iIL13Pred

iIL13Pred predicts interleukin-13 (IL-13)-inducing peptides using machine learning to support identification of peptide modulators relevant to IL-13–mediated inflammatory conditions such as asthma, autoimmune disorders, neoplastic diseases, and COVID-19 severity studies.

Key Features:

• Enhanced Feature Selection: Uses minimum redundancy maximum relevance (mRMR) to identify non-redundant, highly relevant peptide features and contrasts this multivariate selection with regularization-based approaches such as linear support vector classifiers with L1 penalty.
• Comprehensive Classifier Suite: Employs seven classifiers: Decision Tree, Gaussian Naïve Bayes, k-Nearest Neighbour, Logistic Regression, Support Vector Machine, Random Forest, and Extreme Gradient Boosting.
• Performance Metrics: Reports validation performance with Area Under the Curve (AUC) of 0.83 and Matthews Correlation Coefficient (MCC) of 0.33.
• Extensive Benchmarking: Benchmarks against state-of-the-art approaches using validation data and external datasets comprising experimentally validated IL-13-inducing peptides.
• Robust Training Datasets: Trains models using an increased number of experimentally validated IL-13-inducing peptide datasets to improve generalizability.

Scientific Applications:

• Inflammatory disease research: Enables prediction of IL-13-inducing peptides relevant to asthma, autoimmune disorders, and neoplastic diseases.
• COVID-19 research: Supports investigation of IL-13 association with COVID-19 severity via predicted IL-13-inducing peptides.
• Drug discovery and biomarker identification: Aids identification of peptide-based therapeutic targets and biomarkers linked to IL-13 activity.
• Personalized medicine: Assists characterization of patient-specific inflammatory peptide profiles related to IL-13.

Methodology:

Feature selection used minimum redundancy maximum relevance (mRMR); classification employed Decision Tree, Gaussian Naïve Bayes, k-Nearest Neighbour, Logistic Regression, Support Vector Machine, Random Forest, and Extreme Gradient Boosting; performance was evaluated using AUC and MCC (AUC=0.83, MCC=0.33) on validation data; benchmarking included validation and external datasets of experimentally validated IL-13-inducing peptides, and training incorporated an increased number of experimentally validated peptide datasets.