Color by degree in R using ggnet2

Question

I've been trying to plot a graph using ggnet2. To do this I use the following code:

library(igraph)
lapply(c("sna", "intergraph", "GGally", "igraph", "network"), require, character.only=T)
data <- read.table('CA-CondMat.txt',sep="\t",header=TRUE)
g = graph.data.frame(data, directed = TRUE)
N = vcount(g)
E = ecount(g)
perc = 0.1
d.g = degree(g,mode='all')/N
new_nodes = sample.int(N,ceiling(perc*N),replace=FALSE,prob =d.g)
new_g = subgraph(g,new_nodes)
dg = degree(g,mode='all')
prob = dg/sum(dg)
png('example_plot2.png')
ggnet2(new_g, size = "degree", node.color = "steelblue", size.cut = 4,
                                          edge.size = 1, edge.color="grey" )
dev.off()

and I get a completely blue graph.

I'm using the package igraph.

What I want to plot is a graph with the nodes' color based on their degree like this one:

Link to the file:
https://snap.stanford.edu/data/ca-CondMat.html

Edit:

Full example added

You need to provide a [reproducible example](http://stackoverflow.com/questions/5963269/how-to-make-a-great-r-reproducible-example) with sample input data. This will make it easier to help you. — MrFlick, May 20 '16 at 16:59
This is still not reproducible because we don't have any data to run it with (we do not have access to "CA-CondMat.txt"). You should provide data in ways described in the link I originally provided. — MrFlick, May 23 '16 at 18:34
Don't add a file. Links can break. Please read the link @Mrflick gave you. Reading up on [ask] might also be a good idea. — Heroka, May 23 '16 at 20:28

Mike Wise · Accepted Answer · 2016-05-25T08:01:29.087

I appreciate a challenge, and graphs are always fun. I think this is what you want (I modified my original to use the file you provided later as I was working on it):

In my code clr-degree is one half of the degree, as this file only has symmetric links and it looked rather boring without the black and green nodes.

I also added library prefixes to all the calls so I could see what was being used from what network library (igraph, network, etc). There is a lot of overlap and inter-dependencies in these libraries.

Note this code should map clr-degree to 0-1 to black, degree 2 to red, degree 3 to green, and >=4 to red:

library(ggplot2)
library(igraph)
library(GGally)

# the following libraries will be required too - used internally
lapply(c("sna", "scales","intergraph", "network"),require, character.only=T)

set.seed(1234)

# data from  https://snap.stanford.edu/data/ca-CondMat.html
data <- read.table('CA-CondMat.txt',sep="")

g = igraph::graph.data.frame(data, directed = TRUE)
N = vcount(g)
E = ecount(g)
d.g = igraph::degree(g,mode='all')/N

# Use new smaller subgraph
perc = 0.05
new_nodes = sample.int(N,ceiling(perc*N),replace=FALSE,prob =d.g)
new_g = igraph::subgraph(g,new_nodes)
dg = igraph::degree(new_g,mode='all')

dg <- dg/2  # for some reason there are only even degrees in this file - so we divide by 2

clrvek = pmax(0,pmin(dg,4))
clrnames = c("0"="lightgrey","1"="black", "2"="blue", "3"="green", "4"="red")

#png('example_plot2.png')
GGally::ggnet2(new_g, 
       color.legend="clr-degree",palette=clrnames,color=clrvek,
       size = "degree",
       edge.size = 1, edge.color="grey",
       legend.position = "bottom") + coord_equal()
#dev.off()

Yielding:

I had already figured out something like this. Now I was trying to improve it by dividing the nodes using `size.cut` and giving each group a certain color but I just can't understand how to do that. — totoedrm, May 25 '16 at 14:58
Well, can you mark this as correct (as it answers the question), and then ask a new one? That is how SO is supposed to work. — Mike Wise, May 25 '16 at 14:59

Color by degree in R using ggnet2

1 Answers1