Step 5 — Co-expression Network

Builds a gene-gene correlation network from your expression matrix and detects modules (communities of co-regulated genes).

Inputs

• The expression matrix carried over from Step 4 (or any CSV).

Pipeline

1. Normalisation — log2(x+1), variance-stabilising, or none.

2. Similarity — Pearson, Spearman, or biweight midcorrelation.

3. Adjacency — soft threshold (WGCNA-style \(a_{ij} = |r_{ij}|^\beta\)) or hard threshold (binary above a cutoff).

4. Module detection — Louvain, Leiden (if installed), or hierarchical clustering with dynamic tree cut.

5. Eigengene — first principal component per module, plotted across samples.

Outputs

• network.gexf — for Cytoscape / Gephi.

• modules.csv — gene → module assignment.

• eigengenes.png — eigengene heat-map.