Searching for SARS-CoV-2 Antigens that Trigger Strong T Cell Response

Yufeng Shen; Associate Professor in the Columbia University Department of Systems Biology and Department of Biomedical Informatics; Associate Director, Columbia Genome Center

In this project we propose to develop a computational method to predict SARS-Cov-2 viral antigens that elicit strong T cell responses based on machine learning and T cell receptor sequence data from recovered COVID-19 patients. This project is motivated by the critical role of T cell response in enhancing the maturation of B cells that recognize viral epitopes. A rationally designed vaccine must include antigens that elicit strong response from T cells, especially CD4 T helper cells. Currently there are published methods that predict suitable viral antigens based on HLA binding affinity. Our method can leverage T cell receptor (TCR) sequences of responding T cell clones in recovered patients. It would be complementary to HLA-based prediction. We will use deep learning methods to embed high-dimensional TCR sequence in lower dimensions, and then train the model to rank putative antigens for a given TCR. We will apply the method to published TCR data from recovered COVID-19 patients to predict candidate SARS-Cov-2 antigens. We are seeking collaborations to validate our predictions.