POPI

The software tool POPI is designed to predict the immunogenicity of MHC-binding peptides, which is important for peptide-based vaccine design. POPI uses a support vector machine (SVM) based system and a genetic algorithm to identify informative physicochemical properties from a dataset of 428 human MHC class I binding peptides. It achieves an accuracy of 64.72% using leave-one-out cross-validation, outperforming two other prediction methods. POPI is the first computational system for predicting peptide immunogenicity based on physicochemical properties and is available as a web server along with the dataset of MHC class I binding peptides.