蛋白质二级结构预测的结构表达方法研究

蛋白质二级结构预测时,描述窗口内氨基酸残基序列的各种物理化学和分子结构参数的数量非常大,而且不能很好地直接体现氨基酸的连接顺序,这样就造成二级结构预测工作既费时效果又低下,因此,对这些数据预处理是非常必要的.研究发现,这些初始数据对应的变换矩阵的本征值谱与氨基酸残基序列情况无关,只与氨基酸的类别、数量性质有关,而其本征函数数据能够直接反映氨基酸残基的序列情况,利用有显著大小的本征值对应的本征函数作为描述参数,在进行蛋白质二级结构预测时,不但能够显著减少描述参数的个数,而且它体现了氨基酸顺序的变化,这将有效提高蛋白质二级结构预测研究效果.

Data Representation Scheme for Predicting the Secondary Structure of Proteins

To predict the secondary structure of the proteins,up to hundreds of the physicochemical and molecular structure properties of the amino acids/residues are selected to describe a segment of proteins.Usually,the large number of describing parameters does not contain the linking property among the amino acides/residues in a segment of protein,so the secondary structure prediction based on this kind of describing scheme always can not obtain good results.The pretreatment of the data is required to improve the prediction results.Basing on the study,the eigenvalues of a square matrix,which is obtained when the parameter matrix of a segment of protein times its transpose,are the same as that of a matrix according to a segment protein in which the locations of the amino acids are exchanged.At the same time,the eigen functions will change by the acids' sequence.So the eigen functions according to the several notable eigenvalues are suggested as describing parameters to predict the secondary structure of proteins.This describing scheme for a segment protein not only involves much less parameters compared with the original ones but also the eigen functions vitally represent the amino acids' sequence property.This scheme has high potential to obtain better prediction results for the secondary structure of proteins.