Dissection of genomic correlation matrices of US Holsteins using multivariate factor analysis

Area of Science:

• Animal Genetics
• Quantitative Genetics
• Genomic Prediction

Background:

• Understanding genetic correlations between traits is crucial for effective breeding programs.
• Genomic selection utilizes genome-wide marker data for predicting breeding values.
• Identifying chromosomal regions influencing multiple traits can refine selection strategies.

Purpose of the Study:

• To compare correlation matrices of direct genomic predictions at the whole genome and chromosomal levels.
• To investigate the genetic architecture of 31 traits in US Holstein bulls.
• To identify specific chromosomes exhibiting distinct covariance structures.

Main Methods:

• Multivariate factor analysis applied to genomic predictions for 31 traits.
• Comparison of correlation matrices at the genome-wide and individual chromosome levels.
• Analysis of covariance structures on specific bovine autosomes (BTA).

Main Results:

• Seven major factors influencing conformation, longevity, yield, and composition traits were identified at the genome level.
• Significant variations in covariance structures were observed on BTA 6, 14, 18, and 20.
• Evidence of segregating quantitative trait loci (QTL) affecting trait groups was found on specific chromosomes, including DGAT1 on BTA 14 and a QTL for calving ease on BTA 18.
• A candidate gene for daughter pregnancy rate was suggested on BTA 28.

Conclusions:

• Chromosomal-level analysis reveals genetic covariances not apparent at the whole-genome level.
• This approach can pinpoint chromosomes with unique covariance structures, potentially harboring QTL with smaller effects.
• The methodology aids in understanding the genetic basis of complex traits and optimizing genomic selection strategies.