Avoid overlap of bipartite network nodes in ggraph

Question 1

I'm plotting a bipartite (two-mode) network using igraph and ggraph.But the nodes are overlapping a lot, even though there is still space in the graphic window.

I would like to plot this using ggraph, and I know it's possible to plot with other packages, such as igraph and bipartite (but my networks do not come out nice enough from these packages).

library(igraph)library(ggraph)# data matrixdata <- matrix(0, nrow = 164, ncol = 32)pos_non_zero <- sample(1:5248, size = 300)data[pos_non_zero] <- sample(1:8, size = 300, replace = TRUE)colnames(data) <- paste0("plan", 1:ncol(data))rownames(data) <- paste0("ani", 1:nrow(data))# igraph objectgr_ob <- igraph::graph_from_biadjacency_matrix(data,weighted=T,directed=F)# ggraphpar(mar=c(0,0,0,0),mai=c(0,0,0,0)) # increasing the graphic window areaggraph(gr_ob, layout="bipartite") +  geom_edge_link(alpha=0.3,color="gray50", aes(width = E(gr_ob)$weight/2)) +  geom_node_point(size = 5,color = "blue",show.legend = F) +  geom_node_label(size=2,aes(label=name), angle=90,fill ="NA", label.size=0) +  guides(edge_color = "none", edge_width = "none")+  scale_y_continuous(expand = c(.15, .15))+  theme_void()

My plot is like this:

Any idea on how to avoid theoverlap of the nodes? The labels overlap are not a problem for me in this case.

Question 2

This post seems relevant:stackoverflow.com/a/46231204/28479453 - even then, aradial graph plot might be a better fit?

Question 3

@lailaps thanks! As soon as I can I'm gonna check this post, but I saw that is done in igraph... there my graph is fitting better, but I don't like much the appearence. But I even going to try to do similar changes in my current code (with ggraph). The radial is not that interesting for me, but I like the look! Maybe for another network.

Question 4

I think a bipartite graph can meet your need.

First, you need to move the rowname to the first column, convert it to the data frame and change it to a long table. I also exclude rows with values "0".

library(dplyr)df <- cbind(Name = rownames(data), data) %>%   as.data.frame() %>%   pivot_longer(!Name, names_to = "Plan", values_to = "n") %>%   mutate(n = as.numeric(n)) %>%   filter(n > 0)

Use the package calledbipartiteD3 to create the chart. But I want to sort the chart based on the 'n' to make the chart even more attractive.

SortPrim <- df %>%  # Sorting for the first column (Name)  group_by(Plan) %>%  summarise(Total=sum(n))%>%  arrange(desc(Total))SortSec <- df %>% # Sorting for the second column (Plan)  group_by(Name) %>%  summarise(Total=sum(n))%>%  arrange(desc(Total))

The chart created below can be further adjusted to meet your preference. I have set it to 'vertical' because it's a very long list. If you prefer a horizontal chart, as you provided in your question, you should change details from line 10. I still prefer the vertical version though. Read the documentation to make further adjustments.

library(bipartiteD3)bipartite_D3(df, colouroption = 'brewer',              Orientation = 'vertical',             ColourBy = 1,             PercentageDecimals=1,              PrimaryLab = 'Plan',             SecondaryLab = 'Name',             SiteNames = '',             SortPrimary = SortPrim$Plan,             SortSecondary = SortSec$Name,             MainFigSize = c(500, 3600),              IndivFigSize = c(200, 1300),             BoxLabPos = c(20, 20),             PercPos = c(110,110),             BarSize = 20,             MinWidth = 10,             Pad=8,             filename = 'Plot')

The cropped version of the plot:

Faisal Elvasador 495 bronze badges · Accepted Answer · 2025-11-09 04:07:59Z

I think a bipartite graph can meet your need.

First, you need to move the rowname to the first column, convert it to the data frame and change it to a long table. I also exclude rows with values "0".

library(dplyr)df <- cbind(Name = rownames(data), data) %>%   as.data.frame() %>%   pivot_longer(!Name, names_to = "Plan", values_to = "n") %>%   mutate(n = as.numeric(n)) %>%   filter(n > 0)

Use the package calledbipartiteD3 to create the chart. But I want to sort the chart based on the 'n' to make the chart even more attractive.

SortPrim <- df %>%  # Sorting for the first column (Name)  group_by(Plan) %>%  summarise(Total=sum(n))%>%  arrange(desc(Total))SortSec <- df %>% # Sorting for the second column (Plan)  group_by(Name) %>%  summarise(Total=sum(n))%>%  arrange(desc(Total))

The chart created below can be further adjusted to meet your preference. I have set it to 'vertical' because it's a very long list. If you prefer a horizontal chart, as you provided in your question, you should change details from line 10. I still prefer the vertical version though. Read the documentation to make further adjustments.

library(bipartiteD3)bipartite_D3(df, colouroption = 'brewer',              Orientation = 'vertical',             ColourBy = 1,             PercentageDecimals=1,              PrimaryLab = 'Plan',             SecondaryLab = 'Name',             SiteNames = '',             SortPrimary = SortPrim$Plan,             SortSecondary = SortSec$Name,             MainFigSize = c(500, 3600),              IndivFigSize = c(200, 1300),             BoxLabPos = c(20, 20),             PercPos = c(110,110),             BarSize = 20,             MinWidth = 10,             Pad=8,             filename = 'Plot')

The cropped version of the plot:

Movatterモバイル変換

Collectives™ on Stack Overflow

Avoid overlap of bipartite network nodes in ggraph

1 Answer1

Comments

Your Answer

Sign up orlog in

Post as a guest

Linked

Related

Hot Network Questions

Subscribe to RSS