INTEGRATING MULTI-OMICS, GENOME EDITING AND ARTIFICIAL INTELLIGENCE FOR PRECISION BREEDING: ADVANCES IN MOLECULAR GENETICS AND CROP IMPROVEMENT

Abstract

Precision breeding is increasingly shaped by the ability to connect molecular measurements with decisions made in breeding populations and fields. Multi-omics can help explain how inherited variation is expressed through regulatory activity, proteins and metabolites; genome editing can test selected hypotheses and create particular alleles; and artificial intelligence (AI) can support prediction, phenotyping and the prioritization of experiments. These contributions are complementary, but they are not interchangeable. Molecular associations may reflect consequences rather than causes, an edit that produces the expected molecular change may not improve crop performance and a model that performs well on a familiar dataset may fail in another population or environment. This review critically examines how multi-omics, genome editing and AI can be combined for crop improvement, drawing on research in tomato, maize, rice, wheat, soybean and other plant systems. The evidence is strongest when candidate discovery is followed by functional testing and replicated agronomic evaluation. We outline a closed-loop approach in which molecular data generate hypotheses, AI helps rank candidates and select informative tests, editing or crossing evaluates causality and field results are used to update subsequent decisions. Priorities include better causal designs, representative training populations, independent validation, reproducible data workflows, realistic economic assessment and wider access to enabling technologies. The value of precision breeding should ultimately be judged by reliable genetic gain under target production conditions, not by the amount of data collected, the apparent sophistication of a model or editing efficiency alone.