Multi-trait/environment sparse genomic prediction using the SFSI R-package

Area of Science:

• Quantitative genetics
• Plant breeding
• Bioinformatics

Background:

• Sparse selection indices (SSIs) predict genetic merit using high-dimensional phenotypes.
• Sparse genomic prediction (SGP) predicts genetic merit using subsets of training data.

Purpose of the Study:

• Introduce a novel framework for multi-trait/environment sparse genomic prediction (MT-SGP).
• Combine features of SSI and SGP into a unified model for enhanced genetic merit prediction.

Main Methods:

• Developed an MT-SGP framework integrating SSI and SGP methodologies.
• Created an R-package (sparse family and selection index) for solving SSI, SGP, and MT-SGP problems.
• Conducted extensive benchmarks using three crop datasets across 30 traits/environments.

Main Results:

• MT-SGP demonstrated comparable or superior prediction accuracy (up to 15% gain) compared to MT-genomic best linear unbiased prediction.
• Identified conditions (sample size, genetic correlation, heritability) favoring MT-SGP's performance gains.
• The R-package provides practical tools for implementing these advanced prediction methods.

Conclusions:

• MT-SGP offers a powerful alternative for genetic merit prediction in complex breeding programs.
• The framework effectively borrows information from correlated traits and genetically similar individuals.
• The developed R-package facilitates the application of MT-SGP in quantitative genetics research and breeding practice.