SEPPA 3.0-enhanced spatial epitope prediction enabling glycoprotein antigens

Area of Science:

• Immunoinformatics
• Computational Biology
• Vaccine Design

Background:

• B-cell epitope prediction is vital for developing immunotherapies and vaccines.
• Glycosylation significantly impacts protein epitopes, yet existing prediction methods do not account for it.

Purpose of the Study:

• To introduce SEPPA 3.0, an enhanced computational tool for predicting B-cell epitopes on glycoprotein antigens.
• To address the limitations of previous methods by incorporating glycosylation effects.

Main Methods:

• Updated parameters using the latest, largest dataset.
• Incorporated micro-environmental features: glycosylation triangles and glycosylation-related amino acid indexes.
• Utilized a logistic regression model with final calibration based on neighboring antigenicity.

Main Results:

• Achieved an AUC of 0.794 via 10-fold cross-validation on internal data.
• On general protein antigens, SEPPA 3.0 yielded an AUC of 0.740 and balanced accuracy (BA) of 0.657.
• On glycoprotein antigens, SEPPA 3.0 achieved an AUC of 0.749 and BA of 0.665, outperforming peer methods.

Conclusions:

• SEPPA 3.0 is the first server capable of accurate epitope prediction for glycoproteins.
• Demonstrates significant advantages over existing tools for both general protein and glycoprotein antigens.
• Provides a valuable resource for immunotherapies and vaccine development, accessible online with batch query support.