Genomic Prediction Including SNP-Specific Variance Predictors.

Area of Science:

• Quantitative genetics
• Bioinformatics
• Animal breeding

Background:

• Genomic selection (GS) aims to improve livestock breeding through genetic predictions.
• Integrating external marker information can enhance GS model accuracy.
• Existing GS models may not fully leverage diverse biological data.

Purpose of the Study:

• To propose a general genomic selection model incorporating external marker information using a hierarchical generalized linear model (hglm) framework.
• To develop an efficient R package (CodataGS) for fitting these models.
• To evaluate the model's performance against established methods like SNP-BLUP.

Main Methods:

• Developed a novel hglm-based model with a link function to integrate external marker data.
• Utilized simulated data to test categorical, continuous, and combined external information models.
• Assessed model performance using a dairy cattle dataset for Milk Yield and Fat Percentage traits.

Main Results:

• The proposed models achieved accuracy improvements of 3.8% to 23.2% over SNP-BLUP in simulated populations.
• The categorical model demonstrated improved accuracy for dairy cattle traits, particularly Fat Percentage.
• Model performance gains were influenced by trait genetic architecture and the uncertainty of external information.

Conclusions:

• The novel hglm-based genomic selection model effectively integrates external marker information, leading to improved prediction accuracies.
• The CodataGS R package provides an efficient tool for implementing these advanced GS models.
• Future advancements in marker information will further enhance the utility of these models in animal breeding programs.