Semi-parametric genomic-enabled prediction of genetic values using reproducing kernel Hilbert spaces methods.

Area of Science:

• Quantitative genetics
• Genomic prediction
• Statistical modeling

Background:

• Genetic value prediction traditionally uses phenotypic and pedigree data.
• Dense molecular markers offer enhanced prediction but pose statistical and computational challenges.
• Multi-factorial traits and high-dimensional genomic data complicate traditional modeling approaches.

Purpose of the Study:

• To explore the application of reproducing kernel Hilbert spaces (RKHS) regressions for genetic value prediction.
• To address challenges in incorporating dense molecular marker data into quantitative genetics models.
• To evaluate RKHS regression for genomic prediction in wheat.

Main Methods:

• Reproducing kernel Hilbert spaces regressions were employed.
• Kernel selection algorithms were discussed and applied.
• Methods were assessed using grain yield data from 599 wheat lines across four environments.

Main Results:

• RKHS regressions provide a flexible framework for incorporating diverse data types, including dense molecular markers.
• The methodology offers computational advantages over many parametric models.
• The proposed methods demonstrated effectiveness in predicting genetic values for a complex trait like grain yield.

Conclusions:

• RKHS regressions are a promising approach for enhancing genomic prediction accuracy.
• This methodology effectively handles the complexity and dimensionality inherent in genomic data.
• The study validates RKHS regression for practical applications in plant breeding and quantitative genetics.