<div class="gmail_quote"><div style=""><div><div>Hi Joachim,<br><br>On 6 April 2009 07:53:30 BST, Joachim Beig <<a href="mailto:beigjoach@fh-bingen.de">beigjoach@fh-bingen.de</a>> wrote:<br>
</div><blockquote type="cite"><div>Dear
Sir or Madam, my name is Joachim Beig, I'm a German student of
bioinformatics at the University Of Applied Sciences, Fachhochschule
Bingen, Germany. At the moment I'm working on my diploma thesis which
is about Feature Extraction of Microscopy Images, and therefore I want
to install the OME-Server to check out the WND-CHARM software packet.
Sadly, it seems that there was a change in the CURL-library, so a
environment-variable called "CURL_MUTE" isn't found from the
installation routine and I'm not able to go further. (I tried to
install OME Server on Ubuntu 8.10 and Fedora 10, unfortunately I have
no Mac or Fedora 4...) So I want to ask you if there is a chance to get
the source-code of this packages to hava just a look on it? I think
it's just a Matlab-Code, isn't it? Or is there a way to get support for
installing the OME-Server? I'm looking forward to hearing from you!
Best regards! Joachim Beig<br></div></blockquote></div></div></div>
<br>I suggest you give the standalone WND-CHARM written in C a try. Ilya Goldberg recently commented on using WND-CHARM, both the standalone version and the one integrated with OME Server. I have included his instructions below. If you have further questions, feel free to write again to the OME users list.<br>
<br>HTH,<br>Curtis<br><br><div class="gmail_quote"><div><div class="h5"><div class="gmail_quote"><div bgcolor="#ffffff" text="#000000">
Ilya Goldberg wrote:
<blockquote type="cite">It depends what you want to do. There are two options right now: <br>
1) run WND-CHARM to see if it works for your particular experimental
system (i.e. use just the pattern recognition algorithm) <br>
To do this all you need is the wndchrm utility which you can download
here: <br>
<a href="http://ome.grc.nia.nih.gov/wnd-charm/wndchrm-1-26.tar.gz" target="_blank">http://ome.grc.nia.nih.gov/wnd-charm/wndchrm-1-26.tar.gz</a> <br>
The only requirements for this utility are the FFTW library and the
libtiff library. <br>
LibTIFF: <a href="http://www.libtiff.org/" target="_blank">http://www.libtiff.org</a> <br>
FFTW 3.1: <a href="http://www.fftw.org/download.html" target="_blank">http://www.fftw.org/download.html</a> <br>
You don't need any OME or OMERO bits to run this. <br>
<br>
The utility is installed using the regular "./configure; make; sudo
make install" set of commands. It will check for the presence of its
required libraries and fail during the configure step if it can't find
them.
<br>
There is a fairly detailed usage manual for this utility published
here: <br>
<a href="http://www.scfbm.org/content/3/1/13" target="_blank">http://www.scfbm.org/content/3/1/13</a> <br>
You can also get a usage message by just executing wndchrm without any
parameters. <br>
Very briefly, it relies on a set of folders full of images, each folder
defining a training class. It trains itself on the set of folders you
provide, and generates a set of HTML reports describing how well it can
distinguish the different classes, their relative similarities, etc.
<br>
<br>
This utility is designed to have many instances of itself running
concurrently on the same set of images. We use it in the lab on a
cluster of a dozen computers each with 2-8 cores, all sharing the same
drive. Then we launch the utility with the same command line as many
times as there are cores/CPUs available. In other words, its not just
a demo utility - we use it regularly in the lab to do science.
<br>
This is all very manual, but its pretty straight-forward to set up and
get running without a lot of conceptual overhead. <br>
<br>
2) Set up OME to run with MATLAB and the MATLAB-based WND-CHARM. <br>
This is quite a bit more involved, but I can help you through this if
that's what you want to do. This type of system is more geared to
people who are using MATLAB to develop new image analysis algorithms.
It can also be used to run WND-CHARM on a cluster, and we've used it in
this mode quite a bit. The problem with this set up is that MATLAB
likes changing its APIs with every minor version release (every 6
months or so), and its become increasingly difficult to keep up.
Probably the future is to use the more stable Java API for MATLAB and
do this through OMERO. MATLAB's various DRM schemes and licensing
shenanigans are making this a less attractive way to go with each
passing release. Still, its heavily used and is well supported by
OMERO, with "best effort" support by OME.
<br>
<br>
The third option, which is not available right now, is to get a
complete, integrated OME/OMERO/WND-CHARM in an easily installable tidy
package without MATLAB requirements.
<br>
That's obviously what we want to provide as soon as we can, but we're
not there yet. <br>
<br>
You can contact Lior or myself if we can help you further. <br>
</blockquote></div></div></div></div></div>
<br>