iTTCA-Hybrid

iTTCA-Hybrid identifies tumor T cell antigens by integrating multiple sequence-based feature representations with machine learning to predict antigenicity for cancer immunotherapy applications.

Key Features:

• Hybrid Feature Representation: Combines amino acid composition, dipeptide composition, pseudo amino acid composition, distribution of amino acid properties within sequences, and physicochemical properties derived from the AAindex.
• Machine Learning Models: Employs support vector machines (SVM) and random forests for classification and prediction of antigenic peptides.
• Performance Metrics: Reports 73.60% accuracy and an area under the curve (AUC) of 0.783, corresponding to reported improvements of 4% in accuracy and 7% in AUC over existing methods.

Scientific Applications:

• Tumor T cell antigen discovery: Predicts tumor-derived peptides that may function as T cell antigens.
• MHC class I epitope identification: Prioritizes candidate peptides likely presented by MHC class I molecules.
• Personalized cancer vaccine and immunotherapy development: Supports selection of antigenic peptides for personalized vaccine and other immunotherapeutic strategies.

Methodology:

Generates peptide data from post‑genomic studies, applies a hybrid feature representation (amino acid composition; dipeptide composition; pseudo amino acid composition; distribution of amino acid properties; AAindex physicochemical properties), and utilizes support vector machine and random forest models to analyze and predict antigenic peptides.