How to rename labels in dendrogram (dendextend)

[Lacona]

Hey all,

The labels of my dendrogram contain too much unnecessary information, so I want to change them into shorter versions. Additionally, all the labels are colored by their group affiliation.

ward <- as.dendrogram(hclust(bc_dist, method = "ward.D2"))
metad <- data.frame(phyloseq::sample_data(gps_rel_abund))
colorCode <- c(`1` = "cyan3", `2` = "tomato", `3` = "mediumorchid", `4` = "chartreuse4")
labels_colors(ward) <- colorCode[metad$group][order.dendrogram(ward)]
par(mar=c(7, 3, 1, 1))
plot(ward)

Now I want to have the labels without the first three and the last two numbers: 012.4A16 -> 4A. But without changing anything else (color, group affiliation) in the dendrogram.
I searched the web for instructions on how to do it, but or the code was too complex for me to understand it, or it changed my group-colors.

Could anyone of you please tell me if it is possible to make these changes and if so, how?

Thank you so much for your help!

[jms]

Hi,
A simple way to do this could be to change the names in your original dataset rather than trying to rename the labels. Based on your screenshot, it looks like the two digits that you want to keep are enough to uniquely identify each observation, is that correct?

In this case and if your original data has a column that contains the names, you could for example use tidyr::separate to create a column with the short names in your original data.

The functions works like this:

suppressPackageStartupMessages(
  library(tidyverse)
)

#sample data
mydf<-data.frame(name=c("012.4C16","002.1B16","010.4A16","006.2C16"))

#split up the names after the forth and sixth digit
mydf%>%
  separate(name,into=c(NA,"shortNname",NA),sep=c(4,6),remove = FALSE)
#>       name shortNname
#> 1 012.4C16         4C
#> 2 002.1B16         1B
#> 3 010.4A16         4A
#> 4 006.2C16         2C

^{Created on 2020-11-09 by the reprex package (v0.3.0)}

If you need help in how to do this in your own data, it would be good if you could post a full reproducible example.

[Lacona]

Thank you for your answer.
I will try this the next days.

[Lacona]

Because the labels are the SampleIDs I decided not to change them.
Finally, I found this solution:

# save as dendrogram
ward <- as.dendrogram(hclust(bc_dist, method = "ward.D2"))
#provide color codes to the labels
metad <- data.frame(phyloseq::sample_data(gps_rel_abund))
colorCode <- c(`1` = "cyan3", `2` = "tomato", `3` = "mediumorchid", `4` = "chartreuse4")
labels_colors(ward) <- colorCode[metad$group][order.dendrogram(ward)]
# check the current label-order and names
name(ward)
# check order and names of the labels
ward%>%labels
# rename and plot
ward %>% set("labels", c("1C31", "2A31", "1B31", "4C31", "2C31", "4A31", "2B31", "1A31", "3B31", "3C31", "3A31", "4B31")) %>% plot(main="Dendrogram day 42")