Bahai, Akash
[Verfasser:in]
;
McHardy, Alice
[Mitwirkende:r];
Dübel, Stefan
[Mitwirkende:r]
Computational methods for characterizing viral neutralization data and predicting linear B-cell epitopes ; Computergestützte Methoden zur Charakterisierung viraler Neutralisierungsdaten und Vorhersage linearer B-Zell-Epitope
Titel:
Computational methods for characterizing viral neutralization data and predicting linear B-cell epitopes ; Computergestützte Methoden zur Charakterisierung viraler Neutralisierungsdaten und Vorhersage linearer B-Zell-Epitope
Beteiligte:
Bahai, Akash
[Verfasser:in]
Erschienen:
TU Braunschweig: LeoPARD - Publications And Research Data, 2022-09-08
Anmerkungen:
Diese Datenquelle enthält auch Bestandsnachweise, die nicht zu einem Volltext führen.
Beschreibung:
Vaccines play a very crucial role in the control and prevention of several infectious diseases worldwide. However, there are many pathogens (such as HCV, HIV) against which a vaccine has not been developed and a preventive vaccine against such pathogens still evades researchers. Identifying neutralizing antibodies and new epitopes on the antigens are some of the ways that could improve our understanding of the immune response against these pathogens and subsequently help in developing new vaccines against them. Experimental methods for these tasks are expensive and time-consuming and this advocates for use of computational methods to accelerate research in these fields. In this thesis, I have developed two computational methods that would aid researchers in these tasks. Firstly, I analysed Hepatitis C neutralization data and then created a neutralization map for HCV using multidimensional scaling. This neutralization cartography allowed us to select a reduced reference panel for testing the breadth and potency of antibodies (and vaccines) against HCV. The panel also allowed us to select patients that develop the most potent antibodies, which are able to neutralize several HCV strains. I have created a pipeline for automating this method and it can potentially be utilized for other viruses as well. Secondly, I have created a new machine learning-based method called EpitopeVec for predicting linear B-cell epitopes on antigenic proteins. Although several methods for predicting linear B-cell epitopes have been published, their generalizability and accuracy in cross-testing is not satisfactory. Therefore, I also benchmarked many recently published methods on several datasets extensively for their generalizability and compared their performance with my method. The performance of the EpitopeVec method was better than the state-of-the-art methods in five-fold cross-validation as well as cross-testing. As the prediction performance of the methods depended on the type of antigens, I also created an EpitopeVec-viral method ...