Design and assessment of a machine learning Model based on physicochemical properties to predict immune relevant determinants of Pathogens

Abstract

newlineEpitope, also known as antigenic determinant or immune relevant determinant, is a part of an newlineantigen that is recognized by the immune system, specifically by antibodies, T-cells, or Bcells newlineand newlineis newlinecapable newlineof newlinestimulating newlinean newlineimmune newlineresponse. newlineThe newlineantigen newlineis newlinea newlinepart newlineof newlinea newlinepathogen newline newlinethat newline newlinethe adaptive immune system recognizes as a foreign object and are often structural newlineproteins that include portions of bacterial cell membranes and spike proteins of the virus. newlineEpitopes bind to helper T-cells, Cytotoxic T-lymphocytes, B-cells, antibodies, and antigenic newlinemolecules depending on the kind of an antigen. Prediction of epitopes is critical for vaccine newlinedevelopment, antibody production and immunodiagnostic tests as they play a crucial role in newlineactivating the immune system of humans. Using a wet-lab experimental approach, identifying newlinethese epitopes involves synthesizing full-length peptides and then performing immunological newlineobservations. While performing wet-lab experiments, all the peptides need to be tested newlineindividually to identify epitopes which makes the task burdensome in terms of cost, time and newlineeffort. Even though continual attempts are made in this field to improve it, the problem newlineremains unsolved and draws the attention of researchers. So an adaptive system based on newlinemachine learning (ML) techniques is desired to increase the accuracy of antigenic epitope newlineprediction. Among other applications, the primary goal of identification of T-and B-cell newlineepitopes in the target antigen is to design an epitope based peptide vaccine (EBPV).