L.GEN.1815-Development and delivery of improved genomic prediction tools for sheep
Comparison of GWAS within sources showed chromosomal regions significant within breed sources.
| Project start date: | 14 May 2019 |
| Project end date: | 06 January 2022 |
| Publication date: | 15 August 2022 |
| Project status: | Completed |
| Livestock species: | Sheep |
| Relevant regions: | National |
|
Download Report
(4.3 MB)
|
|
Summary
The objectives of this project were to develop and deliver knowledge and tools to improve genomic prediction for significant economic traits in sheep, in particular for eating quality and reproduction traits.
Advanced statistical genetic methods were used for the analysis of phenotypic data collected by breeders (mainly for reproduction traits) and collected from the resource flocks (mainly for eating quality traits). Phenotypic data was combined with imputed whole genome sequence genotypes of ~200,000 Australian sheep, to undertake association analysis to find and select predictive SNPs, and to validate increased genomic prediction accuracy from using these SNPs. Different models and methods on using selected SNPs in the Sheep Genetics single step genetic evaluation (OVIS) were explored and tested.
Objectives
The primary objective of the project was to update the set of informative SNPs, using more data and for a larger number of production and reproduction traits including whole genome sequence data, in order to deliver improved genomic prediction accuracy to the Australian sheep industry. The updated set of informative genetic markers was used to evaluate whether the genomic prediction accuracy would be improved compared to the previous test performed in 2018. The project was successful and showed a consistent higher prediction accuracy based on an updated list of informative SNPs.
Key findings
• With the updated and larger data sets Genome wide association studies (GWAS) results showed a larger number of chromosomal regions with quantitative trait loci (QTL) for slaughter and eating quality traits, with numerous new significant chromosomal regions as well as confirmed significant genomic regions detected in previous analyses in 2018. Some chromosomal regions affected several carcase and eating quality traits.
• GWAS results in reproduction traits showed chromosomal regions affecting litter size. GDF9 was found to have a major effect on litter size due to a missense mutation in some maternal breeds. GWAS on tail-length showed one highly significant chromosomal region with a large effect.
• We selected a set of 3,700 significant SNPs across seventeen production and reproduction traits. When including this set of informative markers to the full SNP panel, increased accuracy was observed in single trait genomic prediction (GBLUP) of between 0.04 and 0.16 (absolute value) in EQ traits compared to standard 50k genotypes. Higher prediction accuracy can be achieved by using statistical methods which allow putting more weight on informative SNPs in the analysis, e.g. by fitting them as an additional random effect with a separate genomic relationship matrix. The accuracy improvements were less pronounced in single-step OVIS runs because of complexity of analysis and the difficulty to disentangle accuracy improvement from top SNPs and dependency on the information known prior to genotyping, including information from correlated traits.
Benefits to industry
Adding a new set of predictive SNPs to genotyping arrays for Australian sheep is expected to enhance the genomic prediction accuracy of live animal traits, eating quality and reproduction traits and increase the accuracy of ASBV's produced by the Sheep Genetics single step evaluation in OVIS. The increase would be most notably for younger animals without much information on these traits, allowing earlier selection for hard to measure traits such as eating quality and reproduction, and therefore improving rates of genetic gain for these traits. Tail length is related to fly-strike incidence in Australian sheep and selection on tail length would be very useful in controlling fly-strike incidence.
MLA action
MLA continues to deliver data captured as part of the Livestock Genetics Program and implement the recommendations in this report.
Future research
Ongoing work is needed to find predictive genetic variants for quantitative traits and genetic defects, with more data becoming available at an increased rate. With increased uptake of genotyping, data from breeder flocks, commercial animals and abattoirs might become available and work needs to be done to combine such data, which may be of variable quality.
Further predictive SNPs and possibly causal variants affecting variation in economic traits need to be added to genotyping arrays such that genomic prediction accuracy will increase and start approaching the numbers as achieved in the dairy industry (~0.6 to 0.9). Imputed sequence data is used and its imputation accuracy would benefit from further genome sequencing of key animals.
Ongoing work is also needed to implement these predictive SNPs in genetic evaluation, with expected large increases in the number of genotyped animals and increased variability in genotype array data and phenotypic data quality. It would be useful to develop evaluation methods where imputation of all animals to the same common set of SNP genotypes is not a requirement. For genetic defects a more systematic reporting of affected phenotypes is required.
|
For more information Contact Project Manager: Peta Bradley |
