TSMDA

TSMDA predicts associations between microRNAs (miRNAs) and diseases by integrating target gene and symptom information to prioritize candidate miRNA-disease relationships for experimental validation.

Key Features:

• Integration of target and symptom information: Incorporates miRNA-target gene interaction data and symptom information to inform association predictions.
• Negative sample selection: Employs strategic selection of non-associated miRNA-disease pairs to improve model robustness and class balance.
• Predictive performance: Evaluated with 5-fold cross-validation (AUC 0.989) and blind tests (AUC 0.982), indicating high discrimination accuracy.

Scientific Applications:

• Prioritization of candidate miRNA-disease associations: Ranks potential miRNA-disease links to guide experimental validation.
• Oncology research: Applied to identify putative miRNA associations in breast, prostate, and lung cancers.
• Neurodegenerative disease research: Supports investigation of miRNA involvement in neurodegenerative disorders.

Methodology:

Applies machine-learning techniques to analyze datasets of miRNA-target interactions and symptom profiles, incorporates strategic negative sample selection, and uses 5-fold cross-validation and blind testing for performance evaluation (AUC 0.989 and 0.982).