M2PP

M2PP predicts pathogenic proteins targeted by existing drugs to identify drug–disease target associations and support drug repurposing research.

Key Features:

• Heterogeneous network construction: Constructs a heterogeneous network incorporating target proteins, diseases, and drugs and enriches it with neighborhood similarity information, drug-inferred information, and path information.
• Random forest regression model: Employs a random forest regression model to score unconfirmed target-disease pairs using features derived from the network.
• Validation and performance: Validated via five-fold cross-validation and compared to other state-of-the-art methods, demonstrating superior performance in predicting drug-targeted pathogenic proteins.
• Application to common diseases: Applied to predict high-ranked pathogenic proteins associated with common diseases with supporting evidence from public biomedical literature.

Scientific Applications:

• Drug repurposing: Identify existing drugs that can target newly predicted pathogenic proteins to accelerate therapeutic development.
• Prioritization of disease-associated targets: Prioritize candidate pathogenic proteins for common diseases based on network-derived scores and literature evidence.

Methodology:

Constructs a heterogeneous network of target proteins, diseases, and drugs; enriches nodes/edges with neighborhood similarity, drug-inferred, and path information; applies a random forest regression model to score unconfirmed target-disease pairs; and evaluates performance with five-fold cross-validation.