Gene–Gene Network Builder

A lightweight Python tool for constructing biologically annotated gene–gene interaction networks from simple edge lists, with a focus on network biology workflows and interoperability (e.g. Cytoscape).

---

What This Tool Does

Given a simple edge list, this tool:

1. Constructs an undirected gene–gene network
2. Annotates edges with:
   • interaction type (e.g. physical, regulatory, pathway)
   • evidence/source (e.g. BioGRID, STRING)
3. Annotates nodes with basic network metrics:
   • degree (number of interactions)
4. Exports the network in GraphML (Cytoscape-ready)

---

Input Format

The primary input is a tab- or comma-separated edge list.

Minimal format

• TP53 BRCA1
• BRCA1 RAD51

Annotated format (recommended)

• TP53 BRCA1 physical BioGRID
• TP53 MDM2 regulatory Reactome
• BRCA1 RAD51 pathway KEGG

Notes

• interaction_type is optional (default: unspecified)
• source is optional but strongly encouraged
• Gene symbols are normalized by trimming whitespace and uppercasing
• Self-edges are removed post addition

---

Output

1. GraphML (Primary Output)

• Compatible with Cytoscape
• Preserves node and edge attributes

---

Example Usage

• gene-network --input examples/real_edges.tsv --output network.graphml

Load network.graphml directly into Cytoscape to explore:

• interaction types
• evidence sources
• node degree

Assumptions & Limitations

This tool makes several explicit assumptions:

• Interaction types are user-provided and not inferred
• Gene identifier normalization is minimal (no alias resolution)
• The network is undirected by design for simplicity
• These choices are intentional to avoid introducing incorrect biological inferences.

Future Extensions

Directional edges for regulatory networks

Confidence weighting based on evidence

Integration with pathway or expression datasets

Identifier mapping (HGNC ↔ Ensembl)