Fig. 1

Overall architecture of the iIL13Pred design: The positive (IL-13 inducing peptides) and negative dataset (non-IL-13 inducing peptides) were obtained from IL13Pred (Jain et al. [19]).The positive and negative datasets were divided into 80:20 as training and testing data. The compositional features of Pfeature algorithm were used to compute features of IL-13 and non-IL-13 inducing peptides. Non-redundant and highly relevant feature selection tool mRMR was used to identify highly discriminatory and non-redundant features. Seven machine learning classifiers with five-fold internal cross validation was performed followed by an external validation on testing datasets. Best classifiers was then used to evaluate independent experimentally validated IL-13 inducing peptides. Abbreviations:IL-13, Interleukin-13; iIL13Pred, improved IL-13 prediction; mRMR, minimum redundancy maximum relevance; ML, Machine Learning; DT, Decision Tree; RF, Random Forest; SVM, Support Vector Machine; LR, Logistic Regression; GNB, Gaussian Naïve Bayes; KNN, k-Nearest Neighbour; XGB, eXtreme Gradient Boosting