Accuracies of prediction
Dinesh Thekkoot, PhD
University of Alberta and Genesus Inc.
The aim of animal breeding is to genetically improve populations. In other words, the next generation should be superior to the current generation. The challenge is to identify the best animals to be parents of the next generation. To identify a future parents we need to know their true breeding (genetic) value (TBV). In real life it is impossible to observe the TBV of an animal. Thus we use an estimate of the TBV called an estimated breeding value (EBV). Parents of the next generation are selected based on their EBVs.
An important question is how accurate an estimate is the EBV of the TBV? Accuracy of EBV is defined as the correlation between true and estimated breeding values. This correlation gives a sense of how close the EBV is to the TBV. Accuracy of EBV can range from 0 (an inaccurate estimate) to 1 (EBV being a perfect estimate of TBV). The higher the EBV accuracy the lower the risk of selecting animals with low TBV as parents. EBV accuracy also influences the genetic gain from a selection program. The genetic gain in a breeding program is proportional to the accuracy of the EBV. The higher the EBV accuracy with no change in other factors, the greater will be the genetic gain from selection.
The genetic gain per year is calculated using the formula
Where is the genetic gain per year, is the intensity of selection, is the EBV accuracy , is the additive genetic standard deviation of the trait under selection and is the generation interval expressed in years. In most of the pig breeding programs, the intensity of selection and generation interval are close to being optimized without having a negative effect on genetic variation. Therefore, for any further increase in we need to introduce methods that can increase the accuracy of EBV.
The EBV accuracy depends on many factors like, heritability of the trait, amount and quality of the information available on the selection candidate and relatives, genetic relationships among the all animals, etc. In general, the higher the amount of information available, the higher will be the EBV accuracy.
EBV accuracies using different methods
- Best linear unbiased prediction (BLUP) animal model. This method is used in most of the genetic programs today and utilizes all available information for calculating EBVs. The EBVs are estimated using phenotypes on the selection candidates and relatives along with the pedigree relationships. Along with prediction, BLUP animal models simultaneously correct the phenotypes for systematic environmental effects. For example, this method accounts for genetically related animals at different farms while estimating EBVs. This is the most accurate method for predicting EBVs utilizing both pedigree and phenotypes.
- Genomic Selection (GS). Young animals that do not have their own phenotype will have low EBV accuracies using BLUP animal model methods. Depending on the trait, it might take a considerable amount of time to get an own phenotype or some animals may never get an own phenotype. For example, it takes around 12 months to get litter information on a gilt and animals selected as parents will never get carcass and meat quality phenotypes. Hence it would be good if we can increase the EBV accuracies for these types of traits. Genomic prediction is a recently developed method to address this issue. It predicts EBVs using molecular marker information accurately without the need for own phenotype. Studies have been conducted at Genesus and elsewhere, to estimate accuracies using GS. Results show that for a large number of traits accuracies from GS are higher than BLUP methods based on phenotypes and pedigree information.
- Combining genomic, phenotypic and pedigree information can result in higher accuracies than GS. Single Step genomic BLUP (SSBLUP) is the method that combines these three sources of information. Currently this method provides the highest EBV accuracies for young animals. Studies conducted at Genesus have shown that SSBLUP EBV accuracies were around 30-50% higher than BLUP based EBVs. A higher accuracy of prediction means a greater genetic change per year.
Accuracies of EBVs and expected genetic change to selection for some traits using two different methods
Traits | Genetic standard deviation | BLUP | SSBLUP | ||
Accuracy | Genetic change/year* | Accuracy | Genetic change/year* | ||
Total born | 1.15 | 0.31 | 0.63 | 0.42 | 0.85 |
Sow back fat at farrowing (mm) | 2.49 | 0.36 | 1.57 | 0.53 | 2.32 |
Sow body weight at farrowing (kg) | 8.61 | 0.32 | 4.84 | 0.40 | 6.04 |
*Assuming a selection proportion of 10% and generation interval of 1 year for all traits and for both methods
As can be seen in the above table the accuracy of the EBV and the resulting genetic change per year increased by adding genomic information (BLUP compared to SSBLUP) for all three traits. Clearly the impact of genomic information on genetic change per year is significant. At Genesus we produce the most accurate EBVs for our selection candidates. We have invested heavily to collect accurate and up-to-date information on all growth, efficiency, reproduction, carcass and meat quality traits, for all animals in our nucleus, multiplier and production herds. This high quality data along with genomic information helps us to estimate the most accurate EBVs possible, assuring the highest genetic response and maximum profitability for Genesus customers.
Categorised in: Featured News, Global Tech
This post was written by Genesus