IEPAPI: a method for immune epitope prediction by incorporating antigen presentation and immunogenicity

Area of Science:

• Computational immunology
• Bioinformatics
• Cancer immunotherapy

Background:

• CD8+ T cells recognize peptides presented by HLA-I molecules, crucial for cancer immunotherapy.
• Existing prediction methods for HLA-I binding and antigen presentation lack precision due to unaddressed T-cell receptor recognition.
• Direct modeling of T-cell immune responses is limited by the underexplored mechanisms of T-cell receptor recognition.

Purpose of the Study:

• To develop a novel computational method, IEPAPI, for predicting immune epitopes by integrating antigen presentation and immunogenicity.
• To enhance the accuracy of neoantigen screening for T-cell vaccine design.

Main Methods:

• IEPAPI utilizes a transformer-based feature extraction block for peptide and HLA-I protein representations.
• It integrates antigen presentation prediction into the immunogenicity prediction branch to model biological process connections.
• The method simulates the interplay between antigen presentation and T-cell receptor recognition.

Main Results:

• IEPAPI outperformed NetMHCpan4.1 and mhcflurry2.0 on 100% and 76% of HLA subtypes in antigen presentation prediction, respectively.
• IEPAPI demonstrated superior precision on two independent neoantigen datasets compared to existing approaches.
• The method effectively incorporates antigen presentation and immunogenicity for improved epitope prediction.

Conclusions:

• IEPAPI offers a significant advancement in predicting T-cell immune responses and identifying cancer neoantigens.
• The method provides a vital tool for the design of effective T-cell vaccines.
• Integrating antigen presentation and immunogenicity prediction enhances the accuracy of epitope discovery.