1a. Objectives (from AD-416):
Use the BovineSNP50 assay to provide high-accuracy predictions of genetic merit to U.S. beef breeds; enable the adoption of whole genome enabled animal selection (WGEAS) by developing low-density and low-cost SNP assays for: intermediate-accuracy genetic prediction, mate selection, and parentage verification and traceability; develop, adapt and optimize statistical methodologies to: fully integrate SNP genotype or haplotype effects into existing genetic evaluation technologies, and supplement or replace pedigree data; and collaborate and coordinate U.S. and European Union WGEAS activities.
1b. Approach (from AD-416):
Genetic prediction using high-density SNP data will be implemented using MTDFREML. Implementation of more sophisticated strategies will follow using the MTGSAM programs that will be modified to accommodate extensions to the prediction model. Collaboration with a biotechnology company to develop a 384-SNP assay that is expected to dramatically decrease genotyping costs and increase sample throughput. A machine learning approach using a two-step feature subset selection algorithm will be evaluated for SNP selection for this assay. Develop BLUP approaches for the prediction of genetic merit in non-pedigreed populations using molecular relationship matrices. We shall manage this coordination and collaboration via e-mail and teleconference calls, however, we shall also meet at least annually in conjunction with the PAG or ISAG meetings alternating between the U.S. and Europe to coordinate activities.
3. Progress Report:
Award money was disbursed to the Beltsville Area. Award money was, in turn, disbursed to collaborators, including University of Nebraska. An objective for the beef project is the integration of Molecular Breeding Values (MBVs) into the genetic evaluations for Expected Progeny Differences (EPDs). The beef industry scenario differs greatly from the dairy in that due to intellectual property issues surrounding the Single Nucleotide Polymorphism (SNP) identity for commercially available DNA tests, there is a need to integrate the MBV itself rather than SNP genotypes or other strategies requiring genotypic information. We developed methods for this integration for a within-breed evaluation for multiple traits (both linear and threshold traits) including maternal effects. We also developed strategies to incorporate multiple MBVs (evolving panels or from different commercialization companies) and allow for a MBV with variable accuracy (which can result when imputing from a lower-density to a higher-density panel) in a computationally efficient manner. The impact of Bayes-based estimates of individual animal MBV reliabilities on genetic parameters was evaluated and alternative methods for incorporating individual animal reliabilities in genetic evaluations were compared. Methodology for the estimation of SNP effects for binary traits was developed to handle selective genotyping based on an animal phenotype. The method for multiple traits and multiple MBVs within a single breed was adopted by Angus Genetics Inc. (AGI) for evaluations of carcass traits. This research supports the related in-house project to use genotypic data and resulting bovine haplotype map to enhance genetic improvement in dairy cattle through development and implementation of whole genome selection and enhanced parentage verification approaches (Objective 2).