Translating functional molecular knowledge into crop-breeding success.

Sequence-based deep learning predicts genetic variant effects at base-pair resolution, enabling precision breeding in crops and overcoming evolutionary constraints.

• Why it matters: Traditional plant breeding is limited by evolutionary barriers and slow phenotypic improvements, creating a need for more targeted and efficient methods to enhance crop traits.
• What they did: The study employs high-quality genome sequence data and deep learning models to predict the impact of genetic variants, linking these predictions to important agronomic traits for crop improvement.
• The result: This approach allows breeders to prioritize variants for selection or editing, paving the way for gene introgression, removal of deleterious mutations, and designing new plant ideotypes to meet future environmental challenges.