Breeding Values) can be calculated for an animal before it has any phenotypic performance recorded. This additional information improves the accuracy of the gEBVs compared to the EBVs at the same age under the same circumstances. It is then obvious that there is a gain for traits that are measured late in life, are sex-linked, difficult, or expensive to measure. This gain in accuracy can also be observed for traits with low heritability as the genotype is not impacted by the environment.
Early selection among full sibs
Being able to access an individual’s genotype allows geneticists to distinguish between full sibs without phenotypic data. Because of this we can identify and keep as candidates the most promising animals and focus on their measure and evaluation.
This higher accuracy leads to a better selection of future breeding animals (see FAQ “WHAT ARE THE DRIVERS FOR GENETIC PROGRESS?”). Because of this better choice of animals to keep, genetic progress is increased each generation. These gains are cumulative generation after generation.
What if an animal has not been genotyped?
Originally, only animals which had a genotype could be included in genomic evaluations. The global cost to run a genomic program was therefore very high if all animals were to be tested. In species where the value of an individual animal is low, this cost was a limiting factor and a strategy needed to be developed to link animals evaluated with genomics and other without genomics into a single set of EBVs and index.
In 2009, a method (Legarra et al.) was developed to enable the common evaluation of both genotyped and non-genotyped animals: the ss-gBLUP (single-step Genomic Best Linear Unbiased Prediction).
In this method, an appropriate association matrix is constructed and joins all available information (genotypes, performances, and pedigree) to be used by a ss-gBLUP to evaluate candidates. After this ss-gBLUP, gEBVs can be used the same way as EBVs and all animals are evaluated the same way.
How do we apply genomic selection at Choice?
At this time in Choice ‘s breeding program, genotyping is mainly applied on male candidates in dam lines. Since boars are selected with a higher selection intensity and have more offspring than females in a pig breeding system, their influence on the population is higher. Because of this investing in testing more on boars make sense to better spread the genomic gain.
Since ss-gBLUP combines all information sources, all candidates (male and females) benefit from the better accuracy, even if they are not all genotyped.
In practice, the GEBVs are used the same way as EBVs to construct indexes to increase the global performance of the line.
Choice’s selection program adds the genomic technology to the tools used to improve our animal’s performance. This advanced method is based on the high-quality data collection on our candidates which allows Choice to make increased progress on our lines, and to bring the best possible product to our customers.
Archibald AL, Bolund L, Churcher C, Fredholm M, Groenen MAM, Harlizius B, Lee K-T, Milan D, Rotschild MF, Uenishi H, Wang J, Schook LB & the Swine Genome Consortium (2010). Pig genome sequence – analysis and publication strategy, BMC Genomics, 11: 438.
Legarra A, Aguilar I, Misztal I (2009). A relationship matrix including full pedigree and genomic information. Journal of Dairy Science, 92(9): 4656-4663.
Meuwissen TH, Hayes BJ, Goddard ME (2001). Prediction of total genetic value using genome-wide dense marker maps, Genetics, 157(4): 1819-1829.
vanRaden PM (2008). Efficient Methods to Compute Genomic Predictions. Journal of Dairy Science, 91, 4414-4423.