Network Graph in R with ggraph and igraph

Package

Author

Thomas Lin Pedersen

Sample data

ggraph works on graph objects created with igraph. Build one from an edge list data frame using graph_from_data_frame(). Each row is a connection between two nodes.

# install.packages("igraph")
library(igraph)

edges <- data.frame(
  from   = c("Alice", "Alice", "Bob",  "Carol", "Carol", "Dave", "Eve"),
  to     = c("Bob",   "Carol", "Dave", "Dave",  "Eve",   "Frank","Frank"),
  weight = c(3, 5, 2, 4, 1, 3, 2)
)

g <- graph_from_data_frame(edges, directed = FALSE)

Basic ggraph

ggraph() initializes the plot and sets the layout algorithm. Add geom_edge_link() for the edges and geom_node_point() for the nodes. theme_graph() removes the axes and grey background.

# install.packages("ggraph")
library(ggraph)

ggraph(g, layout = "fr") +
  geom_edge_link() +
  geom_node_point(size = 5) +
  theme_graph()

Basic network graph in R with ggraph geom_edge_link and geom_node_point

Node labels

Network graph with node labels in ggraph using geom_node_text

Add geom_node_text() and map name to label. Set repel = TRUE to push labels away from the node centers so they don’t overlap.

# install.packages("ggraph")
library(ggraph)

ggraph(g, layout = "fr") +
  geom_edge_link() +
  geom_node_point(size = 5) +
  geom_node_text(aes(label = name), repel = TRUE) +
  theme_graph()

Layouts

The layout argument controls how nodes are positioned. "fr" (Fruchterman-Reingold) is the default force-directed layout. Other common options are "kk" (Kamada-Kawai), "circle", "star" and "tree".

# install.packages("ggraph")
library(ggraph)

ggraph(g, layout = "circle") +
  geom_edge_link() +
  geom_node_point(size = 5) +
  geom_node_text(aes(label = name)) +
  theme_graph()

Network graph with circular layout in ggraph

Node color and size

Network graph with nodes colored and sized by degree in ggraph

Map degree(g) to size and color inside aes() to encode how many connections each node has. degree() is an igraph function that counts edges per node.

# install.packages("ggraph")
library(ggraph)

ggraph(g, layout = "fr") +
  geom_edge_link(alpha = 0.4) +
  geom_node_point(aes(size  = degree(g),
                      color = degree(g))) +
  scale_color_viridis_c(guide = "none") +
  scale_size(range = c(4, 10), guide = "none") +
  theme_graph()

Edge width

Map a numeric edge attribute to width inside aes() and use scale_edge_width() to set the output range. Here the weight column encodes connection strength.

# install.packages("ggraph")
library(ggraph)

ggraph(g, layout = "fr") +
  geom_edge_link(aes(width = weight), alpha = 0.6) +
  geom_node_point(size = 5) +
  scale_edge_width(range = c(0.5, 3), guide = "none") +
  theme_graph()

Network graph with edge width mapped to weight in ggraph

Directed graph

Directed network graph with arrows in ggraph

Set directed = TRUE in graph_from_data_frame() and pass an arrow() to geom_edge_link(). Use end_cap to keep the arrowhead from overlapping with the node.

# install.packages(c("igraph", "ggraph"))
library(igraph)
library(ggraph)

gd <- graph_from_data_frame(edges, directed = TRUE)

ggraph(gd, layout = "fr") +
  geom_edge_link(arrow   = arrow(length = unit(3, "mm"),
                                 type   = "closed"),
                 end_cap = circle(3, "mm")) +
  geom_node_point(size = 5) +
  theme_graph()

Curved edges

Replace geom_edge_link() with geom_edge_arc() to draw curved edges. The strength argument controls how much the arcs bend — 0 is straight, 1 is a semicircle.

# install.packages("ggraph")
library(ggraph)

ggraph(g, layout = "fr") +
  geom_edge_arc(strength = 0.3, alpha = 0.6) +
  geom_node_point(size = 5) +
  geom_node_text(aes(label = name), repel = TRUE) +
  theme_graph()

Network graph with curved edges using geom_edge_arc in ggraph