iIL13Pred: improved prediction of IL-13 inducing peptides using popular machine learning classifiers

Area of Science:

• Bioinformatics
• Computational Biology
• Immunology

Background:

• Interleukin-13 (IL-13) is a key inflammatory cytokine implicated in various diseases, including asthma and COVID-19 severity.
• Understanding IL-13 regulation is crucial for developing novel therapeutic strategies.
• Characterizing molecules that regulate IL-13 induction is of significant therapeutic interest.

Purpose of the Study:

• To develop an improved prediction method for IL-13-inducing peptides.
• To identify highly relevant and non-redundant features for accurate peptide classification.
• To enhance the performance of existing IL-13 prediction models.

Main Methods:

• Utilized the Pfeature algorithm for peptide feature computation.
• Employed Minimum Redundancy Maximum Relevance (mRMR) for multivariate feature selection.
• Investigated seven machine learning classifiers: Decision Tree, Gaussian Naïve Bayes, k-NN, Logistic Regression, SVM, Random Forest, and XGBoost.

Main Results:

• The proposed iIL13Pred method achieved improved Area Under the Curve (AUC) and Matthews Correlation Coefficient (MCC) scores of 0.83 and 0.33, respectively, on validation data.
• mRMR feature selection identified discriminatory features for IL-13-inducing peptides, outperforming previous methods.
• The model demonstrated enhanced performance metrics compared to the state-of-the-art IL13Pred approach.

Conclusions:

• The iIL13Pred method offers superior performance in predicting IL-13-inducing peptides compared to existing methods.
• Benchmarking on validation and external datasets confirmed the robustness and improved accuracy of iIL13Pred.
• A user-friendly web server is available for rapid screening of IL-13-inducing peptides, facilitating research and therapeutic development.