摘要
提出一种预测蛋白质二级结构的模式识别方法。该法首先对大量已知结构的蛋白质实验数据进行分析，找出鉴别蛋白质不同结构成分的有效信息，即设计分类器，然后实现对未知蛋白质二级结构的预测。用此方法对６４０个实验样本进行了研究，得到较高的预测精度，表明方法是有效的。还对实验结果进行了分析；讨论了有限样本对分类器性能的影响。
A pattern recognition method for the prediction of secondary structure of proteins is presented in this paper. By means of analysing and learnig of a lot of known structured proteins, a classifier is designed to predict the structure of unknown proteins. The experiment results for 604 samples indicate a higher correctness rate. The influence of finite sample size on classification. performance is also discussed in the paper. This method will be further developed.
出处
《清华大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
1992年第1期108-114,共7页
Journal of Tsinghua University(Science and Technology)
关键词
模式识别
蛋白
二级结构
预测
pattern recognition, protein, secondary structure, prediction
