J-Express forum
December 07, 2019, 03:33:43 PM *
Welcome, Guest. Please login or register.

Login with username, password and session length
News: J-Express 2011 released. Download from jexpress.bioinfo.no
 
   Home   Help Search Login Register  
Pages: [1]
  Print  
Author Topic: RMA?  (Read 8971 times)
christopher
Newbie
*
Posts: 2


View Profile
« on: September 28, 2005, 10:34:39 AM »

When will JExpress support RMA? I find the current tools (R, RMAExpress) to cumbersome to use, and would really like to take advantage of the other tools, especially the visualization features, of JExpress, in combination with RMA.
Logged
Narve Sætre
Newbie
*
Posts: 5


View Profile Email
« Reply #1 on: September 28, 2005, 10:36:32 AM »


Hi,

we are currently working on the RMA functionality. The implementation is more or less complete, and it is included in the latest beta version (available for Norwegian users). We need to do some more testing and data verification before we can publish the final version, but stay tuned, it will be here soon.
Logged
Narve Sætre
Newbie
*
Posts: 5


View Profile Email
« Reply #2 on: September 28, 2005, 10:42:06 AM »


By the way, we are also doing some performance work, especially on the memory requirements for huge datasets. I would be interesting to hear what kind of data you're working on: The number of datasets and the size of them. As you probably know, RMA computations can be quite memory intensive, but we're doing some tuning right now and so far the results are looking good.
Logged
christopher
Newbie
*
Posts: 2


View Profile
« Reply #3 on: September 28, 2005, 10:52:17 AM »


Great news, keep up the good work! My typical dataset is 40-60 datasets, 600-1000x600-1000. Hopefully I can run this on my desktop computer, so I don't have to bug the IT department and have them run the program on their precious big servers...

By the way, a dataset of the size given above takes about 500-1000 MB of disk space. Of course, the files are usually transferred in a compressed format. The same data in a zip-file takes about 150MB. It would be nice if JExpress could read compressed files directly, that would save up both space and time. This is of course assuming that reading compressed files doesn't slow down the computations. I don't see why the major software vendors haven't implemented this simple but useful feature.
Logged
Narve Sætre
Newbie
*
Posts: 5


View Profile Email
« Reply #4 on: September 28, 2005, 11:01:26 AM »


Thanks for your input, christopher. You should be able to run those datasets on a "normal" desktop (1GB RAM, 1+GHz processor). In fact I have used datasets of approx. that size on my laptop.

Regarding compressed files: This feature is already partially implemented in JExpress: After reading a CDF / CEL file for the first time, JExpress automatically saves the dataset in a custom compressed format (typically a quarter of the original size). Next time you use the same dataset, JExpress will use the compressed files. This greatly improves performance on large datasets. Currently you need to have the original CDF/CEL file, but if this is a popular feature request we can make JExpress work without the original CDF/CEL files.

Note that we don't currently support reading original compressed files directly. This is an interesting request; I can see that JExpress could be able to read CDF.gz files directly. This will be considered for a future release. The performance should be about the same or even higher, as hard disks quite often is the bottleneck in these days.
Logged
Kristian Flikka
Newbie
*
Posts: 1


View Profile
« Reply #5 on: September 29, 2005, 09:47:12 AM »

Hmm, why not just include reading original compressed files it in the upcoming 2.7 version? If it is so easy, I mean... Grin
Logged
Narve Sætre
Newbie
*
Posts: 5


View Profile Email
« Reply #6 on: September 29, 2005, 10:12:12 AM »

Hehe, well the code to read a .gz file is easy. However, we need some interface changes, and decide on how to handle file extensions, and what to do if there is a .CDF and a .CDF.GZ file in the same directory (which to use) etc. So I doubt this will be in release 2.7, but perhaps the next release Smiley
Logged
Pages: [1]
  Print  
 
Jump to:  

Powered by MySQL Powered by PHP Powered by SMF 1.1.14 | SMF © 2006-2008, Simple Machines LLC Valid XHTML 1.0! Valid CSS!