A Novel Deep Neural Network Technique for Drug-Target Interaction.

Area of Science:

• Bioinformatics
• Computational Chemistry
• Drug Discovery

Background:

• Drug discovery is costly and time-consuming.
• Drug repositioning and repurposing accelerate the process.
• Accurate drug-target interaction (DTI) prediction is crucial for efficient drug discovery.

Purpose of the Study:

• To present MPS2IT-DTI, a new computational model for DTI prediction.
• To develop a novel method for encoding molecular and protein sequences into images.
• To utilize a deep learning approach, specifically a convolutional neural network, for DTI prediction.

Main Methods:

• Encoding molecule and protein sequences into image representations.
• Developing a deep learning model (convolutional neural network) for DTI prediction.
• Training and evaluating the MPS2IT-DTI model on the Davis and KIBA datasets.

Main Results:

• MPS2IT-DTI demonstrates competitive performance compared to state-of-the-art methods.
• Achieved a concordance index of 0.876 and MSE of 0.276 on the Davis dataset.
• Achieved a concordance index of 0.836 and MSE of 0.226 on the KIBA dataset.

Conclusions:

• MPS2IT-DTI is a viable model for DTI prediction, balancing performance and model complexity.
• Representing sequences as images bypasses the need for embedding layers common in NLP-based approaches.
• The image-based encoding offers a novel approach to DTI prediction, potentially improving efficiency in drug discovery.