-
Notifications
You must be signed in to change notification settings - Fork 5
Installation
Installation of the new Connectome Mapper should be simpler thanks to a debian package download for Ubuntu versions >=11.10. For older versions of Ubuntu / different distribution, please look at the manual installation. Prerequesites are needed anyway, for all the installation choices.
- Installed and configured version of Freesurfer (http://surfer.nmr.mgh.harvard.edu/). $FREESURFER_HOME should have been declared and the Freesurfer setup script should have been sourced as described here: http://surfer.nmr.mgh.harvard.edu/fswiki/SetupConfiguration. You can check if it is the case by looking for SUBJECTS_DIR in your exports variable (type export on the command line).
- Installed and configured version of FSL (http://fsl.fmrib.ox.ac.uk/fsl/fslwiki/). $FSLDIR should have been declared and the FSL setup script should have been sourced as described under "Installation" here: http://fsl.fmrib.ox.ac.uk/fsl/fslwiki/FSL%20FAQ
- Installed version of Diffusion Toolkit. Diffusion toolkit executables ('dtk', 'odf_recon', ...) should be in the $PATH, and $DSI_PATH needs to be set to the folder containing the diffusion matrices.
Installation is composed of a debian package file (cmp_2.x.x_all.deb, containing the python cmp and cmtklib packages) and compiled binaries (32/64 bit versions available).
- Install the .deb package with the Ubuntu Software Center (default if you double click on the package on Ubuntu) or using the dpkg command (sudo dpkg -i cmp_2.x.x_all.deb). This will install all the needed dependencies.
- Install the compiled binaries needed by the Connectome Mapper by putting them somewhere in the PATH (e.g. copy all the files of the archive to /usr/local/bin)
- Install our forked version of Nipype (http://nipy.sourceforge.net/nipype/). For now, we require a modified vesion Nipype interfaces that is available on our Github repository (https://github.com/LTS5/nipype). To install it clone to your machine the nipype fork by typing
git clone git://github.com/LTS5/nipype.git, and run the install script withsudo python setup.py install. You will have to remove already installed versions of nipype if they were installed through apt-get (installation location:/usr/lib/pyshared) as it will take precedence over versions installed through the setup.py script.
Manual installation is divided between the Python libraries needed by the Connectome Mapper and the CMTKlib and the libraries needed by the DTB binaries. Files for manual installation is the zipped archive of the Connectome Mapper.
- As we will use
easy_installin order to have access to the latest libraries even on older systems the python-setuptools package is needed. Ipython is strongly recommended for debugging purposes. Debian/Ubuntu command:sudo apt-get install python-setuptools ipython - Python libraries needed: traits, traitsui, pyface, nibabel, numpy, networkx, scipy. Easy_install command:
sudo easy_install traits traitsui pyface nibabel numpy networkx scipy nose - Install our forked version of Nipype (http://nipy.sourceforge.net/nipype/). For now, we require a modified vesion Nipype interfaces that is available on our Github repository (https://github.com/LTS5/nipype). To install it clone to your machine the nipype fork by typing
git clone git://github.com/LTS5/nipype.git, and run the install script withsudo python setup.py install. You will have to remove already installed versions of nipype if they were installed through apt-get (installation location:/usr/lib/pyshared) as it will take precedence over versions installed through the setup.py script. - Libraries needed by the DTB binaries: boost (module program-options), nifti, blitz. Debian/Ubuntu command: sudo apt-get install libboost-program-options-dev libnifti-dev libblitz0-dev
- Extract the source code and install the Connectome Mapper from the Bash Shell using following commands:
tar xzf LTS5-....tar.gz,cd LTS5-cmp-..../,sudo python setup.py install
If some libraries are missing the Connectome Mapper will refuse to launch. The error message should explain which binaries/libraries are missing. In case of DTB binaries problems, try recompiling the DTB sources under /src.