基于一种新的基团编码的蛋白质二级结构预测  

A new method for protein secondary structure prediction based on radical group

作　　者：张帅燕[1] 刘毅慧[1] 

机构地区：[1]齐鲁工业大学信息学院,济南250353

出　　处：《智能计算机与应用》2017年第3期13-16,20,共5页Intelligent Computer and Applications

基　　金：国家自然科学基金(61375013);山东省自然科学基金(ZR2013FM020)

摘　　要：提出一种新的氨基酸编码方式,即基团编码,基团编码是对20种氨基酸进行的编码方式,含有42个属性,然后采用这种新的编码方式进行蛋白质二级结构预测。所有的氨基酸都可以有这几种基团来表示,这种基团编码方式中包含氨基酸或蛋白质中原子稳定结构的信息。实验中采用3折交叉验证,分别采用不同的滑动窗口数,通过支持向量机(SVM)来进行蛋白质二级结构预测,验证2组数据的准确率,可以发现氨基酸的不同的编码方式对预测精度会产生影响。经过实验对比,包含氨基酸内部稳定结构信息的基团编码方式的准确率比正交编码要高出1.2%。In this paper, a new encoding method is proposed called the radical group encoding, which is used to predict the secondary structure. The radical group encoding is the encoding method to encode 20 amino acids, it has 42 features. Almost all amino acids can be represented by the encoding method, it contains the information of stable structure of atoms in amino acids or proteins. The 3 fold cross validation experiments are used with different window length, the secondary structure of proteins uses the Support Vector Machine （SVM） to predict. After the two experiments, to compare the radical group encoding with the quadrature encoding, the conclusion can be achieved. The radical group encoding method, which contains the information of the internal stable structure of amino acid, is better than the quadrature encoding. The accuracy of radical group encoding is 1.2% higher than quadrature encoding.

关 键 词：蛋白质二级结构预测 基团编码 正交编码 SVM 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]