关键词: 水稻, 智能育种, 数据库, 人工智能, 多组学数据, 表型预测, 基因挖掘

Abstract:

Traditional rice breeding is characterized by long cycles, low selection efficiency and experience-dependent decision-making, which can hardly meet the demands for efficient and precise breeding in the modern seed industry. The explosive growth of multi-omics data and the rise of artificial intelligence (AI) have opened up a new avenue for intelligent rice breeding. To systematically integrate database resources and AI technologies and establish a data-driven intelligent rice breeding and selection system, this paper reviews the functional characteristics and typical platforms of four major categories of rice databases, including rice genome, transcriptional regulation, gene interaction networks and germplasm resource pedigrees, with representative platforms covering RiceVarMap, Ribo-uORF, RiceNet v2 and the Chinese Rice Variety Pedigree Database. It also summarizes the research progress of AI applications in multiple scenarios, such as intelligent phenotypic prediction, breeding efficiency improvement, environmental adaptability evaluation, key gene mining and genomic analysis. Furthermore, a breeding application framework integrating databases and AI is constructed, which consists of core modules including data integration and management, high-throughput data processing and compression, in-depth genomic analysis, multi-dimensional phenotypic prediction and intelligent decision support. The results indicate that the in-depth integration of databases and AI can remarkably enhance the efficiency of gene mapping, the accuracy of phenotypic prediction and the rationality of parent selection, as well as improve the scientificity and effectiveness of intelligent breeding decision-making. The breeding cycle can be shortened from 8-10 years to 3-5 years. At present, the development still faces bottlenecks such as inconsistent data standards, insufficient multimodal data fusion and lagging field application. In the future, it is necessary to formulate unified data standards, develop multimodal large models and strengthen collaborative verification among industry, academia and research institutes, so as to accelerate the realization of precise, efficient and intelligent rice breeding. This paper provides a systematic reference for the research, development and industrial application of intelligent rice breeding technologies.

Key words: rice, intelligent breeding, database, artificial intelligence, multi-omics data, phenotypic prediction, gene mining