Is there a way to create a dendrogram of arrays(samples) like the Eisen et al method used in dChip? I don't want a heatmap, just the clustering of arrays. I tried kmeans but that doesn't produce something I can plot as a dendrogram. Thanks Marc

try hc<-hclust(dist(t(exprs(Your.exprset)))) plot(hc, hang=-1, labels=A.suitable.vector) this will give you a complete clustering of samples with the euclidean distance. This is fine for rma data that is already in log2 units. You can change the clustering "method" of hclust check ?hclust, or you can use a different distance method something like 1-cor, but you will have to then convert it to a distance metric using my.metric<- 1-cor(exprs(Your.exprSet)) hc<-hclust(as.dist(my.metric)) or something like that -----Original Message----- From: mcolosim@brandeis.edu To: bioconductor@stat.math.ethz.ch Sent: 17/08/04 23:34 Subject: [BioC] Clustering like Eisen Is there a way to create a dendrogram of arrays(samples) like the Eisen et al method used in dChip? I don't want a heatmap, just the clustering of arrays. I tried kmeans but that doesn't produce something I can plot as a dendrogram. Thanks Marc _______________________________________________ Bioconductor mailing list Bioconductor@stat.math.ethz.ch https://stat.ethz.ch/mailman/listinfo/bioconductor ********************************************************************** This email and any files transmitted with it are confidentia...{{dropped}}

Hi Marc It's all in the docs :-) The functions you will be interested in are: dist() cor() as.dist() hclust() as.dendrogram() cutree() plot(..., horiz=TRUE) If you are using cor() as your distance measure, you will need to convert it from a similarity to a distance by taking 1-cor(). Also heatmap will draw you a nice dendrogram of rows without clustering the columns if you set the 3rd argument to NA. Mick -----Original Message----- From: Stephen Henderson [mailto:s.henderson@ucl.ac.uk] Sent: 18 August 2004 09:06 To: 'mcolosim@brandeis.edu '; 'bioconductor@stat.math.ethz.ch ' Subject: RE: [BioC] Clustering like Eisen try hc<-hclust(dist(t(exprs(Your.exprset)))) plot(hc, hang=-1, labels=A.suitable.vector) this will give you a complete clustering of samples with the euclidean distance. This is fine for rma data that is already in log2 units. You can change the clustering "method" of hclust check ?hclust, or you can use a different distance method something like 1-cor, but you will have to then convert it to a distance metric using my.metric<- 1-cor(exprs(Your.exprSet)) hc<-hclust(as.dist(my.metric)) or something like that -----Original Message----- From: mcolosim@brandeis.edu To: bioconductor@stat.math.ethz.ch Sent: 17/08/04 23:34 Subject: [BioC] Clustering like Eisen Is there a way to create a dendrogram of arrays(samples) like the Eisen et al method used in dChip? I don't want a heatmap, just the clustering of arrays. I tried kmeans but that doesn't produce something I can plot as a dendrogram. Thanks Marc _______________________________________________ Bioconductor mailing list Bioconductor@stat.math.ethz.ch https://stat.ethz.ch/mailman/listinfo/bioconductor ********************************************************************** This email and any files transmitted with it are confidentia...{{dropped}} _______________________________________________ Bioconductor mailing list Bioconductor@stat.math.ethz.ch https://stat.ethz.ch/mailman/listinfo/bioconductor

Marc, See ?hclust. Sean On Aug 17, 2004, at 6:34 PM, mcolosim@brandeis.edu wrote: > > Is there a way to create a dendrogram of arrays(samples) like the > Eisen et al > method used in dChip? I don't want a heatmap, just the clustering of > arrays. > > > I tried kmeans but that doesn't produce something I can plot as a > dendrogram. > > Thanks > > Marc > > _______________________________________________ > Bioconductor mailing list > Bioconductor@stat.math.ethz.ch > https://stat.ethz.ch/mailman/listinfo/bioconductor