R visNetwork + igraph weighted network visualization with visEdges

Question

I have a very simple question regarding combining igraph with visNetwork. I want to weight the edges with visEdges(value=E(graph)$weight), but that does not work. Here is a toy example to illustrate the problem:

     test
     [,1] [,2] [,3] [,4] [,5]
[1,]    0    1    3    7    1
[2,]    4    0    8    9    5
[3,]   10    3    0    8    3
[4,]    5    1    5    0    7
[5,]    8    2    7    4    0

library(igraph); library(visNetwork)

test.gr <- graph_from_adjacency_matrix(test, mode="undirected", weighted=T)

If I now try to visualize it as a weighted graph, it doesn't plot it:

test.gr %>%
  visIgraph(layout = "layout_in_circle") %>%
  visEdges(value = E(test.gr)$weight)

If I use

test.gr %>%
  visIgraph(layout = "layout_in_circle") %>%
  visEdges(value = 10)

instead, I get a plot:

but this is of course not what I want. I want different edge widths according to E(test.gr)$weigth.

Can you tell me how I can do this?

Curet

Answer 1

With visNodes and visEdges you can set global options for all nodes and edges. With e.g. visNodes(shape = "square"), your make all nodes squares. You understand that this is not the right route to set individual edge widths.

To achieve what you want, you could convert the igraph object to a visNetwork-list. Then, you could add an 'expected' column with the name "value". visNetwork will then use that column to give weight to the edges.

Maybe this is not thé best solution, but it works.

test <- as.matrix(read.table(header = FALSE, text = "
    0    1    3    7    1
    4    0    8    9    5
   10    3    0    8    3
    5    1    5    0    7
    8    2    7    4    0"))

library(igraph)
library(visNetwork)

## make igraph object
test.gr <- graph_from_adjacency_matrix(test, mode="undirected", weighted=T)

## convert to VisNetwork-list
test.visn <- toVisNetworkData(test.gr)
## copy column "weight" to new column "value" in list "edges"
test.visn$edges$value <- test.visn$edges$weight

test.visn$edges
# from to weight value
#   V1 V2      4     4
#   V1 V3     10    10
#   V1 V4      7     7
#   V1 V5      8     8
#   V2 V3      8     8
#   V2 V4      9     9
#   V2 V5      5     5
#   V3 V4      8     8
#   V3 V5      7     7
#   V4 V5      7     7

visNetwork(test.visn$nodes, test.visn$edges) %>%
  visIgraphLayout(layout = "layout_in_circle") 

This yields the following graph:

Please, let me know whether this is what you want.

Patric

You saved my life, that's exactly what I was looking for. Definitely, visNetwork should implement this. The edges difference are too heavy by the way. How could we avoid that? – According