From: Adaikalavan Ramasamy (firstname.lastname@example.org)
Date: Sun 19 Jan 2003 - 18:15:36 EST
I am attempting to replicate what Cluster 3.0 and Treeview (both by Mike
Eisen) to cluster both microarray genes and arrays does using R with
hclust. I basically utilized the plot.mat function in sma library with
some layout() and hclust().
1. Can I know if some has already written such a function or available
in some package.
2. If not, would appreciate if someone could take the time to test the
function as I only had time to test on a few datasets. Any feed back is
much appreciated. The function is called ecluster.fn and I have also
attached a simple example on how to use it.
So far my code works (attached) fine. The only problem is that subtree
are ordered in terms of the 'tightest' leftmost (according to the
details section of hclust) . So sometimes I have blocks of red, green
and then red (again) although the branches can be mannually reorder ed
so that most of the reds are to one side and most of the reds. This gets
a bit more complicated when rows are also to be ordered. I don't time to
manually reorder every dataset and want to automate the process.
3. How do we create a criterion to arrange the subtrees in such a way to
show most visual distinction. Sorry if this is a bit vague.
Thank you very much.
This archive was generated by hypermail 2.1.3 : Tue 01 Jul 2003 - 09:11:19 EST