Cotton Genomics and Genetics 2025, Vol.16, No.3, 148-162 http://cropscipublisher.com/index.php/cgg 148 Research Insight Open Access AI-Assisted Genomic Prediction Models in Cotton Breeding Jinhua Cheng, Mengting Luo Institute of Life Science, Jiyang College of Zhejiang A&F University, Zhuji, 311800, China Corresponding email: mengting.luo@jicat.org Cotton Genomics and Genetics, 2025, Vol.16, No.3 doi: 10.5376/cgg.2025.16.0015 Received: 28 Apr., 2025 Accepted: 07 Jun., 2025 Published: 29 Jun., 2025 Copyright © 2025 Cheng and Luo, This is an open access article published under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Preferred citation for this article: Cheng J.H., and Luo M.T., AI-Assisted genomic prediction models in cotton breeding, Cotton Genomics and Genetics, 16(3): 137-147 (doi: 10.5376/cgg.2025.16.0015) Abstract Cotton is an important economic crop related to the national economy and people's livelihood, but traditional breeding faces challenges such as long cycle, low efficiency and difficulty in improving yield and quality simultaneously. As a new technology of molecular breeding, genomic selection (GS) improves breeding accuracy and efficiency by utilizing whole genome marker information, and shows great potential in crop breeding. In recent years, the rapid development of artificial intelligence (AI) technology has injected new impetus into agricultural breeding. The application of machine learning and deep learning to crop genome big data analysis is expected to accelerate the breeding process of crops such as cotton. This study reviews the current status and challenges of cotton breeding, the basic principles of genomic prediction breeding, and the application progress of artificial intelligence algorithms in cotton breeding. The research progress of genomic prediction of major cotton traits such as yield, stress resistance and fiber quality is discussed in detail. Typical cases in Australia, the United States and China are cited to analyze the practice of cotton intelligent breeding. The current challenges in data quality and model generalization ability, multi-omics data integration, model interpretability, etc. are analyzed, and the future development direction of the integration of artificial intelligence and genomic prediction is prospected. This study hopes to break through the bottleneck of traditional breeding, improve the efficiency and accuracy of cotton breeding, and cultivate new varieties with high yield, high quality and multi-resistance. Keywords Cotton breeding; Genomic selection; Phenotypic prediction; Deep learning; Intelligent breeding 1 Introduction Cotton (Gossypium hirsutumLinn.) is one of the most important natural fiber crops in the world, and it is also an important cash crop and textile industry raw material in my country. Its yield and quality directly affect the textile industry and farmers' income. After years of development, cotton breeding has made remarkable progress, especially after the promotion of insect-resistant transgenic cotton, my country has become the second major country after the United States to have transgenic cotton varieties with independent intellectual property rights. However, cotton breeding still faces many challenges. Traditional breeding mainly relies on phenotypic selection and experience accumulation, with a long breeding cycle and low efficiency, and it is difficult to respond to new challenges of climate change and pests and diseases in a timely manner. At the same time, cotton yield and fiber quality are often negatively correlated, and it is extremely difficult to maintain or improve quality while increasing yield. For example, in the past, it was difficult to take into account both fiber length and strength in breeding, which once became a technical bottleneck. Conventional breeding has limited improvement in soil salinity, drought and other adversity resistance, and cotton production is still deeply affected by drought, salinity and disease. These factors have led to severe challenges in my country's high-quality and high-yield cotton breeding, which requires new technical means to break through (Sun et al., 2022). The development of molecular breeding has provided new ideas for cotton breeding. Genomic selection (GS) was proposed by Meuwissen et al. in 2001. It has developed significantly in the past two decades and has been verified in crops such as wheat and corn to improve selection accuracy and accelerate the breeding process. GS predicts individual breeding values by weighted estimation of high-density markers across the whole genome, overcoming the limitation of traditional marker-assisted selection that only uses a few major QTLs, and has shown great potential in improving crop yield, stress resistance and quality (Budhlakoti et al., 2018; 2022). With the decline in the cost of high-throughput sequencing and genotyping, cotton genome sequencing and variation map
RkJQdWJsaXNoZXIy MjQ4ODYzNA==