SNF-NN

SNF-NN predicts novel drug-disease interactions by integrating similarity network fusion (SNF) with neural networks to support computational drug repositioning.

Key Features:

• Similarity Network Fusion (SNF): Fuses heterogeneous drug-related and disease-related similarity matrices together with known drug-disease interactions to capture complex relationships.
• Deep Learning Integration: Uses a neural network model fine-tuned for interaction prediction to enhance predictive performance.
• High-performance Computing: Leverages high-performance computing resources for model training and network fusion computations.
• Robust Evaluation: Evaluated using stratified 10-fold cross-validation, reporting performance metrics of [Formula: see text] = 0.867 and [Formula: see text] = 0.876.
• Comparative Analysis: Compared against nine baseline machine learning methods and seven state-of-the-art methods, with reported metric ranges of [Formula: see text] 0.879–0.931 and [Formula: see text] 0.856–0.903.
• Clinical Validation: Cross-references predicted interactions with clinical trials and published studies for validation.

Scientific Applications:

• Drug Repositioning: Predicts novel drug-disease interactions to identify new therapeutic uses for existing drugs.
• Pharmaceutical Research and Development: Supports prioritization of drug candidates and optimization of the drug development process to reduce time and costs compared to traditional methods.
• Integrative Genomics Analyses: Integrates genomics data and large-scale publicly available datasets to enhance computational drug-repositioning analyses.

Methodology:

Heterogeneous similarity information and known drug-disease interactions are processed through similarity selection and Similarity Network Fusion (SNF), and the fused network is input to a neural network model that is fine-tuned to predict interactions and evaluated with stratified 10-fold cross-validation.