SMART:a sparse mask attention recurrent transformer framework for crop genome-to-phenotype predictionOA
Innovation of breeding technology has played a key role in improving crop cultivars and will continuously be the driving force for developing high-yield and climate-smart crop cultivars to meet future food demands[1].With the advance of molecular and computational technologies,we are entering the era of Breeding 4.0,which deeply integrates the cutting-edge molecular tools with both phenotypic and genotypic big data of crops to breed new cultivars based on a pre-designed blueprint[1].Predicting phenotypic outcomes of breeding materials based on their genomic information,or genome-to-phenotype(G2P)prediction,is one of the crucial components of Breeding 4.0,so as to speed up the breeding process[1].Machine learning,especially deep learning,is an effective approach for exploring G2P due to its exceptional ability to capture nonlinear relationships between genomic features and phenotype.To that end,we developed a novel hybrid framework,termed SMART(Sparse Mask Attention Recurrent Transformer),for crop G2P prediction(Fig.1).
Zijie Shen;Enhui Shen;Qian-Hao Zhu;Longjiang Fan;Chu-Yu Ye
Institute of Fundamental and Frontier Sciences,University of Electronic Science and Technology of China,Chengdu,610054,China Institute of Crop Science&Institute of Bioinformatics,College of Agriculture&Biotechnology,Zhejiang University,Hangzhou,310058,China Hainan Institute,Zhejiang University,Sanya,572025,China School of Electronic and Communication Engineering,Shenzhen Polytechnic University,Shenzhen 518055,ChinaInstitute of Crop Science&Institute of Bioinformatics,College of Agriculture&Biotechnology,Zhejiang University,Hangzhou,310058,ChinaCSIRO Agriculture and Food,Canberra,ACT 2601,AustraliaInstitute of Crop Science&Institute of Bioinformatics,College of Agriculture&Biotechnology,Zhejiang University,Hangzhou,310058,China Hainan Institute,Zhejiang University,Sanya,572025,ChinaInstitute of Crop Science&Institute of Bioinformatics,College of Agriculture&Biotechnology,Zhejiang University,Hangzhou,310058,China
农业科技
crop cultivarscrop genome phenotype predictionmolecular computational technologieswebreed new cultivarsbreeding technologyimproving crop cultivarsphenotypic outcmolecular computational technologies
《Crop Design》 2026 (2)
P.48-50,3
supported by National Key R&D Program of China(2023ZD04076)Shenzhen Polytechnic University Postdoctoral Fund(6025331008K).
评论