摘要
采用一种信息离散性度量方法对CB396数据集中的蛋白质数据进行二级结构预测,预测准确率达到72.1%.为了提高预测准确率,将FDOD算法结合PSI-BLAST进行多重序列比对,使预测准确率提高到75.6%,证明了该方法的有效性.在此基础上,利用20种疏水标度改进FDOD方法,以减小计算量.最后,结合长程作用,对预测准确性的影响因素进行了讨论.
The secondary structure prediction is important in protein structure prediction since protein secondary structure is the basis of the tertiary structure. A method of information discrepancy-FDOD （ Function of Degree of Disagreement） was used to predict protein secondary structure in this work. Data were selected from the CB396 database and prediction accuracy was 72.1%. Besides, multiple alignment of PSI-BLAST was used to combine with FDOD, with the prediction accuracy increased to 75.6%. Furthermore, in order to reduce computation complexity, hydrophobic values were introduced to improve the algorithm and the impact factors of hydrophobic values. Long-range interaction is discussed in the prediction of protein secondary structure.
出处
《应用科学学报》
CAS
CSCD
北大核心
2008年第2期150-154,共5页
Journal of Applied Sciences
基金
上海市重点学科建设资助项目(No.T0102)
关键词
二级结构预测
信息离散性
疏水特性
长程作用
protein secondary structure prediction, disagreement degree, hydrophobic values, long-range interaction
作者简介
通信作者：严壮志，博士，教授，博导，研究方向：生物医学图像与生物信息处理；E-mail：zzyan@shu．edu．cn
