MHCSeqNet

MHCSeqNet predicts peptide binding affinity to Major Histocompatibility Complex (MHC) molecules using deep learning to support neoepitope selection for immunotherapy and cancer vaccine development.

Key Features:

• Deep Learning Architecture: Leverages neural network architectures developed for natural language processing to model amino acid sequences of both MHC alleles and epitope peptides as tokenized sequences.
• Generalization Capability: Generalizes across unseen MHC class I alleles and predicts binding for peptides of varying lengths.
• Performance Excellence: Demonstrated superior performance relative to state-of-the-art predictors on MHC binding affinity datasets and MHC ligand peptidome datasets.

Scientific Applications:

• Cancer vaccine development: Aids selection of neoepitopes that bind strongly to MHC molecules for candidate inclusion in therapeutic cancer vaccines.
• Neoepitope identification for immunotherapy: Supports identification and prioritization of immunogenic peptides for synthetic peptide vaccine design and other immunotherapeutic strategies.

Methodology:

Translates biological sequences into formats amenable to NLP techniques, models amino acid sequences of MHC alleles and epitope peptides as sentences treating individual amino acids as words, and applies sequence representation strategies with neural network architectures from natural language processing to predict peptide–MHC binding affinity; the architecture accommodates new MHC allele sequences.