NetMHCcons

NetMHCcons predicts peptide binding affinities to known Major Histocompatibility Complex (MHC) class I molecules to support identification of T cell epitopes.

Key Features:

• Integration of Multiple Prediction Methods: NetMHCcons uses a consensus approach that combines NetMHC, NetMHCpan, and PickPocket to improve prediction accuracy.
• Allele characterization threshold: For MHC alleles with at least 50 data points including a minimum of ten binders, NetMHCcons combines NetMHC and NetMHCpan for prediction.
• Prediction for poorly characterized alleles: For alleles lacking sufficient characterization, NetMHCpan is used as the primary predictor.
• Neighbor-based method selection: When the training data includes close neighbor alleles, NetMHCpan maintains superior performance, whereas combinations of NetMHCpan and PickPocket are used when an allele has more remote neighbors in the training set.

Scientific Applications:

• Epitope identification: Predict peptide–MHC class I binding affinities to identify candidate T cell epitopes.
• Vaccine design: Support selection of peptide epitopes for vaccine development against infectious diseases and cancers.
• Immunological research: Aid studies of cell-mediated immunity by mapping peptide presentation to T cells.
• Cancer immunotherapy: Assist identification of tumor-associated neoepitopes for therapeutic strategies.

Methodology:

NetMHCcons implements a consensus prediction approach by integrating NetMHC, NetMHCpan, and PickPocket and selects predictors based on allele-specific training-data thresholds (≥50 data points with ≥10 binders) and neighbor similarity in the training set.