Introduction to RISK¶
RISK (Regional Inference of Significant Kinships) is a next-generation tool for biological network annotation and visualization. It integrates community detection algorithms, statistically rigorous overrepresentation analysis, and high-resolution visualization to uncover structured relationships in complex networks.
RISK is designed to:
- Identify biologically coherent modules in large-scale networks
- Perform fast and flexible overrepresentation testing
- Generate publication-ready visualizations
- Scale to networks with hundreds of thousands of edges on standard hardware
- Generalize beyond biology to interdisciplinary networks
RISK Features¶
| Category | Capabilities |
|---|---|
| Clustering | Louvain, Leiden, Markov Clustering, Spinglass, Walktrap, Greedy Modularity, Label Propagation |
| Statistical Tests | Permutation, Hypergeometric, Chi-squared, Binomial |
| Network Formats | Cytoscape, Cytoscape JSON, GPickle, NetworkX |
| Annotation Formats | JSON, CSV, TSV, Excel, Python dictionary |
| Scalability | Efficient analysis of networks with 500k+ edges on standard hardware |
| Visualization | High-resolution outputs (SVG, PNG, PDF) |
| Cross-domain Use | Demonstrated on physics citation networks (Supplementary Fig. S7) |
Example Applications¶
- Functional module identification in S. cerevisiae PPI and GI networks
- Chemical–genetic interaction (Chem-GI) mapping to infer compound targets
- Microbial interaction networks to reveal host–pathogen relationships
- Cross-domain application to non-biological networks (e.g., physics citation networks)
Interactive Examples¶
For a complete workflow, see tutorial.html, which demonstrates all key steps—from loading a network to generating publication-ready figures—in a single notebook. You can also download the notebook and data as a ZIP.
The tutorial applies RISK to the Saccharomyces cerevisiae protein–protein interaction (PPI) network (3,839 nodes, 30,955 edges; Michaelis et al., 2023). It illustrates:
- Network clustering and module detection
- Annotation-based overrepresentation analysis
- Advanced visualization and interactive exploration
- Parameter export and reproducibility tools
This notebook complements this documentation—use it for end-to-end examples and contextual guidance.
You can also launch the Quickstart notebook on Binder for an interactive session without local installation.