AllerTOP

AllerTOP predicts the allergenic potential of proteins from amino acid sequences by analyzing physicochemical properties and classifying sequences with machine learning to identify potential allergens.

Key Features:

• Physicochemical Property Analysis: Leverages detailed physicochemical properties of proteins, focusing on amino acid sequences to assess allergenic potential.
• Amino acid E-descriptors and transformations: Encodes sequences using amino acid E-descriptors and applies auto- and cross-covariance transformations to capture sequence-derived features.
• Machine Learning Classifiers: Implements logistic regression (LR), decision tree (DT), naïve Bayes (NB), random forest (RF), multilayer perceptron (MLP), and k nearest neighbours (kNN) for classification, with kNN achieving 85.3% accuracy in 5-fold cross-validation.

Scientific Applications:

• Food Safety: Identifies potential allergens in food products to inform safety assessments.
• Product Development: Detects allergenic proteins in commercial formulations such as detergents and washing powders to guide hypoallergenic product design.
• Medical Research and Diagnostics: Assesses the allergenic potential of therapeutic proteins and diagnostic reagents for patient safety evaluations.

Methodology:

Protein sequences are encoded with amino acid E-descriptors and transformed using auto- and cross-covariance, then classified with LR, DT, NB, RF, MLP, and kNN and evaluated by 5-fold cross-validation (kNN 85.3% accuracy).