A learning-based method for drug-target interaction prediction based on feature representation learning and deep...

Area of Science:

• Bioinformatics
• Computational Chemistry
• Drug Discovery

Background:

• Drug-target interaction (DTI) prediction is crucial for efficient drug discovery.
• Traditional experimental methods are costly and time-consuming.
• Computational prediction of DTIs offers a cost-effective and time-saving alternative.

Purpose of the Study:

• To develop a novel computational method for predicting drug-target interactions.
• To enhance the accuracy and efficiency of drug discovery processes.
• To reduce the experimental costs and time associated with identifying new drug candidates.

Main Methods:

• Feature extraction from heterogeneous networks using Jaccard similarity and random walk.
• Dimensionality reduction and feature selection via denoising autoencoders.
• Development of a convolutional neural network (CNN) for DTI prediction (DTI-CNN).

Main Results:

• The DTI-CNN model achieved an average AUROC score of 0.9416 and AUPR score of 0.9499.
• DTI-CNN demonstrated superior performance compared to three existing state-of-the-art methods.
• The proposed method effectively predicts drug-target interactions.

Conclusions:

• The DTI-CNN method shows enhanced performance over existing approaches.
• The proposed computational model is well-designed for DTI prediction.
• This approach offers a promising tool for accelerating drug discovery.