Multitrait genomic prediction method in approximate genome-based kernel model

Area of Science:

• Plant breeding
• Genomics
• Quantitative genetics

Background:

• Simultaneous selection for multiple traits is crucial for developing superior crop varieties.
• Existing genomic prediction methods face computational challenges with large datasets and multiple traits.

Purpose of the Study:

• To develop an efficient, large-scale multi-trait genomic prediction method.
• To evaluate the performance of the new method against existing approaches in terms of predictive accuracy and computational efficiency.

Main Methods:

• Development of the approximate genome-based kernel model for multi-trait genomic prediction (MT-RHPK).
• Simulation studies to compare MT-RHPK with multi-trait genomic best linear unbiased prediction (MT-GBLUP) and single-trait genomic best linear unbiased prediction (ST-GBLUP).
• Validation using 14 paired traits from bread wheat and rice datasets.

Main Results:

• MT-RHPK demonstrated significantly faster computational time than MT-GBLUP with comparable or better predictive accuracy.
• MT-RHPK showed improved predictive accuracy for low-heritability traits when genetic correlations were positive.
• Performance varied based on heritability and genetic correlation between traits, with MT-RHPK generally performing well across diverse scenarios.

Conclusions:

• MT-RHPK is a practical and computationally efficient tool for large-scale, multi-trait genomic prediction.
• The method facilitates the simultaneous improvement of multiple traits in crop breeding programs.
• MT-RHPK offers advantages in speed and accuracy, particularly for complex trait selection.