End-to-End Deep Learning Approach for Drug-Target Interaction Prediction
Authors
Abstract
The discovery of potential Drug-Target Interactions (DTIs) is a determining step in the drug discovery and repositioning process. Establishing effective computational methods, instead of the traditional in vivo or in vitro methods, is decisive to find new leads in a considerable short of amount of time. Deep Learning have shown to outperform state-of-art methods in multiple classification problems.In this work we evaluate an experimental setup that exploits the particular ability of Convolutional Neural Networks (CNNs) to obtain 1D representations from protein sequences (amino acid sequence) and SMILEs (Simplified Molecular Input Line Entry System, which represent the chemical structure as a string). This can be interpreted as features that express local dependencies or patterns that can then be used in a Fully Connected Neural Network (FC), acting as a binary classifier.
The results achieved show that the use of CNNs to obtain representations of the data, instead of using the traditional descriptors of proteins sequences and chemical structures, lead to a more effective performance.