基于图形表示方法的蛋白质亚细胞定位预测模型

发布时间：2018-10-09 18:42

【摘要】：图形表示方法作为生物序列比较分析的方法之一,由于其具有可视性、易于数值刻画的特点,已经被广泛应用在生物信息学的研究中。本文主要工作是提出两种新的图形表示并将其分别应用在序列相似性分析和亚细胞定位预测中。本文基于核苷酸三联体、氨基酸的疏水指数值和异参数迭代函数提出了一种新的图形表示,并给出一种数值刻画来量化不同序列之间的相似性。应用此方法本文分别比较了九个物种的ND5蛋白序列和十二个物种的β-珠蛋白序列的相似性,并利用得到的距离矩阵构建它们的进化树,得到的进化树与物种的进化关系一致。另外,利用相关系数我们比较本文方法和传统的经典算法Clustal W以及其它图形表示方法和ClustalW的结果,比较的结果显示本文方法在序列相似性分析的研究中是有效的。亚细胞定位预测一直是生物信息学中的一个热点问题。本文中,我们结合图形表示方法和BP神经网络方法提出了一种新的亚细胞定位预测模型。首先应用新的蛋白质序列图形表示和对应的数值刻画计算蛋白质序列之间的距离矩阵,将其标准化后导入BP神经网络,得到了一个新的亚细胞定位预测模型。进一步地,利用构造的预测模型,本文在两个数据集ZD98和CL317上进行了试验,在这两个数据集上的整体预测精度分别为94.9%、87.4%。另外,利用个体敏感度和整体预测精度这两个指标,将本文预测方法与已有文献中的亚细胞定位预测方法在同样的数据集ZD98和CL317上的预测结果作比较,比较的结果显示本文预测模型能有效的预测蛋白质的亚细胞定位。
[Abstract]:As one of the methods of comparative analysis of biological sequences, graphic representation has been widely used in the study of bioinformatics because of its visibility and easy numerical description. The main work of this paper is to propose two new graphical representations and apply them to sequence similarity analysis and subcellular location prediction respectively. Based on the hydrophobic index value of nucleotide triad, amino acid and the iterative function of different parameters, a new graphical representation is proposed, and a numerical characterization is given to quantify the similarity between different sequences. Using this method, the similarity of ND5 protein sequences of nine species and 尾 -globin sequences of 12 species are compared, and their evolutionary trees are constructed by using the distance matrix. The evolutionary tree obtained is consistent with the evolutionary relationship of species. In addition, the correlation coefficient is used to compare the proposed method with the traditional classical algorithm Clustal W and other graphical representation methods and the results of ClustalW. The comparison results show that the proposed method is effective in the study of sequence similarity analysis. Subcellular localization prediction has been a hot topic in bioinformatics. In this paper, we propose a new subcellular location prediction model based on graphical representation and BP neural network. Firstly, the distance matrix between protein sequences is calculated by using a new graphic representation of protein sequences and corresponding numerical characterization, and then introduced into BP neural network to obtain a new subcellular location prediction model. Furthermore, using the constructed prediction model, this paper has carried out experiments on two data sets, ZD98 and CL317, and the overall prediction accuracy on these two data sets is 94.9 and 87.4, respectively. In addition, using the two indexes of individual sensitivity and global prediction accuracy, we compare the prediction results on the same data set ZD98 and CL317 with the subcellular localization prediction method in the previous literature. The results show that the prediction model can effectively predict the subcellular localization of proteins.
【学位授予单位】：浙江理工大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：Q811.4;TP183

【相似文献】