NetCleave: an open-source algorithm for predicting C-terminal antigen processing for MHC-I and MHC-II

Area of Science:

• Immunology
• Bioinformatics

Background:

• Cell surface antigen presentation is crucial for immune response.
• C-terminal antigen processing is a critical bottleneck in peptide presentation pathways, defining the downstream peptidome.

Purpose of the Study:

• To introduce NetCleave, an open-source algorithm for predicting C-terminal antigen processing in both MHC-I and MHC-II pathways.
• To evaluate NetCleave's performance against existing tools using immunogenicity datasets.

Main Methods:

• Developed NetCleave, a neural network-based algorithm.
• Trained the network on 46 physicochemical descriptors of cleavage site amino acids.
• Assessed prediction accuracy using AUC and evaluated performance on four independent immunogenicity datasets.

Main Results:

• NetCleave achieved high prediction accuracy for MHC-I (AUC=0.91).
• Prediction for MHC-II remains challenging (AUC=0.66).
• NetCleave demonstrated superior performance compared to other prediction tools on immunogenicity datasets.

Conclusions:

• NetCleave is a highly accurate tool for predicting C-terminal antigen processing, particularly for MHC-I.
• C-terminal processing prediction holds promise for identifying immunogenic peptides, potentially aiding vaccine and immunotherapy development.