Estimation of genetic covariances with method R

Area of Science:

• Quantitative genetics
• Statistical genetics
• Animal breeding

Background:

• Method R is a computationally inexpensive approach for estimating variance and covariance.
• Accurate estimation of (co)variance components is crucial in quantitative genetics and animal breeding.

Purpose of the Study:

• To investigate the properties of Method R for estimating (co)variance components.
• To focus on the specific application of Method R in covariance estimation.
• To develop and evaluate alternative algorithms for Method R implementation.

Main Methods:

• Development of theoretical Method R formulas for univariate and bivariate models.
• Analysis of regression curves for Method R in single-trait and covariance estimation.
• Implementation and testing of an alternative algorithm using a transformation matrix.

Main Results:

• Method R regression curves were continuous and monotonic for single-trait models, influenced by animal information and variance ratios.
• Covariance estimation using Method R resulted in a monotonic but discontinuous regression curve.
• An alternative algorithm based on a transformation matrix demonstrated reliable convergence for all tested models, including those with covariances.

Conclusions:

• Method R can be adapted for covariance estimation, overcoming limitations of its standard implementation.
• The alternative algorithm enables reliable covariance estimation with Method R, even in large-scale models.
• Method R, with modifications, provides a viable and efficient tool for estimating covariances in complex genetic models.