Drug Target Interaction Prediction by using Deep Learning Technique

Abstract

The development of drug-target interactions (DTIs) is a decisive step for drug discovery and reuse process, because the effect of Antibiotics is now declining. Several methods have been proposed for this problem, but they rarely use a combination of protein and synthetic materials. In this paper, deep learning approach, and an easy-to-use library for DTI prediction using neural networks in learning, from proteins (amino acid sequences) properties of networks and Simplified Molecular Compounds Input line entry system (SMILES) array are used. The outcome shows that using convolutional neural network (CNN) instead of conventional Annotations to acquire statistics representations can enhance overall performance. The deep learning approach outperformed machine learning strategies in successfully classifying effective and interactions. The proposed approach uses BLASTP for protein sequence dataset, that contain real-global goal interaction records. The DTiGEMS+ tool is used for integrating various features of the drug and target. The proposed approach achieves 96% accuracy as compare to the existing drug prediction strategies.