在R中绘制漂亮且对齐的彩色树状图

Question

I am using the Sparcl package ( https://cran.r-project.org/web/packages/sparcl/sparcl.pdf ) to plot dendrograms in R. In my specific problem, I am clustering the groups according to one criterion, and I want to visualize by coloring based on another criterion (the point of this is to show that the cluster coincides (or does not), with another characteristic. I have been able to do this with the Sparcl package, to highlight the nodes that I want to emphasize:

df <- read.delim("the_data_matrix.txt");
d <- dist(as.matrix(df))
hc = hclust(d)
y[]='black'
y[list_of_nodes$V1]='red' # This will allow me to color only certain branches red, leaving the others black

If I plot with the standard plotting function, I can control various parameters, such as labels and text size with hang and cex (but cannot color any branches) (In the picture this is "Dendrogram 1"):

plot(hc,hang=-10,cex=.1)

On the other hand, if I plot using the ColorDendrogram function within Sparcl, I can get a colored dendrogram, but lose formatting options (In the picture this is "Dendrogram 2"):

ColorDendrogram(hc, y = y, branchlength = 4)

ColorDendrogram gave me errors when I used hang and cex to control text size and placement. My Question

Does anyone know how to fix this, either within the Sparcl package or another one? I would like to have flexibility of color that ColorDendrogram has, but not lose formatting capabilities.

Answer 1

Try the package dendextend ( vignette ), which should give you all flexibility:

library(dendextend)
d1 <- mtcars %>% dist %>% hclust %>% as.dendrogram
d2 <- mtcars %>% dist(method="minkowski") %>% hclust(method="single") %>% as.dendrogram
vals <- grep("Merc", rownames(mtcars), val=T) # highlight branches leading to "Merc..."

par(mfrow=c(2, 1))
d1 %>% set("by_labels_branches_col", value = vals) %>% set("hang_leaves", -10) %>% set("labels_cex", .1) %>% plot
d2 %>% set("by_labels_branches_col", value = vals) %>% plot