Prediction and association mapping of agronomic traits in maize using multiple omic data

Area of Science:

• Plant breeding and genetics
• Genomics
• Bioinformatics

Background:

• Genomic selection (GS) accelerates plant breeding by enabling early selection before phenotypes are measured, offering advantages over marker-assisted selection for polygenic traits.
• Metabolome and transcriptome data are emerging as valuable sources for enhancing phenotype prediction in plant breeding.
• Maize serves as a model organism for investigating the utility of diverse omics data in breeding programs.

Purpose of the Study:

• To compare the predictive abilities of genomic, transcriptomic, and metabolomic data sources for phenotype prediction in maize.
• To evaluate the performance of eight different statistical methods across these omics data and six agronomic traits.
• To assess the efficacy of genome-wide association study (GWAS) methods, including a novel modified LASSO, for identifying genetic variants influencing complex traits.

Main Methods:

• Utilized maize data to compare predictive abilities of genomic, transcriptomic, and metabolomic data using eight statistical prediction methods.
• Performed genome-wide association studies (GWAS), transcriptome-wide association studies (TWAS), and metabolome-wide association studies (MWAS).
• Employed the genome-wide efficient mixed model association (GEMMA) method and a modified least absolute shrinkage and selection operator (LASSO) method for association analyses.

Main Results:

• Best linear unbiased prediction (BLUP) demonstrated superior performance across traits and omics data.
• Genomic prediction exhibited higher accuracy compared to transcriptomic and metabolomic predictions.
• The modified LASSO method outperformed GEMMA in GWAS, showing higher power and lower Type 1 error rates in simulation studies.

Conclusions:

• Genomic data provides the most predictive power for plant breeding applications compared to transcriptomic and metabolomic data.
• The modified LASSO method represents a significant advancement for association studies, offering improved statistical power and accuracy.
• Integrating diverse omics data and advanced statistical methods can accelerate genetic gains in crop improvement programs.