Differences among methods to validate genomic evaluations for dairy cattle.

Area of Science:

• Animal Genetics and Breeding
• Genomic Evaluation Methods
• Dairy Cattle Production

Background:

• Genomic evaluations are crucial for dairy cattle breeding, but their validation requires careful methodology.
• Previous validation methods may not fully account for the complexities of genomic data and selection.
• Accurate validation is essential for reliable genetic improvement in dairy populations.

Purpose of the Study:

• To investigate and compare two distinct methods for testing the predictive accuracy of genomic evaluations in dairy cattle.
• To assess the impact of different validation approaches on the interpretation of genomic evaluation results.
• To determine the optimal method for validating genomic predictions, particularly for first-crop bulls.

Main Methods:

• Utilized USDA genetic evaluation data from August 2006 and April 2010 for Holstein, Jersey, and Brown Swiss cattle.
• Trained models using predicted transmitting abilities (PTA) from August 2006 (Method 1) and April 2010 (Method 2).
• Validated predictions using deregressed PTA of unproven bulls that became proven by April 2010, analyzing bias, regression coefficients, and coefficients of determination.

Main Results:

• Genomic information generally enhanced predictive ability across breeds and traits.
• Method 2 yielded higher coefficients of determination but was influenced by data non-independence.
• Method 1 allowed for accurate evaluation of increased accuracy from genomic information in first-crop bulls.

Conclusions:

• Both validation methods revealed some bias in prediction equations across breeds.
• Method 1 is advised for validating genomic evaluations due to its ability to provide unbiased assessment, especially for first-crop bulls.
• The choice of validation method significantly impacts the interpretation of genomic evaluation performance.