[BioC] mva functions

w.huber at dkfz-heidelberg.de w.huber at dkfz-heidelberg.de
Mon Mar 24 14:14:18 MET 2003


On Mon, 24 Mar 2003, Jason Skelton wrote:
> The files I'm trying to run through hclust etc have the following sizes

Hi Jason,

You probably want to reduce the number of genes from 12488 to something
like a few hundred before clustering. With 10^4 genes the distance matrix
(which needs to be calculated as an intermediate step by hclust) has size
of 10^8, which corresponds to roughly 1 Gigabyte. I would not expect that
all 12488 rows of your data carry relevant information, esp.  if you have
only 4 conditions.

Also, it could be hard to display the cluster carpet or heatmap with more
than a few hundred rows since your display may not have that many pixels.

Best regards
Wolfgang



More information about the Bioconductor mailing list