Related Experiment Videos
National evaluation for calving ease, gestation length and birth weight by linear and threshold model methodologies
Deukhwan Lee1, Ignacy Misztal, J Keith Bertrand
1Department of Animal and Dairy Science, University of Georgia, Athens, GA 30605, USA.
Journal of Applied Genetics
|June 25, 2002
Summary
Bayesian threshold models offer accurate calving ease predictions in Gelbvieh cattle but require significant computational resources. Linear models are faster for linear traits, while threshold models are better for calving ease, though computationally intensive.
Area of Science:
- Animal Genetics and Breeding
- Quantitative Genetics
- Statistical Genomics
Background:
- Accurate genetic evaluation for traits like calving ease, gestation length, and birth weight is crucial in beef cattle breeding programs.
- Traditional linear models (LM) are computationally efficient for continuous traits.
- Linear-threshold models (LTM) offer a framework for analyzing categorical traits like calving ease, but their computational demands can be high.
Purpose of the Study:
- To compare the performance of linear models (LM) with Bayesian linear-threshold models (LTM) for analyzing calving ease, gestation length, and birth weight in Gelbvieh cattle.
- To evaluate the accuracy and computational efficiency of empirical Bayes (TMEB) and Monte Carlo (TMMC) methodologies within LTM.
- To assess the suitability of different statistical approaches for genetic parameter estimation in large cattle datasets.
Main Methods:
- Analysis of a large dataset (393,097 calving ease, 129,520 gestation length, 412,484 birth weight records) from Gelbvieh cattle.
- Implementation of mixed linear models (LM) for continuous traits and linear-threshold models (LTM) for calving ease (CE).
- Comparison of solutions from LM with TMEB and TMMC methodologies for LTM, focusing on genetic correlations and computational costs.
Main Results:
- High correlations (0.85-0.86 for direct, 0.75-0.78 for maternal effects) were observed between LM and TMEB for calving ease.
- TMEB and TMMC showed near-perfect agreement (0.98-1.00) for calving ease genetic effects.
- Linear traits showed high correlations (>0.97) between LM and TMEB, but lower correlations (0.91) between LM and TMMC, indicating convergence issues with TMMC. TMMC required significantly longer computing time (6 days) and more memory compared to LM and TMEB.
Conclusions:
- Bayesian implementation of threshold models (TMEB) provides accurate estimates for calving ease, comparable to linear models, while being computationally feasible.
- While TMMC is also accurate, its computational cost is prohibitive for large-scale genetic evaluations.
- TMEB offers a practical approach for analyzing categorical traits like calving ease in large cattle populations, balancing accuracy and computational efficiency.