In the proposed work, we will develop software tools to predict T- and B-cell epitopes of allergenic and viral proteins. The approach is based on novel quantitative descriptors of the physical-chemical properties of amino acids developed recently by our group. The primary goal of the new approach is to use a minimal number of variables to establish the classification procedures and QSAR models. The novel descriptors of physical-chemical properties of amino acids will be used in combination with a partial least squares approach to reduce the number of variables in the discriminant analysis and in artificial neural networks. Algorithms based on multivariate classification, K-nearest-neighbor methods, support vector machines and neural networks will be developed and assessed by cross-validation for their ability to predict T- and B-cell epitopes in proteins. The resulting QSAR models/database approach can then be used to identify immunogenic epitopes in the proteins of pathogens for vaccine development and drug design. IgE epitopes, archived in our web-based, relational Structural Database of Allergenic Proteins (SDAP), will be used to develop the Bcell epitope prediction methods. Stereochemical variability plots will also be used to predict functional and immunological determinants on proteins from Dengue virus (DV). This information can aid in the design of vaccines that better stimulate neutralizing T- and B-cell responses to diverse variants of DV. The validated suite of software tools to identify and classify immunogenic peptides will be made available to the scientific community as a Web server, similar to SDAP. Collaborations with experimental groups will enable the practical applications of the tools, which include predicting the allergenicity of novel foods and drugs, improving specific immunotherapies for allergy and asthma, and vaccine design.
Showing the most recent 10 out of 24 publications
