Clustering can be a very useful tool to see if your classes are defined by your feature list or only a certain number of features.
First use xcms to collect the data and fill peaks.
library(xcms)
library(faahKO)
cdfpath <- system.file("cdf", package = "faahKO")
cdffiles <- list.files(cdfpath, recursive = TRUE,full=T)
xset <- xcmsSet(cdffiles)
xsg <- group(xset)
xsg <- retcor(xsg)
xsg <- group(xsg,bw=10)
xsg <- fillPeaks(xsg)
And now for clustering and creating the dendrogram. Notice here that we have taken all of the features. You can of course create an index that will allow you to only have certain features.
## plot the dendrogram for all of the features
plot(hclust(dist(t(groupval(fill, "medret", "into")))))
## an index can be easily made and then used to restrict the number of features used in clustering.
Topic: xcms Cluster Dendrogram