Guangming Daily, Beijing, November 6: Reporter Yang Shu learned from the Shenzhen Institute of Agricultural Genomics of the Chinese Academy of Agricultural Sciences (Shenzhen Branch of Lingnan Modern Agricultural Science and Technology Guangdong Laboratory) that Zhou Yongfeng’s team from the institute proposed a method of using artificial intelligence to conduct grape cultivation. Compared with traditional methods, the new breeding method can increase the breeding efficiency by 4 times and greatly shorten the grape breeding cycle. This research is expected to achieve precise design and breeding of grapes, accelerate grape variety innovation, and provide methodological reference for the breeding of other perennial crops. Relevant research results were published in the international journal "Nature Genetics" on November 4.
Zhou Yongfeng, a researcher at the Shenzhen Institute of Agricultural Genomics of the Chinese Academy of Agricultural Sciences, said that grapes are a perennial crop. It takes about three years to plant a grape seed from germination to fruiting. But if you want to cultivate good grape varieties, it will take even longer. At present, the main method of choice in the breeding community is still cross-breeding. This method often requires decades of screening and is extremely workload-intensive. Moreover, due to the highly complex grape genome, the hybrid effect of the offspring is often not ideal after cross-breeding.
Since the 21st century, breeders have proposed molecular breeding, which analyzes and predicts based on massive genome genetic variation data to improve breeding efficiency and accuracy. Among them, obtaining comprehensive and accurate crop genome data is key.
Zhou Yongfeng's team began focusing on grape design breeding in 2015, and released the first complete reference genome map of grapes in 2023. Subsequently, the team continued sequencing and assembly, and constructed the first most comprehensive and accurate grape pan-genome to date.
In order to further clarify the relationship between grape genes and traits, Zhou Yongfeng's team selected more than 400 representative grape varieties from nearly 10,000 grape varieties, and conducted 29 tests for three consecutive years, including ear size, skin color, etc. Agronomic traits were investigated and a grape genotype map and trait map were constructed. On this basis, Zhou Yongfeng's team used quantitative genetic analysis to identify 148 gene loci significantly related to agronomic traits, of which 122 loci were discovered for the first time.
Faced with the above grape genome and trait data, Zhou Yongfeng's team introduced machine learning technology in artificial intelligence to analyze the complex network relationship between genotype and trait data, and built the first grape genome-wide selection model. Compared with cross-breeding, which requires judgment based on the phenotype of grapes after maturity, this whole-genome sequencing breeding model can use computer software to predict the traits of grapes after maturity during the seedling stage. The results show that the prediction accuracy of the multigene score prediction model that combines structural variation information is as high as 85%.
Through this model, breeding experts can quickly and accurately assess the genetic potential of large quantities of grape breeding materials to better select superior varieties. At the same time, seedlings that do not meet the conditions are eliminated as early as possible, reducing unnecessary cost investment and greatly improving the efficiency of grape breeding. At present, relevant research results have been applied for and approved 6 national invention patents and 1 international patent has been applied for.