如何用ggraph复现RCy3（Cytoscape）生成的图结构？坐标复用后图结构差异的原因解析

I’ve run into this exact issue before—there are two key reasons why your ggraph plot doesn’t match the Cytoscape layout, and both are easy to fix once you know what to look for:

1. Opposite Y-Axis Directions #

Cytoscape and ggraph (which builds on ggplot2) use reverse y-axis orientations:

• In Cytoscape, the y-axis increases as you move down the screen
• In ggraph/ggplot2, the y-axis increases as you move up the screen

This means your imported y-coordinates are flipped vertically by default, making the entire graph look mirrored or structurally different.

2. Mismatched Node Order #

The node order returned by getNodePosition() doesn’t automatically match the node order in the igraph object created by createIgraphFromNetwork(). Since ggraph maps layout coordinates to nodes by their position in the igraph object, a misalignment here will scramble your nodes entirely.

How to Fix It #

Let’s adjust your code to address both issues:

Step 1: Flip the Y-Coordinate #

Reverse the y-values to align with ggraph’s coordinate system:

mylayout_Miserables$y <- -mylayout_Miserables$y

Step 2: Align Node Order #

Ensure the rows in your layout data frame match the node order in your igraph object. We’ll use node names to sort the layout correctly:

# Get the node names from your igraph object
igraph_node_names <- V(mygraph_Miserables)$name
# Reorder the layout data frame to match igraph's node sequence
mylayout_Miserables <- mylayout_Miserables[match(igraph_node_names, rownames(mylayout_Miserables)), ]

Full Corrected Code #

Here’s the updated version of your script with both fixes applied:

library(RCy3)
library(ggraph)
library(igraph)

# Load and prep data
lesmis <- system.file("extdata","lesmis.txt", package="RCy3")
dataSet <- read.table(lesmis, header = FALSE, sep = "\t")

# Create cleaned igraph object
gD <- simplify(graph.data.frame(dataSet, directed=FALSE))

# Create network in Cytoscape
createNetworkFromIgraph(gD, new.title='Les Miserables')

# Extract data from Cytoscape
mygraph_Miserables <- createIgraphFromNetwork()
mylayout_Miserables <- getNodePosition()

# Clean up coordinate columns
mylayout_Miserables$x <- as.numeric(as.character(mylayout_Miserables$x_location))
mylayout_Miserables$y <- as.numeric(as.character(mylayout_Miserables$y_location))
mylayout_Miserables <- mylayout_Miserables[, c("x", "y")]

# Fix 1: Flip y-axis to match ggraph
mylayout_Miserables$y <- -mylayout_Miserables$y

# Fix 2: Align layout node order with igraph
igraph_node_names <- V(mygraph_Miserables)$name
mylayout_Miserables <- mylayout_Miserables[match(igraph_node_names, rownames(mylayout_Miserables)), ]

# Get node labels and colors
vlabels_1 <- V(mygraph_Miserables)$name
color_labels_1 <- getNodeColor(vlabels_1)

# Generate matching plot
ggraph(mygraph_Miserables, layout = mylayout_Miserables) + 
  geom_edge_link(color = "orange", width=0.7) + 
  geom_node_point(size=5, color=color_labels_1) + 
  geom_node_text(aes(label = vlabels_1), size=2, color="gray50", repel=T) + 
  theme_void()

After making these changes, your ggraph plot should match the structure of the Cytoscape network perfectly.

One quick note: If you manually adjusted nodes in Cytoscape after creating the network, the imported coordinates will reflect those changes—so the plot will match the adjusted Cytoscape layout, not the initial ggraph kk layout. That’s expected behavior!

内容的提问来源于stack exchange，提问作者Alain Paris