基于Transformer的大豆基因组预测方法
Journal: Agricultural Science DOI: 10.12238/as.v8i3.2800
Abstract
随着大豆育种的需求日益增加,传统育种方法面临着时间长、成本高以及环境因素影响较大的挑战。因此,基因组预测技术成为了一种提升大豆育种效率的关键工具。研究的目的在于应用Transformer模型,结合大豆SNP数据,构建一个准确、高效的大豆基因组预测模型,探索其在基因型到表型预测中的应用潜力。首先,通过数据预处理和特征选择,将大豆SNP数据转换为适合Transformer模型输入的格式。其次,结合Transformer模型中的自注意力机制,探索如何捕捉SNP间的长距离依赖关系,从而提高性状预测的准确性。为了进一步优化模型性能,采用了贝叶斯优化算法,自动化搜索超参数配置,以提高计算效率和预测精度。实验结果表明,与传统的CNN、LSTM等模型相比,基于Transformer模型的大豆基因组预测框架,在单性状预测及多性状联合预测任务中均表现出较高的精度。此外,贝叶斯优化进一步提升了模型的超参数选择效率,减少了手动调参的工作量。
Keywords
大豆；单核苷酸多态性；Transformer；Bayes优化
Full Text
PDF - Viewed/Downloaded: 0 TimesReferences
[1] Cabanos C,Matsuoka Y,Maruyama N.Soybean proteins/pep tides: A review on their importance, biosynthesis, vacuolar sorting,and accumulation in seeds[J].Peptides,2021,143:170598.
[2] Vargas-Almendra A, Ruiz-Medrano R, Núñez-Muñoz L A, et al.Advances in Soybean Genetic Improvement[J].Plants, 2024,13(21):3073.
[3] Du H, Fang C, Li Y, et al. Understandings and future challenges in soybean functional genomics and molecular breeding[J]. Journal of Integrative Plant Biology, 2023, 65(2): 468-495.
[4] Vaswani A, Shazeer N, Parmar N, et al. Attention is all you need[J]. Advances in neural information processing syste ms,2017,30.
[5] Wu C,Zhang Y,Ying Z, et al. A transformer-based genomic prediction method fused with knowledge-guided module[J]. Briefings in Bioinformatics,2024,25(1):bbad438.
[2] Vargas-Almendra A, Ruiz-Medrano R, Núñez-Muñoz L A, et al.Advances in Soybean Genetic Improvement[J].Plants, 2024,13(21):3073.
[3] Du H, Fang C, Li Y, et al. Understandings and future challenges in soybean functional genomics and molecular breeding[J]. Journal of Integrative Plant Biology, 2023, 65(2): 468-495.
[4] Vaswani A, Shazeer N, Parmar N, et al. Attention is all you need[J]. Advances in neural information processing syste ms,2017,30.
[5] Wu C,Zhang Y,Ying Z, et al. A transformer-based genomic prediction method fused with knowledge-guided module[J]. Briefings in Bioinformatics,2024,25(1):bbad438.
Copyright © 2025 陈艾灵
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License
