diffuStats

diffuStats performs label propagation and diffusion analysis over biological networks to contextualize molecular entities and prioritize novel candidates.

Key Features:

• Collection of Graph Kernels: Provides a collection of graph kernels for modeling diffusion processes over biological networks.
• Diverse Diffusion Scores: Implements multiple diffusion score definitions to assess and rank molecular entities.
• Posterior Statistical Normalization: Performs posterior statistical normalization on diffusion scores to produce statistically robust results.
• Parallel Permutation Analysis: Supports parallel permutation analysis of normalized scores for benchmarking and scoring-method selection.

Scientific Applications:

• Molecular Candidate Prioritization: Propagates labels across networks to prioritize novel molecular candidates based on network context.
• Gene Function Prediction: Uses diffusion-based scores to predict gene function within interaction networks.
• Disease Biomarker Discovery: Identifies and ranks candidate biomarkers in disease-related molecular networks.
• Protein–Protein Interaction Analysis: Contextualizes proteins within protein–protein interaction networks using diffusion methods.

Methodology:

Computational steps include selection of graph kernels, computation of diffusion scores, posterior statistical normalization, and parallel permutation analysis for benchmarking.