KIRMES Source Code
Welcome to KIRMES!
Run the example to see how it works:
python kirmes.py -e -v -p Example/positives.fasta -n Example/negatives.fasta -q Example/query.fasta -r Example/resume.pck
To see all available options, run:
python kirmes.py --help
Use your own FASTA files to run KIRMES!
KIRMES consists of a series of object-oriented Python scripts. It requires several other programs to run, that are not included in any download package due to copyright and redistribution issues.
A *nix-like shell will be necessary, we have tested Mac OS X and Ubuntu Linux. Feel free to try others.
KIRMES relies on these programs and Python modules that need to be installed and available in the (Python-)path for KIRMES to run:
Python Interpreter: you will need Python 2.5, which is probably already installed, with a couple of extras:
- matplotlib (pylab)
- For the SVMs to work, you will need the large-scale machine learning toolbox Shogun, available at http://www.shogun-toolbox.org Shogun has its own requirements that need to be met, e.g. the installation of SVMLight
- For the Gibbs sampling programs supported by KIRMES, you need to download the ones you want to use (optional):
- KIRMES has a cool Galaxy interface, and you will need to install the Python-based Galaxy web service: instructions
First, install all the packages for Python.
Then install SVMLight.
Afterwards, you can download and compile the shogun source code. Be sure to use the latest version. Compile shogun with the option:
which enables the python bindings that KIRMES requires.
The command-line version of KIRMES is now ready to use.
Then you can download and install Galaxy. It needs to be checked out via their mercurial (hg) versioning system.
Follow the instructions to set up the server and then create a directory in the tools subdirectory called KIRMES. Here, you can unzip the KIRMES files and afterwards add it as a tool to the tool_conf.xml in the root directory, e.g. with the following line:
<section name="KIRMES" id="KIRMES"> <tool file="mlb/kirmes/KExamplesInstructions.xml" /> <tool file="mlb/kirmes/KKMC.xml" /> <tool file="mlb/kirmes/KIMS.xml" /> <tool file="mlb/kirmes/KIML.xml" /> <tool file="mlb/kirmes/KTraining.xml" /> </section>
Install the Gibbs samplers you wish to use and put them in your system's path variable, or directly into the KIRMES directory in the Galaxy instance.
Finally, you can start Galaxy by running:
in the Galaxy directory. As soon as it is up and running, you can connect to it via http://localhost:8080 (if you didn't change the port during setup). You should see KIRMES as one of the tools in the left-hand pane.
First Steps in Galaxy
Start by uploading a FASTA file to Galaxy, or preferably two, with a positive and a negative set. You can first run the training part of KIRMES and afterwards test it on a third set of genes via the Classifier part.