generative-ai
Generative AI refers to machine learning systems capable of producing new content—such as sequences, structures, or predictions—by learning patterns from large datasets. In plant science, these tools are being applied to accelerate discovery by generating novel protein or genome sequences, predicting gene function, and modeling complex biological interactions that would be difficult to study experimentally. This opens new avenues for crop improvement, stress tolerance research, and understanding the molecular underpinnings of plant development.
open_in_new WikipediaPubMed · 2026-04-13
Scientists used a generative AI system called TargetGAN to invent artificial gene switches for plants that are far more powerful than anything found in nature. The best synthetic switch boosted gene activity 128-fold compared to a standard benchmark, opening new possibilities for precisely engineering crop traits.
29 out of 2,909 validated AI-generated synthetic promoters exceeded the activity of all tested natural promoters, proving synthetic designs can transcend natural limits.
The top synthetic candidate, SP1482, achieved 128-fold higher gene expression than the widely used 35S minimal promoter benchmark.
The TargetGAN framework — trained on 76,851 natural promoters — can generate promoters targeting user-specified activity levels, with moderate predictive accuracy (correlation of 0.64 between predicted and measured activity).