水稻蛋白质二级结构预测  

Prediction of Rice Protein Secondary Structure

作　　者：高光芹[1] 宛新生[1] 李晓 黄家荣[2] 王潇然[3] GAO Guangqin;WAN Xinsheng;LI Xiao;HUANG Jiarong;WANG Xiaoran(College of Science,Henan Agricultural University,Zhengzhou 450002,China;College of Forestry,Henan Agricultural University,Zhengzhou 450002,China;College of Life Science,Henan Agricultural University,Zhengzhou 450002,China)

机构地区：[1]河南农业大学理学院,河南郑州450002 [2]河南农业大学林学院,河南郑州450002 [3]河南农业大学生命科学学院,河南郑州450002

出　　处：《贵州大学学报（自然科学版）》2022年第2期45-49,65,共6页Journal of Guizhou University:Natural Sciences

基　　金：国家自然科学基金资助项目(31401450)。

摘　　要：针对水稻蛋白质二级结构预测研究,查阅了国家水稻数据中心文献资源,基于国际蛋白质数据库(protein data bank,PDB),选择具有代表性的蛋白质(5XQI)作为样本,应用BP神经网络建模技术,对水稻蛋白质二级结构进行预测研究。结果表明:先用氨基酸描述子量化一级结构,再用主成分分析综合描述子,能简化模型结构,提高模拟预测准确度和运行速度;构建标量型的人工神经网络模型和仿真函数预测式,简捷直观,应用方便;适宜的模型结构为21∶20∶3,即21个输入层节点、20个隐含层神经元、3个输出层神元的BP神经网络模型结构;模型的整体拟合准确度为0.85,H、E、C三种二级结构的拟合准确度分别为0.92、0.79、0.81;整体预测准确度为0.72,三种二级结构的预测准确度分别为0.79、0.65、0.71。基于BP神经网络的水稻蛋白质二级结构预测模型的拟合、预测准确度比以往同类研究高,为水稻蛋白质二级结构预测提供了一种新的研究方法。In view of the scarcity of rice protein secondary structure prediction research,based on the international protein database,PDB,the representative protein(5XQI)was selected as a sample,and BP neural network modeling technology was applied to predict rice protein secondary structure.The results show that quantifying the first-order structure with amino acid descriptors and then synthesizing the descriptors with principal component analysis can simplify the model structure and improve the accuracy and speed of simulation prediction.The scalar artificial neural network model and simulation function prediction formula are constructed,simple and intuitive,easy to use.The appropriate model structure is 21∶20∶3,that is,the BP model structure has 21 input layer nodes,20 hidden layer neurons and 3 output layer primitives.The overall fitting accuracy of the model is 0.85,and the fitting accuracy of H,E and C secondary structures is 0.92,0.79 and 0.81,respectively.The overall prediction accuracy is 0.72,and the prediction accuracy of the three secondary structures is 0.79,0.65 and 0.71,respectively.Compared with previous similar studies,the accuracy has been improved.This paper,employing a unique BP artificial neural network mathematical model,has provided a new method for studying the secondary structure of rice protein.

关 键 词：水稻 蛋白质 二级结构 预测 BP神经网络 

分 类 号：Q518.1[生物学—生物化学]