NMF-DR

NMF-DR predicts novel drug-disease associations for drug repurposing by applying non-negative matrix factorization to heterogeneous drug-disease interaction networks.

Key Features:

• Recommender System Framework: Operates as a recommender system that suggests potential new uses for existing drugs based on established drug-disease relationships.
• Data Integration: Integrates multiple data sources for drugs and diseases and constructs a heterogeneous drug-disease interaction network from two types of disease similarities (including those derived from known drug-disease associations) and multiple drug similarity views.
• Improved NMF-Based Completion: Employs an enhanced non-negative matrix factorization technique to complete the drug-disease adjacency matrix and assign prediction scores to unknown drug-disease pairs.
• Prediction Performance Evaluation: Demonstrated superior prediction performance over several existing methods in experimental evaluations of drug-disease association prediction.

Scientific Applications:

• Drug repurposing candidate identification: Identifies novel therapeutic indications for existing drugs by predicting drug-disease associations.
• Prioritization for validation: Ranks predicted drug-disease pairs to support further exploration and experimental validation by researchers and pharmaceutical teams.

Methodology:

Constructs a heterogeneous drug-disease interaction network from integrated drug and disease similarities (including disease similarities derived from known drug-disease associations and multiple drug similarity measures) and applies an enhanced non-negative matrix factorization to complete the drug-disease adjacency matrix and score unknown drug-disease pairs within a recommender system framework.