Re: [Jtreeview-users] k-means results

[Alok S. <al...@ca...>]

I think I've figured it out.  To view k-means results, you don't open 
the Eisen1998_K_G10.kgg file, you open the Eisen1998_K_G10.cdt file. 
All that java treeview does is display white lines between the k-means 
clusters - you were probably expecting something else. It's possible, 
and indeed straightforward, to extend java treeview to do more 
interesting things with "partitioned" data sets (i.e. kmeans, SOMs, 
etc) and non-exclusive partitioning (fuzzy-k, fuzzy-c, etc). 
Essentially, you would just add annotation columns in the data 
structure corresponding to the partitioning (very easy), and then have 
a view that would make use of that annotation column (less easy, but 
straightforward). It would actually be a cleaner than what currently 
goes on in the k-means viewer, which messes with the data matrix itself 
in ugly ways.

	-Alok

On Wednesday, May 26, 2004, at 03:47  PM, Antonio Garcia wrote:

>
> Tried with your data. The pcl and cdt files are displayed fine.
>
> Upon opening the kgg file, I get the following error:
>
> Found 2468 lines...
> eweight on line 0
> parsing kgg
> error parsing KGG: LoadException 6:Wrong number of genes found! Found 
> 2467
> Expected 2466
>
> ignoring k-means clusters
> Found 1 tabs in header...
> Found 2468 total lines...
> eweight on line 0
>
>
> Ideas? A.
>
>
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> Antonio Garcia-Martinez
> UC-Berkeley Physics/Joint Genome Institute
> http://cryptologia.com
>
>         .=$=.   .=$=.           .=$=.   .=$=.
> @       @ | | | @ | | | @       @ | | | @ | | |
> | @   @ | | | @   @ | | | @   @ | | | @   @ | |
> | | @ | | | @       @ | | | @ | | | @       @ |
> ~'   `~$~'           `~$~'   `~$~'           `
>
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>
>