Technical note: Genetic principal component models for multitrait single-step genomic evaluation.

Area of Science:

• Animal genetics
• Quantitative genetics
• Statistical genetics

Background:

• Multivariate linear mixed models are computationally intensive for genetic evaluation, particularly with genomic data.
• Genomic relationship matrices often have dense inverse structures, increasing computational burden.

Purpose of the Study:

• To describe a reparameterization of the multivariate linear mixed model using principal components.
• To reduce computational requirements for genetic evaluation, especially in genomic contexts.

Main Methods:

• Reparameterization of the multivariate linear mixed model to principal components.
• Utilizing a sparser coefficient matrix in the mixed model equations.
• Incorporating dimension reduction by omitting principal components with negligible eigenvalues.

Main Results:

• The reparameterization yields an equivalent model with reduced computational needs.
• This approach is particularly beneficial for single-step genomic evaluation with large genomic relationship matrices.
• Dimension reduction further enhances computational savings.

Conclusions:

• The principal component reparameterization offers a computationally efficient alternative for genetic evaluation.
• This method has significant implications for handling large-scale genomic datasets in animal breeding.
• The framework facilitates substantial computational savings in genomic prediction models.