🧩 CENTRA identifies modules using Louvain clustering.

🔍 Each module is tested for functional coherence via Overrepresentation Analysis.

📊 The resulting modules often align with coordinated biological processes and pathways.

🌿 Some genes reveal structure not by connections alone, but by how complex their local neighborhood is.

CENTRA uses Local Fractal Dimension to capture this, a measure of how dense the surroundings of a gene become in the network.

✨ High LFD can hint at elevated information density.

⭐ Some genes stand out because of who they’re connected to.

CENTRA uses eigenvector centrality to identify these influence hubs, genes linked to other highly connected players across the network.

✨ High eigenvector values can hint at strong regulatory relevance.

🔀 Some genes quietly sit at the crossroads of biological information flow.

CENTRA uses betweenness centrality to find these hidden mediators, genes that bridge modules and shape how information moves through a network.

✨ High betweenness can hint at regulatory leverage.

How do we define a key gene in a network?

CENTRA uses multiple centrality measures:
🔹 Betweenness
🔹 Eigenvector
🔹 Local Fractal Dimension

These combine to reveal functional importance across contexts.

CENTRA is available as interactive browser app:

🧭 Explore topic-specific networks
🔍 Search genes or terms
🌐 Visualize modules, metrics, enrichments

🎨 What if you could explore gene networks by topic, not just predefined categories?

CENTRA clusters gene sets from MSigDB into 27 topics using Latent Dirichlet Allocation, a topic modeling method designed for text.

🕸️ Each topic becomes its own biological network.

🚨 Our CENTRA paper is out in NAR Genomics and Bioinformatics!
CENTRA shows how centrality and fractal geometry uncover functional master regulators across biological networks. You can explore everything instantly, completely open access.

doi.org/10.1093/narg...