GraphCrunch

GraphCrunch 2 performs network modeling, quantitative comparison, topological alignment, and topology-based clustering of protein-protein interaction (PPI) networks to support comparative and functional analyses.

Key Features:

• Network Modeling: Implements popular random network models and compares them with empirical PPI data across multiple network properties.
• Network Alignment (GRAAL): Incorporates the GRAph ALigner algorithm (GRAAL) for topological alignment of any pair of networks to reveal extended regions of topological and functional similarity.
• Topology-Based Clustering: Provides an algorithm that clusters nodes by topological characteristics to uncover relationships between interaction patterns and phenotypes.
• Network Comparison Metrics: Uses quantitative network properties such as degree distribution and clustering coefficients for model-to-real network comparisons.
• Performance Optimization: Utilizes parallel processing on multi-core CPUs to accelerate computationally intensive analyses.

Scientific Applications:

• Model Evaluation: Comparing PPI networks to random models to determine whether specific graph model families better represent eukaryotic or viral PPI networks.
• Cross-Species Functional Inference: Revealing functional similarities between proteins in yeast and human PPI networks through topology-based clustering.
• Phenotype–Topology Associations: Identifying relationships between interaction patterns and phenotypes via topology-driven clustering analyses.

Methodology:

Implements random network models and quantitative comparisons to empirical PPI networks using network properties (e.g., degree distribution, clustering coefficients), applies the GRAAL topological alignment algorithm, performs topology-based node clustering, employs heuristics for practical network comparison, and uses parallel processing on multi-core CPUs.