Venn diagrame from contingency table in R

Question

I have a data like contingency table, which display abundance of data, but I want to draw venn diagram from this data fram.

Structure of my data:

species_abundance<-data.frame(Genus = c("Parasphingorhabdus", "Loktanella", "Cytobacillus", "Paracoccus", "Paucisalibacillus", "Kytococcus", "Salinibacterium", "Acinetobacter baumanni","Marinococcus","Bacillus"),
               S3 = c(0, 0, 1, 1, 0, 0, 1,0,4,0),
               S5 = c(0, 0, 0, 1, 1, 0, 1,0,3,5),
               S7 = c(3, 1, 0, 2, 0, 1, 0,0,3,1),
               S9 = c(0, 1, 0, 3, 0, 0, 0,1,2,0)

How can I draw venn diagramm from this data frame in order to find unique and shared species accross different stations (S3, S5, S7.......)?

If I convert the data as given below and I try using Venny2 I will get such image, similar image and finding I want to do using R, kindly help

species_abundance1<-data.frame(S3 = c("", "", "Cytobacillus", "Paracoccus", "", "", "Salinibacterium","", "Marinococcus", ""),
                          S5 = c("", "", "", "Paracoccus", "Paucisalibacillus", "", "Salinibacterium","", "Marinococcus","Bacillus"),
                          S7 = c("Parasphingorhabdus", "Loktanella", "", "", "", "Kytococcus", "","", "Marinococcus","Bacillus"),
                          S9 = c("", "Loktanella", "", "", "", "", "","Acinetobacter baumanni", "Marinococcus",""))

Allan Cameron · Answer 1 · 2023-01-13T17:27:47.600

There are a couple of ways of getting a 4-variable Venn diagram in R, but Venn diagrams beyond this number of categories are extremely complicated and not a great way of visualising data. Here is an example of a 5-category Venn diagram from Wikimedia commons:

A 7-category Venn can't even be drawn using ellipses, and involves a complex floral shape as can be seen in the linked article.

In any case, you can see that a Venn even with 5 categories would not be a very user-friendly way to represent the data.

In your case, the natural way to present this kind of data would be via a heatmap. You first need to reshape your data into long format.

library(tidyverse)

species_abundance %>%
  pivot_longer(-Genus, names_to = 'Site', values_to = 'Count') %>%
  mutate(Site = factor(Site, unique(Site))) %>%
  ggplot(aes(Site, Genus, fill = factor(Count))) +
  geom_tile(color = 'black') +
  geom_text(aes(label = ifelse(Count == 0, '', Count))) +
  coord_equal() +
  scale_fill_manual(guide = 'none', 
                    values = c('white', 'lightyellow', 'yellow', 'orange')) +
  theme_minimal(base_size = 16)

Addendum

If you really want a 5-category Venn diagram which shows the number of species that the 5 sites have in common, you can do:

library(VennDiagram)

grid::grid.newpage()

with(sign(species_abundance[-1]),
     draw.quintuple.venn(sum(S3), sum(S5), sum(S7), sum(S9), sum(S10),
        sum(S3 == 1 & S5 == 1),  sum(S3 == 1 & S7 == 1),
        sum(S3 == 1 & S9 == 1),  sum(S3 == 1 & S10 == 1),
        sum(S5 == 1 & S7 == 1),  sum(S5 == 1 & S9 == 1),
        sum(S5 == 1 & S10 == 1), sum(S7 == 1 & S9 == 1),
        sum(S7 == 1 & S10 == 1), sum(S9 == 1 & S10 == 1),
        sum(S3 == 1 & S5 == 1 & S7 == 1),
        sum(S3 == 1 & S5 == 1 & S9 == 1),
        sum(S3 == 1 & S5 == 1 & S10 == 1),
        sum(S3 == 1 & S7 == 1 & S9 == 1),
        sum(S3 == 1 & S7 == 1 & S10 == 1),
        sum(S3 == 1 & S9 == 1 & S10 == 1),
        sum(S5 == 1 & S7 == 1 & S9 == 1),
        sum(S5 == 1 & S7 == 1 & S10 == 1),
        sum(S5 == 1 & S9 == 1 & S10 == 1),
        sum(S7 == 1 & S9 == 1 & S10 == 1),
        sum(S3 == 1 & S5 == 1 & S7 == 1 & S9 == 1),
        sum(S3 == 1 & S5 == 1 & S7 == 1 & S10 == 1),
        sum(S3 == 1 & S5 == 1 & S9 == 1 & S10 == 1),
        sum(S3 == 1 & S7 == 1 & S9 == 1 & S10 == 1),
        sum(S5 == 1 & S7 == 1 & S9 == 1 & S10 == 1),
        sum(S3 == 1 & S5 == 1 & S7 == 1 & S9 == 1 & S10 == 1),
        category = c("S3", "S5", "S7", "S9", "S10"),
        fill = c("orange", "red", "green", "blue", "yellow"),
        cex = 2,
        cat.cex = 2,
        cat.col = 'black'
))

