Genomic selection for genotype performance and stability using information on multiple traits and multiple...

Area of Science:

• Plant breeding and genetics
• Statistical modeling in agriculture
• Genomic data analysis

Background:

• Genomic selection (GS) is crucial for efficient plant breeding.
• Analyzing multi-environment trial (MET) data is complex, especially with multiple traits.
• Existing methods often struggle to fully utilize genotype by environment by trait interaction (GETI).

Purpose of the Study:

• To develop a single-stage genomic selection approach incorporating multiple traits and environments.
• To extend the factor analytic linear mixed model (FA-LMM) for GS in complex datasets.
• To enable breeders to leverage GETI for enhanced selection accuracy.

Main Methods:

• Developed a partially separable factor analytic linear mixed model (SFA-LMM).
• The SFA-LMM uses a three-way separable structure: factor analytic matrices for traits and environments, and a genomic relationship matrix.
• Incorporated a diagonal matrix to model specific GEI and GTI patterns for each trait/environment.

Main Results:

• The SFA-LMM demonstrated a better fit compared to simpler separable models.
• It achieved a comparable fit to more complex non-separable and partially separable approaches.
• The SFA-LMM includes fewer parameters, making it more efficient for large-scale genomic datasets.

Conclusions:

• The SFA-LMM provides an informative framework for utilizing GETI in genomic selection.
• This approach enhances selection accuracy across correlated traits and environments.
• The developed model supports simultaneous selection for performance and stability, advancing plant breeding analyses.