NTD-DR: Nonnegative tensor decomposition for drug repositioning.

Area of Science:

• Computational biology
• Pharmacology
• Bioinformatics

Background:

• Drug repositioning accelerates drug discovery by finding new uses for existing drugs.
• Traditional methods are time-consuming and costly.
• Integrating diverse data sources can improve prediction accuracy.

Purpose of the Study:

• To propose a novel computational method, nonnegative tensor decomposition for drug repositioning (NTD-DR), for identifying potential drug applications.
• To leverage multi-modal data for enhanced drug repositioning predictions.
• To validate the efficacy of NTD-DR against existing state-of-the-art methods.

Main Methods:

• Constructed a three-dimensional tensor of drug-target-disease associations using pairwise networks.
• Integrated drug, target, and disease similarity information.
• Employed nonnegative tensor decomposition (NTD-DR) for prediction.

Main Results:

• NTD-DR outperformed competing methods in predicting drug-disease associations, as measured by AUC and AUPR.
• Case studies confirmed the reliability of NTD-DR predictions for five diseases.
• The method identified more known associations and novel, literature-validated associations compared to other approaches.

Conclusions:

• NTD-DR is a powerful and reliable computational approach for drug repositioning.
• The method effectively integrates heterogeneous data to improve prediction accuracy.
• NTD-DR accelerates the drug discovery pipeline by identifying promising novel drug-disease associations.