Prediction of ITK inhibitor kinases activity based on posterior probabilistic weighted average based ensemble voting classification

Abstract

ITK inhibitor is used for the treatment of asthma and activity of inhibitor prediction helps to provide better treatment. Few researches were carried out for the analysis and prediction of kinases activity. Existing methods applied for the inhibitor prediction have limitations of imbalance dataset and lower performance. In this research, the Posterior Probabilistic Weighted Average Based Ensemble voting (PPWAE)ensemble method is proposed with various classifier for effective prediction of kinases activity. The PPWAE model selects the most probable class from the classification method for prediction. The co-train model has two advantages: Features are trained together to increases the learning rate of model and probability is measured for each model to select the efficient classifier. Support Vector Machine (SVM), Random Forest (RF), Decision Tree (DT), Classification and Regression Tree (CART), and Nave Bayes were among the classifiers employed. The results suggest that the Probabilistic Co-train ensemble technique performs well in kinase activity prediction. In the prediction of ITK inhibitor activity, the suggested ensemble method has a 74.27 percent accuracy, while the conventional SVM method has a 60% accuracy.