MiRAGE: mining relationships for advanced generative evaluation in drug repositioning

Area of Science:

• Computational biology
• Pharmacology
• Bioinformatics

Background:

• Drug repositioning accelerates drug discovery and reduces costs by finding new uses for existing drugs.
• Integrating diverse biological data sources is key to discovering novel drug-disease associations (DDAs).
• Existing methods may not fully capture complex drug-disease relationships, highlighting the need for advanced integration techniques.

Purpose of the Study:

• To develop a novel computational method, MiRAGE, for effective drug repositioning.
• To enhance the identification of drug-disease associations (DDAs) by integrating multiple biological data sources.
• To overcome the limitations of current methods in capturing complex drug-disease mechanisms.

Main Methods:

• MiRAGE employs a three-step framework: negative sampling with hard negative mining, classification using random forest models, and feature selection based on feature importance.
• The method integrates diverse biological data to identify effective drugs for specific diseases based on coupled mechanisms.
• Evaluation was performed on multiple benchmark datasets to assess performance against state-of-the-art algorithms.

Main Results:

• MiRAGE demonstrated superior performance compared to existing state-of-the-art algorithms across various metrics.
• The method consistently outperformed others in uncovering novel drug-disease associations (DDAs).
• Case studies on Parkinson's disease and schizophrenia successfully identified top candidate drugs, validated by prior research.

Conclusions:

• MiRAGE is an effective and versatile computational tool for drug repositioning.
• The method provides valuable insights for therapeutic discoveries and addressing unmet medical needs.
• Integrating diverse biological data through MiRAGE significantly enhances the identification of novel drug-disease associations.