Machine learning algorithms translate big data into predictive breeding accuracy.

Area of Science:

• Plant breeding
• Genomics
• Machine learning

Background:

• Extensive genomic, phenotypic, and environmental data are crucial for modern plant breeding.
• Statistical machine learning (ML) algorithms can identify relevant features and build robust predictive models.
• Understanding genotype-by-environment (G×E) interactions is key to improving crop performance in diverse conditions.

Purpose of the Study:

• To review the transformative impact of big data and ML on genomic-enabled prediction in plant breeding.
• To discuss how these technologies enhance prediction accuracy and understanding of G×E interactions.
• To highlight the optimization of breeding strategies through analysis of large, diverse datasets.

Main Methods:

• Leveraging historical breeding data for ML analysis.
• Utilizing cross-validation techniques for model robustness and reliability.
• Analyzing multi-trait genomics, phenomics, and environmental covariables.

Main Results:

• ML algorithms automatically identify key genomic and environmental features.
• Enhanced prediction accuracy for new plant lines.
• Deeper insights into genetic factors influencing performance across environments.

Conclusions:

• Big data and ML are revolutionizing plant breeding by improving prediction accuracy.
• ML facilitates a better understanding of genotype-by-environment interactions.
• These approaches optimize breeding strategies through automated analysis of extensive datasets.