Statistics and data science unlock the predictive power of quantitative genetics

Area of Science:

• Plant breeding and genetics
• Statistical genetics
• Data science in agriculture

Background:

• Plant breeding is shifting from empirical methods to a predictive science driven by large datasets (genomic, phenotypic, environmental, multi-omics).
• Quantitative genetics provides the theoretical basis, but modern implementation requires advanced statistical and data science methodologies for actionable insights.

Purpose of the Study:

• To review how statistics and data science enhance the predictive power of quantitative genetics in plant breeding.
• To highlight the integration of diverse data types and methodologies for improved breeding decisions.
• To identify challenges and future directions for data-driven plant breeding.

Main Methods:

• Genomic selection
• Genome-wide association studies (GWAS)
• Mixed models
• Multivariate analysis
• Machine learning (ML)
• Data integration
• Uncertainty quantification

Main Results:

• Statistics and data science operationalize and extend quantitative genetics principles.
• These tools form an interconnected pipeline supporting various breeding decisions (parent selection, genotype advancement, etc.).
• Key challenges include 'large p, small n', population structure, G×E interactions, and data integration.

Conclusions:

• Statistics and data science unlock the predictive potential of quantitative genetics for efficient, sustainable, and decision-oriented plant breeding.
• Future frameworks must be integrated, interpretable, probabilistic, and scalable, combining diverse data while aligning with biological meaning and breeding goals.