Increased accuracy of artificial selection by using the realized relationship matrix.

Area of Science:

• Animal Breeding and Genetics
• Quantitative Genetics
• Genomic Prediction

Background:

• Pedigree-based relationship matrices in BLUP (Best Linear Unbiased Prediction) rely on average relationships.
• Dense marker genotypes provide realized genomic relationships, reflecting actual identity by descent (IBD).
• Genomic selection (GS) uses marker effects to predict breeding values.

Purpose of the Study:

• To demonstrate the increased accuracy of breeding values by replacing pedigree-based relationships with realized genomic relationships in BLUP.
• To show the equivalence between this BLUP approach and standard genomic selection.
• To develop deterministic predictions for the accuracy of genomic breeding values (GEBVs).

Main Methods:

• Construction of a realized relationship matrix from dense marker genotypes.
• Implementation of BLUP using the realized relationship matrix instead of the pedigree-based one.
• Development of deterministic prediction equations for GEBV accuracy based on effective loci, family structure, and population size.

Main Results:

• Replacing the average relationship matrix with the realized relationship matrix in BLUP significantly enhances breeding value prediction accuracy, particularly for individuals lacking phenotypes.
• The BLUP method using realized relationships is mathematically equivalent to genomic selection assuming normally distributed QTL effects.
• Deterministic predictions accurately estimate GEBV accuracy, showing it can approach unity with sufficient genotyped and phenotyped relatives.

Conclusions:

• Utilizing realized genomic relationships in BLUP offers a powerful and accurate method for predicting breeding values, especially in the absence of individual phenotypes.
• The developed deterministic prediction framework provides valuable insights into the factors influencing GEBV accuracy.
• Sufficiently large numbers of genotyped and phenotyped relatives are crucial for achieving high accuracy in genomic prediction within populations.