[R] about rect.hclust

Peter Langfelder peter.langfelder at gmail.com
Sat Apr 26 06:53:08 CEST 2014


On Fri, Apr 25, 2014 at 6:35 PM, chris Jhon <cjhon217 at gmail.com> wrote:
> Hi,
>
> I am  using hclust and cutree to cluster a data frame y and cut it into few
> clusters as follows
>
> y
>   V1 V2 V3 V4
> A  1  2  3  4
> B  5  6  7  8
> C  9 10 11 12
> D 13 14 15 16
> E 17 18 19 20
>> clu<-hclust(dist(y),method="complete")
>  clu<-hclust(dist(y),method="complete")
>> clu
> Call:
> hclust(d = dist(y), method = "complete")
> Cluster method   : complete
> Distance         : euclidean
> Number of objects: 5
>> plot(clu)
>> ct<-cutree(clu,k=3)
>> rect.hclust(clu,3,border="red")
>> ct
> A B C D E
> 1 1 2 2 3
> The question is how to plot the cluster number on the dendrogram plot?

Package WGCNA (which I maintain) contains the function
plotDendroAndColors that allows you to indicate (multiple) clusters
under a dendrogram by colors, for example as in this image:

http://labs.genetics.ucla.edu/horvath/CoexpressionNetwork/BranchCutting/Example-Dendrogram-10.png

In the simplest use, you could call it as

plotDendroAndColors(clu, ct, "Clusters", main = "My clusters")

It will show cluster 1 as black, 2 as red etc. If you have more
clusters than colors in the standard pallette (which only contains 8
different colors), you can use

plotDendroAndColors(clu, labels2colors(ct), "Clusters", main = "My clusters")

Function labels2colors converts integer numeric labels to discrete
colors; the color sequence used by labels2colors can be seen by typing

standardColors()

Hope this helps,

Peter




More information about the R-help mailing list