用非线性预测方法研究蛋白质序列的特性(Ⅰ)

Nonlinear Prediction Analysis of Properties in Protein Sequences(Ⅰ)

通过详细HP模型将一条蛋白质序列构造为3个特征序列。其中蛋白质序列取自蛋白质分类数据库CATH,该库将蛋白质序列根据不同结构域分为4类(α主类,β主类,αβ类,低二级结构类)。然后采用非线性预测方法,研究每类蛋白质序列的特征序列,得到特征序列的误差比值(E-D)图。从图形上发现,每类蛋白质序列的每条特征序列的曲线起伏波动不断,具有特异性,且4类蛋白质序列对应特征序列的E-D图之间也有很大不同。

In this manuscript, three characteristic sequences for a protein sequence were constructed according to the detailed HP model. Protein sequences were selected from the protein structure classification database CATH. The database classified protein into four classes （mainly alpha, mainly beta, mainly alpha-beta and few secondary structures） based on the different domain structures. By using the nonlinear prediction methods, three characteristic sequences E-D graphs of protein sequences in every four classes were obtained. It was found that each characteristic sequences E-D graphs of protein sequences fluctuated all along and specific. Moreover, the difference of E-D graphs for corresponding characteristic sequences among four classes of protein sequences was very large.