Expanding genomic prediction in plant breeding: harnessing big data, machine learning, and advanced software

Area of Science:

• Agricultural Science
• Genetics
• Computer Science

Background:

• Genomic selection (GS) is increasingly vital for enhancing genetic gains in plant breeding.
• The efficiency of GS relies on sophisticated statistical and computational methods.

Purpose of the Study:

• To review statistical machine learning (ML) methods and software that enhance genomic selection (GS) efficiency.
• To explore how ML democratizes genomic prediction (GP) methodologies in plant breeding.

Main Methods:

• Review of statistical ML tools for genomic-enabled prediction.
• Examination of ML approaches for diverse phenotype predictions (continuous, binary, categorical, count).
• Highlighting deep learning (DL) models in genomic prediction (GP).

Main Results:

• Statistical ML tools significantly enhance GS efficiency, especially with big data.
• Various ML tools are available for predicting different types of plant traits.
• Deep learning models offer unique advantages for genomic prediction.

Conclusions:

• Machine learning and accessible software are key to democratizing genomic selection in plant breeding.
• ML advancements improve the prediction of diverse plant traits, accelerating genetic gains.
• Data management tools are crucial for adopting advanced GS methodologies.