BIOINFORMATICS TO IMPLEMENT GENOMIC SELECTION (BIGS)
Genetics, Breeding, & Animal Health
2012 Annual Report
1a.Objectives (from AD-416):
This research and development project will develop analytical software for Bayesian analysis of genomic information and deliver it within an integrated bioinformatics infrastructure that will enable genomic evaluation using high-throughput SNP genotyping technology in livestock. We will implement these methodologies across a range of economic traits in beef cattle, first using existing genomic and phenotypic records from the U.S. Meat Animal Resarch Center and ultimately enabling researchers to submit data sets for analysis via a web interface. Collectively, we will deliver immediate and on-going opportunities in beef cattle for livestock managers, public researchers and industry geneticists to exploit genomic evaluation.
1b.Approach (from AD-416):
We will enhance existing methods for predicting missing genotypes to applications involving haplotypes and web-enable this software. We will extend theory-based single-trait Bayesian methods for additive effects to practical circumstances that can estimate allelic or haplotypic effects in the context of multiple traits, maternally influenced traits, and categorically expressed traits, to capitalize on both additive and non-additive effects. All these analytical tools will be made available via web interface. Our activities will be motivated and focused on genomic evaluation of beef cattle. Implementation of this research will fully support goals of the 2005-2010 USDA Strategic Plan, by enhancing international competitiveness of the U.S beef cattle and other economically-relevant agricultural industries through practical access to the latest advances in molecular genetic technology.
Statistical Bayesian analyses of U.S. Meat Animal Research Center (USMARC) crossbred genotypes and phenotypes with GenSel software were used to complement genomic best linear unbiased prediction based analytical approaches. Genomic predictions utilizing direct and imputed high density (770,000 markers) genotyping panels trained by USMARC data were provided to cooperators in the Weight Trait Project. Development of a platform-independent version of GenoProb software, which uses complex pedigrees with incomplete marker data to compute probabilities of complete marker genotypes, was abandoned due to a lack of suitable open-source replacements for all of the proprietary software dependencies in GenoProb. While some functions were replaced by open-source equivalents, we determined that eliminating all proprietary dependencies would require substantial re-writing of the existing GenoProb code. Alternatives providing similar functionality to GenoProb, which may be more suitable for incorporation into the project's GenSel system, are being explored.