Research Of Conformational B-cell Epitope Prediction Based On Features Of Epitope And Mimotope

In recent years, the immune information science has become a hot research field. Epitope is the main chemical substance in immune system recognition, and how to predict it accurately is one of the main challenges in the research of immune information science. B-cell epitope is located on the antigen surface, which can be recognized by antibody or B-cell receptor and triggers the body’s immune response. It can be the linear fragments in antigen structure(also called as linear B-cell epitope) or the areas of conformational structure in space(also known as conformational B-cell epitope). B-cell epitope prediction has a very high value in the field of vaccine manufacture, immune response and molecular recognition. However, the methods of biological experiment that be used to identify epitopes need higher cost and price. When we first find the candidate epitope regions by means of computer prediction and then combined with methods of experimental verification and identification can reduce cost and improve efficiency. So, studying out a prediction method of conformational B-cell epitopes which is accurate and effective has a very high application value.In recent years, mimotope-based conformational B-cell epitope prediction is a research focus in the field of epitope prediction. Mimotope is similar to the genuine epitope, and it can be combined with paratope on monoclonal antibody and causes the immune response, so it is thought to have functional equivalence with genuine epitope. The main idea of the prediction method based on mimotope is to map mimotope sequences back to the similar regions on antigenic surfaces, so as to find the genuine epitope. The study is focused on how to use the mimotope information to predict epitope accurately. Existing mimotope-based B-cell epitope prediction methods are generally using the three-dimensional structure of antigen and the sequences information of mimotope to predict epitope, but it seems to have encountered bottleneck to improve performance. There are some attribute features of epitope on the structure and physicochemical properties, and mimotope can mimic the function of epitope, there are also some measure attributes in mimotope. This paper tries to combine some characteristic information of epitope and mimotope to predict the conformational B-cell epitope, so as to break the bottleneck.This paper statistics and makes several comprehensive comparisons from the perspective of physicochemical characteristics to the attributes of epitope and mimotope, which aims to dig out some feature information that contained deeply in the epitope and mimotope sequences which can be used for epitope prediction. Then we use the result of the comparison, we select four features available and combining with Random Forest classification method to form a new method of conformational B-cell epitope prediction, which was based on features of epitope and mimotope. By comparing with other several classical algorithms, the sensitivity of our algorithm has been improved significantly(the value of the sensitivity is 0.53). And all the test samples have a certain number of correct epitopes are predicted, which is not possessed by other methods. Our method also performs well in the sensitivity / specificity scatter chart. These all prove the effectiveness of our approach. The research work in this paper, not only has great significance of theoretical studies on the methods of B-cell epitope prediction, but also will promote the development to practical application.