Prediction of B-cell Antibody Binding Through Statistical Analysis of Epitope Variables : With Application to COVID-19 Vaccine Development

Abstract

A B-Cell is a type of cell in the body’s immune system that develops memory of different pathogens to be used toward future defense against disease. Antigens are the protein portion of a pathogen that functions as a target for antibodies released by B-cells. Epitopes are a portion of an antigen that is recognized by B-cells to initiate a defense mechanism and interacts with an antibody. The antibody it the component of a B-cell that recognizes the presence of a pathogen through memory and binds to the epitope for immunological attack. We analyzed different variables of antigen proteins to predict which variables will contribute most to whether an epitope peptide will bind to an antibody released by a B-cell. We used the program R Studio to apply logistic regression and XGBoost models to an existing data set. Next, we used simple text mining to create our own variables from protein strings. Results from these predictions could be applied toward the development of vaccines against various diseases, including COVID-19.