Despite being far more difficult to read / understand, this also contains less information than the heatmap. For example, I can see from the Venn that there is a single species that only S3 and S5 have in common, but I can see that just as clearly from the heatmap. Furthermore, I can tell you the Genus (it's paracoccus), and how many observations of it there were at each site using the heatmap. You can't do this with a Venn diagram. A Venn is simply the wrong tool for presenting the data you have.

Thank you but I want to draw venn diagramm, if possible kindly help me how to draw for four stations atleast so that I can understand some basics in R — Umar, Jan 13 '23 at 16:48
@Umar a four-category Venn is not a basic thing to do in R, but I can show you how to do it. You need to understand that it _cannot_ be extended to seven sites (as in, it is mathematically impossible to draw an ellipse-based Venn with 7 categories) — Allan Cameron, Jan 13 '23 at 16:50
Thank you very much, I cant express my word for your kind help, it will help me how to edit data and to determine unique and share date within station — Umar, Jan 13 '23 at 16:53
@Umar I have used your original data set to make a quintuple Venn. Personally, I think this is just too difficult to understand. — Allan Cameron, Jan 13 '23 at 17:21
I really thank you, I will follow your suggestion, even I will try using Upset plot as well as heatmap, Once again Thank you — Umar, Jan 13 '23 at 17:30
Oh but in this I cant understand which one is unique species in different station, for example X species is unique in station 3, and y species is common in every stations.Actually I have large sets of data. So I was trying to do, — Umar, Jan 13 '23 at 17:32
we can do such job in venny2, https://bioinfogp.cnb.csic.es/tools/venny/ website but I want to do using R, — Umar, Jan 13 '23 at 17:35
@Umar Can you make the diagram you want from the data in the question using Venny and add it to your question so that I can understand better? — Allan Cameron, Jan 13 '23 at 20:49
@Umar in your sample data, there are no species that are only found at a single location. The number of species found at overlapping sites _is_ recorded correctly in the Venn. It's just difficult to interpret — Allan Cameron, Jan 13 '23 at 22:23
I edited and tried to make it simple, kindly look into it, Thanks — Umar, Jan 14 '23 at 06:06
Let us [continue this discussion in chat](https://chat.stackoverflow.com/rooms/251133/discussion-between-umar-and-allan-cameron). — Umar, Jan 14 '23 at 06:35

score 1 · Answer 2 · answered Jan 16 '23 at 13:26

You might consider my nVennR package:

library(nVennR)
species_abundance<-data.frame(Genus = c("Parasphingorhabdus", "Loktanella", "Cytobacillus", "Paracoccus", "Paucisalibacillus", "Kytococcus", "Salinibacterium", "Acinetobacter baumanni","Marinococcus","Bacillus"),
               S3 = c(0, 0, 1, 1, 0, 0, 1,0,4,0),
               S5 = c(0, 0, 0, 1, 1, 0, 1,0,3,5),
               S7 = c(3, 1, 0, 2, 0, 1, 0,0,3,1),
               S9 = c(0, 1, 0, 3, 0, 0, 0,1,2,0))

ct <- colnames(species_abundance)
ct <- ct[-1]
r <- vector("list", length = length(ct))
names(r) <- ct
for (v in ct){
  t <- species_abundance[species_abundance[,v] != 0, ]$Genus
  r[[v]] <- t
}
myV <- plotVenn(r)

You can also browse the results, as shown in the vignette:

> getVennRegion(myV, c("S7"))
[1] "Parasphingorhabdus" "Kytococcus"   
> getVennRegion(myV, c("S3", "S7", "S9", "S5"))
[1] "Paracoccus"   "Marinococcus"

Venn diagrame from contingency table in R

2 Answers2