A groundbreaking deep-learning framework named DeepWheat has been developed by a Chinese research team to enhance the intelligent design of crop varieties, particularly in wheat breeding. Published in Genome Biology by the wheat gene resource innovation team at the Chinese Academy of Agricultural Sciences, this model accurately predicts gene expression across tissues and cultivars. Given wheat’s complexity, with its intricate genome structure significantly larger than rice and human genomes, DeepWheat’s dual-model system identifies how specific regulatory variations influence gene expression and enables precise forecasting of tissue-specific patterns.
Lu Zefu, the chief scientist, emphasizes DeepWheat’s potential to resolve challenges like trait antagonism, allowing researchers to fine-tune genetic modifications without labor-intensive trial and error. This model not only facilitates targeted genome editing but is also applicable to crops like rice and maize, streamlining the breeding of high-performance varieties. Ultimately, DeepWheat represents a significant advancement in agricultural biotechnology and crop genetics.
