Back

A new and highly effective method for predicting T-cell response targets implemented on SARS-CoV-2 data

Flegr, J.; Zahradnik, D.

2022-09-09 immunology
10.1101/2022.09.09.507307 bioRxiv
Show abstract

Computational T-cell epitope prediction is essential in many immunological projects, including the development of vaccines. T-cells of immunocompetent vertebrate hosts can recognize as non-self only peptides which are present in the parasites proteins and absent in the hosts proteins. This basic principle allows us to predict which peptides can elicit T-cells response. We built on the fact that the specificity of T-cells reacting to SARS-CoV-2 antigens has been recently mapped in detail. Using Monte Carlo tests, we found that empirically confirmed peptides that stimulate T-cells contain an increased fraction of pentapeptides, hexapeptides, and heptapeptides which are not found in the human proteome (p < 0.0001). Similarly, hexapeptides absent in human proteins were overrepresented in peptides that elicited T-cell response in a published empirical study (p = 0.027). The new theory-based method predicted T-cell immunogenicity of SARS-CoV-2 peptides four times more effectively than current empirically based methods.

Matching journals

The top 5 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.