基于信息融合和计算智能的构象性B细胞表位预测方法研究

发布时间：2018-01-09 05:19

本文关键词：基于信息融合和计算智能的构象性B细胞表位预测方法研究　出处：《东北师范大学》2016年博士论文　论文类型：学位论文

【摘要】：抗原能够使机体产生免疫应答,能够使免疫系统产生抗体物质与之反应。在抗原与抗体发生特异性反应时,抗原分子中决定抗原特异性的特殊化学基团被称为抗原表位。抗原表位与抗体结合,引起机体自身的免疫应答。在抗原表位中,可以被B细胞抗原受体或抗体特异性识别的片区被称为B细胞表位。B细胞表位对于机体构建获得性免疫具有重要作用,对抗体疫苗的制备,疾病的预防与治疗起到重大的指导作用。因此近些年来对于B细胞表位的预测成为了一个重要的研究方向。本文从模拟表位和抗原抗体结合特异性、模拟表位与抗原3D结构、表位属性信息等几个方向,采用信息融合和计算智能的方法,对构象性B细胞表位预测算法进行研究,主要成果如下:(1)提出了一个新的基于模拟表位和抗原抗体结合特异性的构象性B细胞表位预测算法。算法在基于模拟表位序列信息的B细胞表位预测方法的基础上,引入抗体对位的信息。在进行预测时不仅考虑抗原本身的信息,而且通过抗原和抗体结合的特异性信息,将抗体与抗原的结合特征引入构象性B细胞表位的预测。与单纯考虑模拟表位信息的几种方法在相同的测试集下比较,本文新提出的预测方法具有更好的预测性能。(2)提出了一个新的基于模拟表位和抗原3D结构信息的构象性B细胞表位预测算法。在进行预测时,首先根据抗原的3D结构信息进行计算,计算抗原的表面氨基酸残基信息(这些表面氨基酸残基更有可能成为表位);然后结合模拟表位的信息进行预测,通过两者的结合,使得抗原的表面氨基酸残基不再是氨基酸残基的简单叠加,而是将表面氨基酸残基按照模拟表位划分为具有统计意义的氨基酸残基区域;最后,为了获得更多的预测表位,对这些区域进行合理的组合,将合理组合后形成的区域中的氨基酸残基作为方法的预测结果。新提出的方法表明:算法的灵敏度、精度、正确率都有了比较明显的提升。(3)提出了一个新的基于表位属性特征信息融合的构象性B细胞表位预测算法。在进行预测时,根据表位体现出来的相关属性特征强度信息,使用计算智能的相关算法对预测的氨基酸残基进行分类,从而预测出表位信息及非表位信息。从最终的预测结果分析,该方法的预测性能比较稳定,同时也说明表位的属性信息对表位的预测算法具有一定的贡献度。但是同时存在的问题是如何发掘有效的构象性B细胞表位属性特征信息,来提高预测的准确性。(4)基于以上的算法,实现相关的在线预测平台,提供在线预测服务。根据不同的输入条件,利用不同的预测方法,实现构象性B细胞表位的预测。使用者可以根据不同的情况,利用预测平台来获得更准确的预测结果。预测平台实现了构象性B细胞表位预测理论研究的实际应用。
[Abstract]:Antigens can make the body immune response, can make the immune system to produce antibody substances to react with it, when the antigen and antibody specific reaction. The specific chemical groups in antigen molecules that determine antigen specificity are called antigen epitopes. The binding of antigen epitopes to antibodies causes the body's own immune response. B cell epitopes are known as B cell epitopes and B cell epitopes, which can be specifically recognized by B cell antigen receptors or antibodies, play an important role in the construction of acquired immune system, and also play an important role in the preparation of antibody vaccines. The prevention and treatment of diseases play an important role in guiding. So the prediction of B cell epitopes has become an important research direction in recent years. In this paper, mimic epitopes and antigen-antibody binding specificity. In order to simulate the 3D structure of epitopes and antigens and the information of epitope attributes, the prediction algorithm of conformational B-cell epitopes is studied by using information fusion and computational intelligence. The main results are as follows:. A new conformational B cell epitope prediction algorithm based on analogue epitope and antigen-antibody binding specificity is proposed. The algorithm is based on the analogue epitope sequence information. In the prediction, not only the information of the antigen itself is considered, but also the specific information of the binding of the antigen and the antibody is taken into account. The binding characteristics of antibodies and antigens were introduced into the prediction of conformational B cell epitopes, which were compared with several methods with only simulated epitope information under the same test set. In this paper, a new prediction method with better prediction performance is proposed. A new conformational B cell epitope prediction algorithm based on simulated epitopes and antigenic 3D structure information is proposed. Firstly, the surface amino acid residues of antigens are calculated according to 3D structure information of antigens (these amino acid residues are more likely to become epitopes; Then combined with the information of analog epitopes to predict, through the combination of the two, the surface amino acid residues of antigen is no longer a simple superposition of amino acid residues. The surface amino acid residues were divided into statistically significant amino acid residues according to the simulated epitopes. Finally, in order to obtain more prediction epitopes, these regions are reasonably combined. The amino acid residues in the region formed by reasonable combination are taken as the prediction results of the method. The new method shows the sensitivity and accuracy of the algorithm. A new conformational B-cell epitope prediction algorithm based on epitope attribute feature information fusion is proposed. The predicted amino acid residues are classified by using the correlation algorithm of computational intelligence according to the information of the characteristic strength of the related attributes reflected by the epitopes. Thus, the epitope information and non-epitope information are predicted. The prediction performance of this method is stable from the analysis of the final prediction results. It also shows that the attribute information of epitope has a certain contribution to the epitope prediction algorithm, but the problem is how to explore the effective conformational B-cell epitope attribute information. To improve the accuracy of prediction. (4) based on the above algorithm, the related online prediction platform is implemented to provide online prediction services. According to different input conditions, different prediction methods are used. To realize the prediction of conformational B cell epitopes. The prediction platform is used to obtain more accurate prediction results, and the practical application of the conformational B cell epitope prediction theory is realized.
【学位授予单位】：东北师范大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：R392

【相似文献】