Integrated model for genomic prediction under additive and non-additive genetic architecture

Area of Science:

• Quantitative genetics
• Animal breeding
• Bioinformatics

Background:

• Genomic selection utilizes genome-wide markers for breeding value estimation and phenotypic prediction.
• Parametric models excel with additive genetic architecture but struggle with non-additive effects (dominance, epistasis).
• Nonparametric approaches capture non-additive effects but often neglect additive components.

Purpose of the Study:

• To develop an integrated genomic prediction model combining strengths of parametric and nonparametric approaches.
• To create a robust model capable of handling both additive and non-additive genetic architectures simultaneously.
• To improve the accuracy of breeding value estimation and phenotypic prediction in livestock and crops.

Main Methods:

• Comparative analysis of parametric (e.g., GBLUP) and nonparametric (e.g., SVM) models.
• Development of an integrated model selecting optimal components from each category.
• Assessment of the integrated model using predictive ability and error variance metrics.

Main Results:

• GBLUP demonstrated strong performance under additive genetic architecture.
• Support Vector Machines (SVM) showed promising results for non-additive genetic architecture.
• The integrated model successfully minimized error variance, handling both additive and epistatic effects.

Conclusions:

• An integrated genomic prediction model offers a robust solution for complex genetic architectures.
• This approach enhances the accuracy of estimating breeding values and predicting phenotypes.
• The developed model provides a significant advancement in genomic selection methodologies.