[BioC] heatmap and hcluster from limma out put object

[Gordon Smyth]

>Date: Tue, 16 Aug 2005 18:01:40 -0300
>From: Marcelo Luiz de Laia <mlaia at fcav.unesp.br>
>Subject: [BioC] heatmap and hcluster from limma out put object
>To: bioconductor <bioconductor at stat.math.ethz.ch>
>
>Hi Bioconductor's Users,
>
>I found in the list archives a step-by-step for to do a heatmap with a
>hclust.
>
>It is in: http://files.protsuggest.org/biocond/html/1127.html
>
>///I tested it and I modifield it to my code, like this:
>
>  [1]dist.x <- dist(exprs(todos.norm),method="euclidean",diag=TRUE)
>  clust.x <- hclust(dist.x, method="average")
>
>  [2]dist.y <- dist(t(exprs(todos.norm)), method="euclidean",diag=TRUE)
>  clust.y <- hclust(dist.y,method="average")
>
>  heatmap(exprs(todos.norm), Rowv=as.dendrogram(clust.x),
>Colv=as.dendrogram(clust.y),
>  cexRow=1, cexCol=1,col=greenred.colors(255))
>  # function greenred is (....).
>
>  x1 <- exprs(todos.norm[1:50,])
>  x2 <- exprs(todos.norm[51:100,])
>  heatmap(x1, cexRow=1, cexCol=1,col=greenred.colors(255))
>
>It works better from normalized data set, but, I would like to run it on
>eBayes object from limma. Is it possible?

You mean you want to cluster contrasts, or to cluster genes based on the 
coefficient estimates? You could do this roughly by replacing 
'exprs(todos.norm)' by fit$coef.

Gordon

>  I would like to see my all
>data, because I have a lot off methabolics pathway and we is very
>curious to see if theses pathways are grouped together.
>
>I suppose that is not need to calculate [1]!??
>
>Thanks
>
>Marcelo