Drug Target Prediction by Multi-View Low Rank Embedding.

Area of Science:

• Computational drug discovery
• Bioinformatics
• Machine learning in pharmacology

Background:

• Drug repositioning is crucial for efficient drug development.
• Predicting drug-target interactions (DTI) often relies on heterogeneous data.
• Existing methods frequently use single data representations, limiting prediction accuracy.

Purpose of the Study:

• To develop and evaluate a single-view (SLRE) and a multi-view (MLRE) approach for DTI prediction.
• To integrate multiple data representations (structural and chemical views) for enhanced prediction.
• To improve drug repositioning strategies through accurate DTI prediction.

Main Methods:

• Proposed a single-view low-rank embedding (SLRE) method for arbitrary data views.
• Extended SLRE to a multi-view learning approach (MLRE) integrating structural and chemical data.
• Evaluated methods against baseline single-view and multi-view approaches.

Main Results:

• Both SLRE and MLRE demonstrated superior performance compared to existing methods.
• The multi-view approach (MLRE) showed significant improvements by integrating diverse data.
• Predicted potential drug-target interactions for 30 FDA-approved drugs.

Conclusions:

• Integrating multi-view representations of drugs and proteins enhances DTI prediction accuracy.
• The proposed MLRE method offers a powerful tool for drug repositioning and development.
• The findings provide valuable insights for identifying novel therapeutic applications of existing drugs.