BLUPbased

BLUP-based refers to methods built on Best Linear Unbiased Prediction, a statistical framework used to estimate the genetic merit of individuals from phenotype data while accounting for relatedness and environmental effects. In animal and plant breeding, BLUP-based evaluations predict breeding values that guide selection decisions.

These methods rely on linear mixed models that decompose observed phenotypes into fixed effects (such as sex or management) and random genetic effects plus residual error. The BLUP estimator arises from solving Henderson's mixed model equations, providing predictions that are linear in the observed data and unbiased under correct model assumptions. Variance components are typically estimated with restricted maximum likelihood, and the genetic relationships among individuals are captured by relationship matrices such as the pedigree-based A matrix or the genomic-based G matrix.

BLUP-based approaches include traditional BLUP (or EBV when variances are unknown) and genomic BLUP (G-BLUP), which integrates dense marker data to define genomic relationships. They can handle unbalanced data and incomplete pedigrees and are widely used to derive estimated breeding values for selection in livestock and crop improvement programs.

Strengths of BLUP-based methods include their ability to use information from related individuals, shrinkage of predictions toward the population mean, and robustness to moderate data imperfections. Limitations involve reliance on correct model specification, assumptions of normality and linearity of effects, potential biases from non-genetic structuring, and computational demands for large datasets. As genomic data become more prevalent, BLUP-based methodologies have evolved to incorporate genomic relationships and marker information in G-BLUP and related frameworks.