DeepNeo

DeepNeo predicts immunogenic neoepitopes presented by major histocompatibility complex (MHC) molecules and recognized by T cells, to identify neoantigens derived from somatic mutations or foreign antigens for research in cancer immunotherapy and viral medicine.

Key Features:

• Deep learning model: Uses a deep-learning approach to identify immunogenic neoepitopes from peptide sequences.
• Peptide–MHC structural modeling: Captures structural properties of peptide-MHC pairs associated with T cell reactivity.
• Immunogenicity beyond binding affinity: Predicts T cell-reactive neoepitopes by integrating information beyond physical binding affinity between mutant peptides and MHC molecules.
• Integration of complex biological interactions: Incorporates multiple biological factors to provide a more comprehensive prediction of neoantigen immunogenicity.
• Updated training data and evaluation: Employs up-to-date training data and improved evaluation metrics with prediction score distributions aligned to known neoantigen behavior.

Scientific Applications:

• Neoantigen discovery in cancer: Identification of tumor-specific mutational neoepitopes for use in cancer immunogenomics studies and vaccine design.
• Viral antigen analysis: Characterization of immunogenic epitopes derived from viral antigens for viral medicine research.
• Personalized cancer vaccine development: Prioritization of candidate neoepitopes for personalized vaccine formulation.
• T cell therapy optimization: Selection of T cell-reactive neoepitopes to inform the design and optimization of T cell-based therapies.

Methodology:

Deep learning trained on peptide–MHC pairs with T cell reactivity that models structural properties of peptide–MHC complexes, integrates factors beyond binding affinity, and uses updated training data with improved evaluation metrics and score distributions.