GIBBSTHUR

GIBBSTHUR implements a Thurstonian model to estimate variance and covariance components and predict breeding values from ranking traits by modeling underlying Gaussian liabilities in equine populations and other continuous or threshold traits.

Key Features:

• MCMC Techniques: Utilizes Monte Carlo Markov Chain methods to obtain posterior estimates and predictions under the Thurstonian framework.
• Data Augmentation: Incorporates a data-augmentation step to sample latent liabilities associated with ranking traits.
• Thurstonian Latent Gaussian Liability Modeling: Models an underlying Gaussian liability that determines observed rankings in competitive events.
• Variance and Covariance Estimation: Provides estimates of variance and covariance components for genetic evaluation.
• Breeding Value Prediction: Produces predictions of breeding values for selection decisions.
• Event Effect and Additive Genetic Competitor Analysis: Estimates event effects and average additive genetic effects of competitors in events.
• Trait Type Support: Supports analysis of ranking traits as well as other continuous and threshold traits.

Scientific Applications:

• Equine breeding evaluation: Applied to breeding evaluation and selection in equine populations using ranking traits from competitions.
• Genetic parameter estimation: Estimation of variance and covariance components for genetic studies and breeding programs.
• Breeding value prediction and selection: Prediction of breeding values to inform selection decisions for desirable genetic traits.
• Competitive event analysis: Quantification of event effects and competitors' average additive genetic effects within competitions.
• Method validation with simulations: Recovery of simulated variance and covariance components for validation of model performance.

Methodology:

GIBBSTHUR employs a Gibbs Sampler scheme integrated with a data-augmentation step to implement the Thurstonian model by sampling latent Gaussian liabilities for estimation of variance and covariance components and prediction of breeding values, and it supports continuous and threshold traits.