NetMHCpan

NetMHCpan predicts peptide binding affinities to known Major Histocompatibility Complex (MHC) molecules to identify peptide–MHC interactions relevant for immunology, vaccine design, and personalized medicine.

Key Features:

• Artificial Neural Networks (ANNs): Core predictive models use ANNs trained on experimental peptide–MHC binding affinity data.
• Pan-specific MHC coverage: Predicts peptide binding across any known MHC molecule and across MHC classes.
• Training on empirical binding data: Models are trained on extensive datasets of measured peptide–MHC binding affinities to generalize to novel peptides.
• Binding-affinity prediction: Produces predictions of peptide–MHC binding strength to prioritize candidate epitopes.

Scientific Applications:

• Immunology research: Identifies peptides likely to bind MHC molecules to support epitope discovery for vaccine development.
• Disease association studies: Explores how variation in peptide binding may influence disease susceptibility or progression.
• Personalized medicine: Predicts individual-specific peptide–MHC interactions to inform personalized therapeutic strategies.
• Immune escape analysis: Analyzes changes in peptide binding to investigate immune escape mechanisms in pathogens.

Methodology:

Applies a machine learning approach in which artificial neural networks are trained on empirical peptide–MHC binding affinity datasets; the models learn sequence and binding-pattern features that correlate with strong or weak binding to predict affinities for new peptides.