BCEPS: A Web Server to Predict Linear B Cell Epitopes with Enhanced Immunogenicity and Cross-Reactivity

Area of Science:

• Immunoinformatics
• Computational Biology
• Vaccine Design

Background:

• Accurate prediction of linear B cell epitopes is crucial for developing antigen-specific antibodies and peptide-based vaccines.
• Existing methods may not sufficiently identify immunogenic epitopes or those inducing cross-reactive antibodies against native antigens.

Purpose of the Study:

• To present BCEPS, a novel web server for predicting linear B cell epitopes.
• To enable the selection of immunogenic epitopes that can elicit cross-reactive antibodies.
• To assist in the design of effective peptide-based vaccines and antibody production.

Main Methods:

• Development of BCEPS, a web server utilizing machine learning models.
• Training models on a dataset of 555 linearized conformational B cell epitopes from antibody-antigen structures.
• Evaluation of model performance using independent datasets, including data from the Immune Epitope Database (IEDB).

Main Results:

• The best performing model, a support vector machine, achieved 75.38% accuracy on the training dataset and 67.05% on an independent IEDB dataset.
• BCEPS allows ranking of predicted epitopes based on flexibility, accessibility, hydrophilicity, and predicted MHC II presentation.
• The server identifies potential issues like ectodomain localization and N-glycosylation sites that could hinder antibody recognition.

Conclusions:

• BCEPS provides a valuable tool for predicting immunogenic linear B cell epitopes.
• The server's features facilitate the selection of epitopes for vaccine development and antibody production.
• Application to SARS-CoV-2 Spike protein demonstrated BCEPS's ability to identify epitopes targeted by neutralizing antibodies.