How to create a network representing different clusters with other information?

Question

I have a dataframe nodes with information that looks like below:

dput(nodes)

structure(list(Names = c("A4GALT", "AASS", "ABCA10", "ABCA7", 
"ABCD4", "ABHD4", "ABTB1", "AC006978.2", "AC009119.2"), type = c("typeA", 
"typeA", "typeC", "typeA", "typeC", "typeC", "typeB", "typeB", 
"typeB"), type_num = c(1L, 1L, 3L, 1L, 3L, 3L, 2L, 2L, 2L), Clusters = c("Cluster1", 
"Cluster1", "Cluster2", "Cluster3", "Cluster3", "Cluster1", "Cluster2", 
"Cluster3", "Cluster2")), row.names = c(NA, 9L), class = "data.frame")

So, in the nodes dataframe, there are 4 columns. Names is the gene name, type is different types, type_num is the number given to each genetype and column Clusters show 3 clusters in which each gene belong to.

Similarly, I have other dataframe edges with with information like below:

dput(edges)

structure(list(fromNode = c("A4GALT", "A4GALT", "A4GALT", "A4GALT", 
"A4GALT", "A4GALT", "A4GALT", "A4GALT", "AASS", "AASS", "AASS", 
"AASS", "AASS", "AASS", "AASS", "ABCA10", "ABCA10", "ABCA10", 
"ABCA10", "ABCA10", "ABCA10", "ABCA7", "ABCA7", "ABCA7", "ABCA7", 
"ABCA7", "ABCD4", "ABCD4", "ABCD4", "ABCD4", "ABHD4", "ABHD4", 
"ABHD4", "ABTB1", "ABTB1", "AC006978.2"), toNode = c("AASS", 
"ABCA10", "ABCA7", "ABCD4", "ABHD4", "ABTB1", "AC006978.2", "AC009119.2", 
"ABCA10", "ABCA7", "ABCD4", "ABHD4", "ABTB1", "AC006978.2", "AC009119.2", 
"ABCA7", "ABCD4", "ABHD4", "ABTB1", "AC006978.2", "AC009119.2", 
"ABCD4", "ABHD4", "ABTB1", "AC006978.2", "AC009119.2", "ABHD4", 
"ABTB1", "AC006978.2", "AC009119.2", "ABTB1", "AC006978.2", "AC009119.2", 
"AC006978.2", "AC009119.2", "AC009119.2"), weight = c(0.005842835, 
0.002253695, 0.014513253, 0.004851739, 0.066702792, 0.009418991, 
0.001136938, 0.000474221, 0.004405601, 0.000666001, 0.005625977, 
0.0333554, 0.004666223, 0.000103131, 0.00026302, 0.004514819, 
0.029632695, 0.001825839, 0.028379806, 0.001403298, 0.008339397, 
0.02393394, 0.004782329, 0.024767355, 0.002986813, 0.00559471, 
0.005961539, 0.064831874, 0.013023138, 0.027935729, 0.006618816, 
0.001134219, 0.012798368, 0.007961242, 0.01640476, 0.007997743
), direction = c("undirected", "undirected", "undirected", "undirected", 
"undirected", "undirected", "undirected", "undirected", "undirected", 
"undirected", "undirected", "undirected", "undirected", "undirected", 
"undirected", "undirected", "undirected", "undirected", "undirected", 
"undirected", "undirected", "undirected", "undirected", "undirected", 
"undirected", "undirected", "undirected", "undirected", "undirected", 
"undirected", "undirected", "undirected", "undirected", "undirected", 
"undirected", "undirected")), row.names = c(NA, -36L), class = "data.frame")

Tried with igraph but didn't look the way I want.

library(igraph)
net <- graph_from_data_frame(d=edges, vertices=nodes, directed=F)

as_edgelist(net, names=T)
as_adjacency_matrix(net, attr="weight")

# Removing loops from the graph:
net <- simplify(net, remove.multiple = F, remove.loops = T) 

# Let's and reduce the arrow size and remove the labels:
plot(net, edge.arrow.size=.4,vertex.label=NA)

And it looks like this:

Can anyone please help me how to create a network like above with the data given above. Any help is appreciated. thank you in advance.

Your example data is not useable The Nodes data frame and the Edges data frame contain different nodes. However, the answer to [Grouped layout based on attribute](https://stackoverflow.com/q/37378744/4752675) shows how to make a plot that groups nodes by cluster — G5W, Aug 28 '20 at 17:15
@G5W thanks a lot for reply. I changed the data in the above post now. Please check now they have same nodes. Could you please help me how to plot that with the above data? thanq — beginner, Aug 28 '20 at 22:13

G5W · Accepted Answer · 2020-08-29T21:01:27.760

This is mostly a repetition of the answer to Grouped layout based on attribute.

I think that you want to group the vertices by the Clusters attribute and color them using the type attribute. I will do that in this answer. Your code creating the network is fine, but a simple plot does not group the vertices by the clusters (I have added coloring the vertices by type).

plot(net, edge.arrow.size=.4,vertex.label=NA, 
    vertex.color=as.numeric(factor(nodes$type)))

What you need is a layout that will emphasize the clusters. The previous answer cited above shows how to do that by generating a different graph with the same vertices but with heavy edge weights between vertices in the same cluster. In your case, it would be

Grouped.net = net
E(Grouped.net)$weight = 1

## Add edges with high weight between all nodes in the same group
for(Clus in unique(nodes$Clusters)) {
    GroupV = which(nodes$Clusters == Clus)
    Grouped.net = add_edges(Grouped.net, combn(GroupV, 2), attr=list(weight=80))
} 

## Now create a layout based on G_Grouped
set.seed(567)
LO = layout_with_fr(Grouped.net)

## Use the layout to plot the original graph
plot(net, layout=LO, edge.arrow.size=.4,vertex.label=NA, 
    vertex.color=as.numeric(factor(nodes$type)))

If you have a large number of vertices, you may also wish to reduce their size with vertex.size=4

thank you very much. Very useful. But I need small help. I'm applying this code for a huge data and it looks like this [https://i.imgur.com/MsPhXOL.png] 1) Here I want to name the clusters in the image. 2) I'm not able to see `typeC` Is there a way to bring those `typeC` on the top. 3) How to move the clusters little far so that I can see the edges (because now I can only see the whole grey color). — beginner, Aug 29 '20 at 21:17
I would start by reducing the size of the vertices. Start with `vertex.size=4` How many vertices of each type do you have? — G5W, Aug 29 '20 at 21:36
yes the image which I gave the link is `vertex.size=4`. typeA (930), typeB (110), typeC (3). Ofcourse typeC are very less but want them to bring on the top. — beginner, Aug 29 '20 at 21:40
Any help with this please https://stackoverflow.com/questions/64122201/how-to-rescale-the-plot-to-push-the-clusters-nodes-a-bit-further-apart-and-nam — beginner, Sep 29 '20 at 15:10

ThomasIsCoding · Answer 2 · 2020-08-29T06:00:29.870

0

I am not sure if the code below works

plot(net,
     edge.width = E(net)$weight,
     vertex.color = factor(V(net)$name),
     mark.groups = split(V(net)$name,V(net)$Clusters))

which gives

edited Aug 29 '20 at 06:00

answered Aug 28 '20 at 22:42

ThomasIsCoding

96,636
9
24
81

How to create a network representing different clusters with other information?

2 Answers2