Prediction of antigenic peptides of SARS- CoV-2 pathogen using machine learning

Area of Science:

• Immunoinformatics
• Computational vaccinology
• Machine learning in immunology

Background:

• Antigenic peptides (APs), or T-cell epitopes (TCEs), are crucial for pathogen recognition and immune response.
• Identifying TCEs is vital for designing effective epitope-based vaccines (EBVs).
• Traditional wet lab methods for TCE identification are laborious, costly, and time-intensive.

Purpose of the Study:

• To develop and validate a robust machine learning model for predicting SARS-CoV-2 TCEs.
• To present an efficient computational approach for identifying potential EBV candidates.
• To leverage XGBoost for high-accuracy TCE prediction.

Main Methods:

• Utilized peptide sequences of TCEs and non-TCEs from the Immune Epitope Database Repository (IEDB).
• Applied feature extraction to identify physicochemical properties of peptides for model training.
• Developed and trained an XGBoost machine learning model.

Main Results:

• The XGBoost model achieved a prediction accuracy of 97.6% on a test dataset.
• Five-fold cross-validation demonstrated a consistent mean accuracy of 97.58%.
• The model outperformed other evaluated machine learning classifiers in TCE prediction.

Conclusions:

• The developed XGBoost model offers a highly accurate and efficient computational tool for predicting SARS-CoV-2 TCEs.
• These predicted epitopes show significant promise as candidates for SARS-CoV-2 epitope-based vaccines.
• Further in vivo and in vitro validation is recommended to confirm the immunogenic potential of the predicted epitopes.