Bayesian discrete lognormal regression model for genomic prediction.

Area of Science:

• Agricultural Science
• Genetics
• Statistical Modeling

Background:

• Genomic selection (GS) is vital for modern breeding, predicting individual performance using genomic data.
• Existing GS models primarily address continuous traits, with limited options for count data.
• Accurate models are crucial for efficient selection of desirable traits in crops and livestock.

Purpose of the Study:

• To introduce a novel Bayesian discrete lognormal regression model for genomic prediction of count traits.
• To evaluate the proposed model's performance against traditional Gaussian and lognormal models.
• To enhance the accuracy and applicability of genomic selection for non-normally distributed traits.

Main Methods:

• Development of a Bayesian discrete lognormal regression model.
• Utilizing a Gibbs sampler for posterior distribution exploration and prediction.
• Application and evaluation on two wheat disease resistance count datasets.

Main Results:

• The proposed Bayesian discrete lognormal model demonstrated competitive performance.
• The model proved effective in predicting count genomic traits.
• Results suggest it is a natural fit for count-based genomic data.

Conclusions:

• The novel Bayesian discrete lognormal model is a viable and effective tool for genomic prediction of count traits.
• This model enhances the capabilities of genomic selection in breeding programs.
• It provides a more appropriate statistical framework for non-normally distributed count data in genetics.