Chinese scientists revolutionise precision grape breeding with artificial intelligence

Chinese researchers have unveiled a groundbreaking AI-based grape breeding method that drastically cuts the time required to develop new grape varieties. The scientific team has achieved a prediction accuracy of 85 per cent in identifying promising grape traits, allowing breeding efficiency to soar to four times that of traditional methods. This is reported by CGTN, a partner of TV BRICS.

This innovation could transform grape breeding by enabling precise design and accelerated development of superior varieties. This breakthrough holds the potential to reshape not only grape cultivation but also the breeding of other perennial crops.

The project builds on years of research, including the first complete grape genome map released in 2023 and the subsequent Grapepan v1.0 genome, which offer new genetic data for breeding purposes. By incorporating machine learning, the team has developed predictive models that allow breeders to analyse genetic potential early in the growth cycle, making it possible to filter out less promising seedlings early on and streamline resources toward more viable plants.

Since 2015, the team has analysed over 400 representative grape varieties, examining 29 traits such as berry size, cluster structure, and skin color. Their efforts resulted in comprehensive genetic and trait maps that form the basis of their predictive AI model, enhancing breeding strategy and reducing costs.

Photo: iStock