Show significant branches in pvclust with ggplot

tidyverse
#1

I can make the following dendrogram plot showing significant branches using dendextend.

pv <- pvclust(t(mtcars), nboot = 100)
pv %>%
    as.dendrogram() %>% 
    hang.dendrogram() %>% 
    pvclust_show_signif(pv)%>%
    plot()

Rplot
Rplot800×477 15.1 KB

Can anyone suggest how to create a similar plot in ggplot2? I can extract segment data from a dendrogram, but I need significance (or au/bp values) from pvlust attached to them.

I tried ggtree and ggdendro, but with no result.

#2 
library(pvclust)
library(tidyverse)
library(dendextend)
pv <- pvclust(t(mtcars), nboot = 100)


pv %>%
  as.dendrogram() %>% 
  hang.dendrogram() %>% 
  pvclust_show_signif(pv) %>% ggplot()

This gives me a result.

#6

This topic was automatically closed 21 days after the last reply. New replies are no longer allowed.

#4

the ddextend documentation shows how to manipulate the graphics with various set commands e.g.

pv %>%
  as.dendrogram() %>%   set("labels_cex", c(.5, .7)) %>% 
  set("branches_lty", c(1, 1, 3, 1, 1, 2)) %>%
  hang.dendrogram() %>% 
  pvclust_show_signif(pv) %>% ggplot()
1 Like
#3

Thanks, I did not try the simplest thing possible! However, I'm not sure how to control the output - there are no geoms. How do I control line thickness, font size, labels etc.?

I would very much prefer to convert pvclust output into a tibble with segments, with au/bp value for each segment. Then, I can plot them the way I need (for example, I need faceting multiple dendrograms in my case).

#5

I just discovered that I can use as.ggend after pvclust_show_signif() it gives me exactly the data I need:

ggd <- pv %>%
  as.dendrogram() %>%
  hang.dendrogram() %>% 
  pvclust_show_signif(pv) %>%
  as.ggdend()

ggd$segments contains coordinates of each segment and its line width, so I can use it in my own ggplot call.

Thanks for pointing me towards the solution!