Breeding value reliabilities for multiple-trait single-step genomic best linear unbiased predictor

Area of Science:

• Animal Breeding and Genetics
• Quantitative Genetics
• Genomic Evaluation

Background:

• Calculating reliabilities for estimated breeding values is crucial for accurate genetic evaluations.
• Traditional methods for single-step genomic best linear unbiased prediction (ssGBLUP) can be computationally intensive for large datasets.
• Accurate reliability estimates are essential for effective selection in livestock breeding programs.

Purpose of the Study:

• To develop and validate approximate multistep methods for calculating reliabilities in single-trait (ST-R²A) and multitrait (MT-R²A) ssGBLUP models.
• To assess the accuracy and computational efficiency of these approximation methods compared to exact calculations.
• To enable reliable genetic evaluations in large-scale animal populations.

Main Methods:

• Developed approximate multistep methods (ST-R²A and MT-R²A) to estimate reliabilities for ssGBLUP.
• Utilized a traditional animal model to estimate nongenomic information for genotyped animals.
• Integrated genomic data and estimated nongenomic information into a genomic BLUP model to approximate total information and reliabilities.
• Accounted for genomic data's increased information in nongenotyped animals via pseudo-record counts.

Main Results:

• The approximate methods (ST-R²A and MT-R²A) showed high correlations with exact ssGBLUP reliabilities (e.g., >0.95 for ST-R²A and >0.99 for MT-R²A).
• Regression coefficients indicated a strong linear relationship between approximate and exact reliabilities.
• The approximation method reduced computing time to approximately 12% of the direct exact approach.
• The methods were validated on both a small dataset (Finnish Red dairy cows) and a large Nordic Holstein dataset.

Conclusions:

• The developed approximate methods provide accurate and computationally efficient ways to calculate breeding value reliabilities in ssGBLUP models.
• These methods are suitable for large-scale genetic evaluations, overcoming computational limitations of exact methods.
• The approach facilitates more widespread and accurate genomic selection in dairy cattle populations.