[BioC] Heatmap without using clustering result

[Wolfgang Huber]

Hi Gregory

"levelplot" in the "lattice" package is a better alternative to
"graphics::image" (however, as far as I know, and I'd love to be
corrected, it does not easily integrate into layouts of multiple panels).

For straightforward image display, "display" in "EBImage" (please use
the development version,
http://www.bioconductor.org/packages/devel/bioc/html/EBImage.html) can
be efficient.

"heatmap.2" in the "gplots" package is a better alternative to
"stats::heatmap".

For even more flexible reordering of rows and columns, have a look at
the example of the "dendrogramGrob" function in the "latticeExtra" package.

	Best wishes
	Wolfgang


------------------------------------------------------------------
Wolfgang Huber  EBI/EMBL  Cambridge UK  http://www.ebi.ac.uk/huber



25/02/2009 23:59 Steve Pederson scripsit
> Hi Greg,
> 
> heatmap() is essentially a call to image() so try image(t(your.data)). I
> found you can get nice control over the colours & the overall appearance
> this way, although it does take a bit more messing around.
> 
> Cheers,
> 
> Steve
> 
> 
> gregory voisin wrote:
>> Hie,
>> I use heatmap() for expression data.
>> my data set is already ordered by Mfuzz clustering.
>>
>> If I use  my heatmap on my ordered.data.set, I get a beautiful heatmap
>> reclustering by hclust.
>>  
>> by consequently, I have 2 possibility.
>> 1. desactive hclust () in heatmap(): like this , I get a heatmap of my
>> ordered.data.set without ordering by hclust()
>>
>> 2. integrate mfuzz, in heatmap function ( heatmap ( eset, hclust =
>> function(x) mfuzz)
>>
>> After severals tries et prospection on Internet, I need your help
>>
>> Greg
>>
>>