Hands‐On Example - iffatAGheyas/bioinformatics-tutorial-wiki GitHub Wiki

6.3.8 Hands-On Example – Co-Expression Network in R & Cytoscape

In this mini-tutorial you’ll:

1. Simulate a small gene expression matrix.
2. Compute pairwise Pearson correlations.
3. Threshold to define edges.
4. Build an igraph network and export edge/node tables.
5. Import into Cytoscape for visualization.

---

A. R Code

# 1. Install & load packages
# install.packages(c("igraph","WGCNA"))
# BiocManager::install("RCy3")
library(igraph)
library(WGCNA)   # for network utilities
# library(RCy3) # optional, for programmatic Cytoscape control

# 2. Simulate expression: 50 genes × 20 samples
set.seed(123)
nGenes   <- 50
nSamples <- 20
expr     <- matrix(rnorm(nGenes * nSamples), nrow = nGenes, ncol = nSamples)
rownames(expr) <- paste0("Gene", sprintf("%02d", 1:nGenes))
colnames(expr) <- paste0("Sample", 1:nSamples)

# 3. Compute Pearson correlation matrix (genes × genes)
corrMat <- cor(t(expr))

# 4. Threshold correlations to define edges (e.g. |r| ≥ 0.8)
thresh   <- 0.8
adjMat   <- corrMat
adjMat[abs(adjMat) < thresh] <- 0
diag(adjMat)                <- 0

# 5. Build igraph object
g <- graph_from_adjacency_matrix(adjMat, mode = "undirected", weighted = TRUE)

# 6. Export edge list (for Cytoscape import)
edges <- as_data_frame(g, what = "edges")
write.csv(edges, "network_edges.csv", row.names = FALSE)

# 7. Export node table with degree (optional attributes)
nodes <- data.frame(
  id     = V(g)$name,
  degree = degree(g)
)
write.csv(nodes, "network_nodes.csv", row.names = FALSE)

# 8. (Optional) Quick plot in R
plot(
  g,
  vertex.label.cex = 0.7,
  vertex.size      = V(g)$degree + 2,
  edge.width       = E(g)$weight * 2,
  main             = "Co-Expression Network (r ≥ 0.8)"
)

• Files generated:

  • network_edges.csv (columns: from,to,weight)

  • network_nodes.csv (columns: id,degree)

B. Visualize in Cytoscape

1. Launch Cytoscape
   Download/install from https://cytoscape.org if needed.

2. Import Network

   1. File → Import → Network → From Table (Text/MS Excel)
   2. Select network_edges.csv, set:
      • Source Interaction = from
      • Target Interaction = to
      • Edge Attribute = weight
   3. Click Import

3. Import Node Attributes

   1. File → Import → Table → From File…
   2. Select network_nodes.csv, choose:
      • Key Column For Network = id
      • Table Contains = Node Table

4. Style & Layout

   • Layout → Prefuse Force Directed (or “yFiles Organic”) to arrange nodes
   • Style → Node Size mapped to degree (larger nodes = higher connectivity)
   • Style → Edge Width mapped to weight (thicker edges = stronger correlation)
   • Style → Node Label = id; adjust font size as needed

5. Export Figure

   1. File → Export → Network View as Graphics…
   2. Choose PDF, SVG, or PNG at desired resolution

Tip: You can also automate Cytoscape from R via the RCy3 package to script import, layout, and styling steps—run Cytoscape with the REST API enabled.