Artificial Intelligence in Immunopeptidomics and Epitope Prediction: A Review of Workflows and Applications

The integration of artificial intelligence into immunopeptidomics and epitope prediction has transformed vaccine development, shifting the paradigm from empirical to data-driven design. This review synthesizes recent advances in AI-driven immunoinformatics across the vaccine development workflow, from antigen discovery to clinical translation. We analyzed key AI methodologies including machine learning, deep learning, and generative models, highlighting their application in epitope prediction, immunopeptidomics workflows, and neoantigen prioritization. Transformer-based architectures and protein language models now dominate epitope prediction tasks, achieving superior accuracy compared to traditional tools. Graph neural networks and AlphaFold-based structural modeling have improved conformational B-cell epitope prediction, while generative AI enables de novo immunogen design. AI-enhanced mass spectrometry workflows have revolutionized immunopeptidomics by improving peptide-spectrum matching and enabling identification of noncanonical epitopes. AI-based tools improved epitope prediction accuracy by up to ۱۵% compared to traditional methods. This review provides a strategic roadmap for implementing integrated AI workflows in vaccine development. We conclude that AI-driven immunoinformatics, coupled with rigorous experimental validation, represents a transformative framework for next-generation vaccine design.

کلیدواژه ها:

Artificial intelligence ، Immunopeptidomics ، Epitope prediction ، Machine learning ، Deep learning ، Vaccine design ، Immunoinformatics ، Neoantigen discovery