How do I look for shared OTUs in an abundance matrix?

Question

[Ming]

Dear all,

Let's assume I have an abundance matrix that looks like this:

             SampleA SampleB SampleC
OTU1         5               6              0
OTU2         3               1              2
OTU3         1               0              1
OTU4         0               0              2

If I have a large abundance matrix, what software or scripts will you recommend constructing a Venn diagram that can look for shared OTUs, or OTUs that are unique to each sample? The software / script should also produce an output that denotes the number of unique and shared OTUs.

Thank you!

[Jean-Karim Heriche]

For more than three sets, Euler diagrams quickly become unreadable. You may want to explore alternatives such as UpSetR.

Answer 1

[zx8754]

Transpose the dataframe/matrix so that OTUs are the sets, then convert into logical, then plot:

# example data
df1 <- read.table(text = "SampleA SampleB SampleC
OTU1         5               6              0
OTU2         3               1              2
OTU3         1               0              1
OTU4         0               0              2", header = TRUE, stringsAsFactors = FALSE)


library(gplots)

venn(t(df1 > 0))

[Jean-Karim Heriche]

Note that the venn() function from gplots is limited to a small number of sets

[zx8754]

True, not sure what OP wants, OTUs as sets or samples as sets, in their example both less than 5 sets. So 4 set Venn should be OK. Agreed, UpSet is more suitable when sets are more than 4.