NetMHCstab

NetMHCstab predicts the stability (half-life) of peptide binding to Major Histocompatibility Complex class I (MHC-I) molecules using artificial neural networks to assess peptide–MHC-I complex stability and its relation to cytotoxic T lymphocyte (CTL) immunogenicity.

Key Features:

• Artificial Neural Networks: Uses ANNs to construct predictors that estimate the half-life of peptide–MHC-I (pMHC-I) complexes.
• Large-scale dataset: Trained on 5,509 peptide stability measurements across 10 HLA class I molecules.
• Affinity integration: Combines stability predictions with affinity predictions from NetMHCcons to improve identification of stable MHC-I complexes and T-cell epitopes.
• Sub-motif identification: Identifies distinct sub-motifs distinguishing stable and unstable binders, highlighting anchor positions such as P2 and P3 at the N-terminus.
• Validation against epitope databases: Validates predictions using known T-cell epitopes and non-epitopes from SYFPEITHI and IEDB.

Scientific Applications:

• Epitope discovery: Prioritizes peptide candidates likely to form stable pMHC-I complexes for T-cell epitope identification.
• Vaccine development: Supports selection of peptide antigens with stable MHC-I presentation for vaccine design.
• Cancer immunotherapy: Aids identification of tumor-specific antigens predicted to be stably presented on MHC-I molecules.
• Immunogenicity assessment: Assesses the contribution of pMHC-I complex stability to CTL immunogenicity.

Methodology:

Trains artificial neural networks on a dataset of 5,509 peptide stability measurements across 10 HLA class I molecules to predict pMHC-I half-life, integrates predictions with NetMHCcons affinity outputs, identifies sub-motifs and anchor-position effects (P2, P3), and validates results against SYFPEITHI and IEDB.