基于噬菌体组合肽库筛选的B细胞表位预测研究

发布时间：2018-07-17 14:58

【摘要】：在体液免疫过程中,B细胞表面受体(BCR)会识别外源抗原蛋白,并产生与抗原蛋白特异性结合的抗体,同时一部分B细胞会被激活分化成为记忆B细胞,并在下次病原体侵入体内时产生更加迅速的免疫应答。抗原表面被B细胞表面受体识别并与抗体特异性结合的区域称为B细胞表位。定位抗原表面B细胞表位对于设计人工疫苗、免疫干预治疗以及高通量的抗体制备而言都具有重要意义。目前,定位B细胞表位最可靠的方法是通过抗原-抗体复合体晶体衍射实验以及核磁共振的方法获得复合体的空间结构。然而这两种实验的方法都需要很高的成本以及大量的人力并且对于设备的要求也很高。随着一些辅助的实验手段的发展以及已知表位数据的增加,人们开始考虑使用计算机进行表位预测。通过预测的方法获得的候选表位可以通过后续的生物实验进行验证。使用这种实验和计算机相结合的方法既可以保证结果的准确又可以节约成本,提高工作效率。基于噬菌体组合肽库筛选的B细胞表位预测是实验方法和计算方法相结合的一种B细胞表位预测方法。方法首先通过噬菌体组合肽库筛选实验获取与抗体亲和度较高的模拟表位序列,然后利用这些模拟表位序列在抗原表面搜索与之相匹配的氨基酸预测候选表位。近年来,随着噬菌体组合肽库筛选获得的模拟表位序列数据和抗原-抗体复合体三维结构数据的不断增长,许多基于噬菌体组合肽库筛选的B细胞表位预测方法被提出,并在几个测试例上运行都得到了较好的预测结果。然而到目前为止,在基于噬菌体组合肽库筛选的B细胞表位预测方面还没有一个通用的标准测试集,同时对算法间性能的分析比较也没有一个完全的评价体系。本文的研究工作主要包括构建基于噬菌体组合肽库筛选的B细胞表位预测标准测试集、建立算法间性能的评价体系、提出更加敏感的基于抗原结构信息和噬菌体组合肽库筛选的B细胞表位预测新方法。首先,本文在对现有的基于噬菌体组合肽库筛选的B细胞表位预测方法研究基础上,整合了MimoDB、PDB、CED和IEDB 4个数据库中的相关信息,构建了一个通用的标准测试集。使用标准测试集及其代表测试集对Mapitope、PepSurf、Pepitope、Pep-3D-Search和EpiSearch 5个公开发表的基于噬菌体组合肽库筛选的B细胞表位预测方法进行了测试,这5个算法或提供源码或提供免费的网络服务。文章通过标准测试集及其代表数据集,并使用敏感性、特异性、准确率和马氏相关系数4个评价参数为基于噬菌体组合肽库筛选的B细胞表位预测方法建立了一个全面的评价体系,并对5个算法的性能进行了综合的评价分析。在综合评价分析基础上,本文提出了一种更加敏感的基于抗原结构信息和噬菌体组合肽库筛选的B细胞表位预测方法。算法首先根据结构特征并使用支持向量机对抗原氨基酸进行分类,实现对抗原的预处理;然后在现有表位预测算法的基础上引入划分的思想,通过将抗原表面氨基酸划分成若干交叠的patch区域进行表位预测。在为每一个patch构建无向图的过程中,算法首次尝试使用可变的距离阈值来定义无向图中顶点的连接。此外,本文第一次采用完备的搜索方法保证了搜索的路径最优。最后,通过与其它5个算法的测试结果进行比较验证,本文提出的算法的敏感度有很明显的提高。本研究不仅对B细胞表位预测方法的理论研究具有重要意义,同时也将推动其向实用方向发展。
[Abstract]:During the humoral immune process, the B cell surface receptor (BCR) recognizes the exogenous antigen protein and produces the antibody specific binding with the antigen protein. At the same time, a part of the B cells will be activated and differentiated into memory B cells and produce a more rapid immune response when the next pathogen invades the body. The antigen surface is identified by the surface receptor of the B cell and The region specifically binding to antibodies is called B cell epitope.
B cell epitopes on the surface of the antigen surface are of great significance for the design of artificial vaccines, immune intervention and high throughput antibody preparation. At present, the most reliable method for locating the epitopes of B cells is to obtain the spatial structure of the complex by means of the diffraction experiment of antigen antibody complex crystal and the square method of nuclear magnetic resonance. However, these two kinds of methods have been used to obtain the space structure of the complex. Experimental methods require a high cost, a large number of manpower and a high demand for equipment. With the development of some auxiliary experimental methods and the increase of known table data, people begin to consider using computers for epitope prediction. The candidate tables obtained by the prediction method can be achieved through subsequent biological reality. Verify that the combination of experiment and computer can ensure the accuracy of the result, save cost and improve work efficiency.
The prediction of B cell epitopes based on phage combination peptide library is a prediction method of B cell epitopes combined with the combination of experimental and computational methods. First, a simulated epitope sequence with high affinity to antibodies is obtained by the screening experiment of phage combination peptide library, and then the epitope sequences are used to search the surface of the antigens on the surface of the antigen. In recent years, a number of B cell epitope prediction methods based on the screening of phage combination peptide library have been proposed, which have been better used in several test cases. But so far, there is not a universal standard test set in the prediction of B cell epitopes based on the phage combination peptide library, and there is no complete evaluation system for the analysis and comparison of the performance between the algorithms.
The research work of this paper mainly includes the construction of the B cell epitope prediction standard test set based on the screening of phage combination peptide library, the establishment of the evaluation system for the performance of the algorithm, and a more sensitive new method for predicting B cell epitopes based on the antigen structure information and the screening of the phage combination peptide library.
First, on the basis of the existing B cell epitope prediction method based on phage combination peptide library, this paper integrates the related information in 4 databases of MimoDB, PDB, CED and IEDB, and constructs a general standard test set. The standard test set and its representative test set are used for Mapitope, PepSurf, Pepitope, Pep-3D-Search, and EpiS. Earch 5 published B cell epitope prediction methods based on phage combination peptide library screening were tested, these 5 algorithms or source code or free network services. The article passes standard test set and its representative data set, and uses 4 evaluation parameters based on sensitivity, specificity, accuracy and Markov correlation coefficient. A comprehensive evaluation system for predicting the B cell epitopes screened by the combined peptide library was established, and the performance of the 5 algorithms was comprehensively evaluated and analyzed.
On the basis of comprehensive evaluation analysis, this paper proposes a more sensitive B cell epitope prediction method based on antigen structure information and phage combination peptide library screening. Firstly, the algorithm is used to classify antigen amino acids based on structural features and use support vector machines to achieve antigen preprocessing; then, the existing epitope prediction algorithm is used. Based on the idea of division, the epitope is predicted by dividing the antigen surface amino acid into a number of overlapping patch regions. In the process of constructing an undirected graph for each patch, the algorithm first attempts to use a variable distance threshold to define the connection of the vertex in the undirected graph. In addition, this paper first uses a complete search method. The path of search is proved to be optimal.
Finally, by comparing with the test results of the other 5 algorithms, the sensitivity of the proposed algorithm is obviously improved. This study is not only of great significance to the theoretical research of the B cell epitope prediction method, but also will promote its practical direction.
【学位授予单位】：东北师范大学
【学位级别】：博士
【学位授予年份】：2011
【分类号】：R392

【参考文献】