A new hybrid model for prediction of drug target interaction using deep learning techniques

One of the important areas is the prediction of the binding relationship between the drug compounds and their biological targets in drug discovery, which has been popularly known as DTI prediction. The success of a drug depends on its ability to specifically bind and modulate its target, often leading to desired therapeutic outcomes. But the off-target interactions can cause unwanted side effects, making the identification of accurate drug-target interactions essential in the initial stages of drug development. Recent progress in deep learning has permitted the development of strong models capable of capturing patterns which are complex in biological data. With the increasing complexity of biological systems and the vast number of potential drug-target pairs, it has become clear that computational methods are required to complement the growth. Attention based RNN with LSTM, GRU is used to further enhance both accuracy of prediction and interpretability of the model. The proposed model is evaluated using two performance metrics namely AUC and AUPR, then compared with all other existing methods.