Ensemble Learning Models for Drug Target Interaction Prediction

Abstract

Drug discovery is a crucial phase before drug development since it is the most essential and distinctive means of testing all medications prior to their medical usage. Other processes in drug discovery include drug target interaction predictions, drug repurposing or repositioning, and drug design. Prediction of drug-target interactions is crucial in these cases. Proteins, enzymes, ion channels, and other components of the human body that aid in the treatment of disease are called targets. The interaction between protein targets in the human body and chemical compounds in medications is known as drug target interaction. In terms of time and money, drug discovery research laboratory experiments are inefficient. Machine learning-based procedures, on the other hand, improved the drug delivery mechanism. However, machine learning-based procedures improved drug discovery and drug target interaction prediction, which aided in the prediction of novel medications and the identification of new uses for current drugs. Different ensemble learning strategies for prediction are compared in this research. Ensemble learning methods are machine learning-based methods that employ numerous independent similar or different models to generate an output or make predictions. Among the several computational methods for predicting drug target interactions, ensemble learning methods are one of the chemogenomic strategies. When compared to single models, Extra Tree and Random Forest are extremely accurate ensemble learning approaches that also give low bias and low variance.