蛋白质特征曲线的研究

提出一种新颖的方案使蛋白质结构信息可视化。在滑动窗口方法基础上，每一个天然氨基酸采用从氨基酸索引数据库中挑选的48种特性参数描述，在某一特定窗口下的所有氨基酸残基的参数就组成一个矩阵，通过矩阵变换得到一个方矩阵，再经过窗口的滑动就得到基于整个蛋白质的所有这些窗口矩阵的本征值矩阵。对本征值矩阵元素作图得到一系列的本征值曲线，这种曲线的轮廓不随窗口的变化而变化，这些曲线被称为蛋白质的特征曲线。为选择合适的窗口宽度、对同一类型蛋白质不同窗口宽度及不同类型蛋白质相同窗口宽度下的本征值矩阵进行了比较研究，对其潜在的用途进行了讨论。

Research of characteristic curves of proteins

To make the information of protein structure visible, a novel scheme is suggested. In this work, each natural amino acid was described basing on 48 amino acid/residues characteristic parameters selected from the amino acid index database. The sliding window method was applied. A desired multidimensional eigenvalue matrix was obtained by all the parameter matrices, each of which was made of all the amino acid residues in one certain sliding window. To compare the desired eigenvalue matrices of different proteins, several different wide windows which include 7, 13, 19 or 21 amino acid residues, were adopted, respectively. The different eigenvalue curves were plotted for all the selected proteins according to different wide windows, which were called characteristic curves of proteins. The elementary comparison of the characteristic curves was analyzed and the undedying functions were discussed.