Cross-view contrastive representation learning approach to predicting DTIs via integrating multi-source information.

Area of Science:

• Bioinformatics
• Computational Biology
• Drug Discovery

Background:

• Drug-target interaction (DTI) prediction is crucial for identifying new drugs and repositioning existing ones.
• Drugs and targets possess biological structural information in sequence data and biochemical functional information in heterogeneous networks.
• Leveraging multi-source data is essential for comprehensive understanding and accurate prediction of complex DTI machinery.

Purpose of the Study:

• To propose a novel model, MOVE (integrating multi-source information for predicting DTI via cross-view contrastive learning), for comprehensive drug and target representation learning.
• To effectively fuse information from diverse data sources, including sequence and network views, for improved DTI prediction.
• To address the challenge of integrating multi-perspective information in DTI prediction.

Main Methods:

• MOVE extracts features from both sequence and network data views.
• A fusion module is employed to integrate these multi-source representations.
• Auxiliary contrastive learning is utilized to enhance the fusion process and representation learning.

Main Results:

• Experimental results on a benchmark dataset demonstrate the effectiveness of the MOVE model.
• MOVE achieves accurate predictions in drug-target interaction tasks.
• The model successfully leverages multi-source information for improved DTI prediction.

Conclusions:

• MOVE provides an effective strategy for integrating multi-source data in DTI prediction.
• The proposed cross-view contrastive learning approach enhances representation learning for drugs and targets.
• The findings highlight the potential of MOVE in accelerating drug discovery and repositioning efforts.