Configuration¶

In order to use vesiclerf, users should have the following software installed and configured appropriately. A working understanding of machine learning and RAMON are helpful but not absolutely required. Their installation and configuration are documented on the linked websites:

A recent version of MATLAB [vesicle-rf] http://www.mathworks.com
CAJAL Toolbox
LONI Pipeline 6.x for distributed computing http://pipeline.loni.usc.edu
macho toolbox http://github.com/openconnectome/macho
vesicle toolbox http://github.com/openconnectome/macho
caffe and pycaffe [vesicle-cnn] http://caffe.berkeleyvision.org/
Caffe-based Object Classification and Annotation (COCA) [vesicle-cnn] https://github.com/iscoe/coca

vesicle-rf and vesicle-cnn have only been tested on mac and linux systems.

Users are encouraged to use the NeuroData AMI, which is pre-setup for vesicle. Because of limitations with the MATLAB license, please contact us for detailed instructions if you have difficulty.

Toolbox Installation Notes¶

More comprehensive documentation on each toolbox is available at the homepage for the respective repositories. Users can get setup quickly for (train,evaluate) processing by taking the following steps from within matlab:

run('/share0/cajal/tools/matlab_install/setupEnvironment.m')
run('/share0/cajal/cajal.m')
cajal.installToolbox('~/code/macho/setup.m')
cajal.installToolbox('~/code/vesicle/setup.m')

LONI Installation Notes (only needed for distributed processing)¶

The workflow has been written to be simple to use and install. The standalone MATLAB code has a driver script. To demo the LONI distributed pipeline, To use this, follow the prerequisites and setup instructions and click RUN. (estimated time to complete: 30-60 minutes prep-time and 10 minutes runtime). If you are a new user, we recommend you install code in our suggested locations so that you don’t need to modify the workflow. The workflow should “just work.” You are welcome to write to our test annotation token, or get your own.

At a high-level, users should:

Make sure that the two environment variables specified (for CAJAL and MATLAB) are on your LONI path for the vesicle package.

MATLAB_EXE_LOCATION='/Applications/MATLAB_R2014b.app/bin/matlab'
CAJAL_LOCATION='/mnt/pipeline/tools/cajal'

Adjust execution paths for modules and temp directories

Detailed LONI tips for new users¶

For LONI, we assume the following installation paths:

LONI: /Applications/LONI6
CAJAL: /share0/cajal
temp directory: /tmp

If running in LONI client mode, you may wish to add environment variables to your bash startup file to facilitate testing.

On a Mac, this is best accomplished by adding the environment variables to your ~.bashrc file. Because of the way that terminals are launched on a mac, it may be helpful to also create a ~/.bash_profile file with the following contents:

To launch LONI Pipeline to Mac specific instructions:

if [ -f ~/.bashrc ]; then
 source ~/.bashrc
fi

When using LONI in a distributed server environment, environment variables are set on launch; this is best accomplished in a local processing environment by setting environment variables as documented above. To launch LONI Pipeline in a way that includes these environment variables, run the following command:

java -cp /Applications/LONI6.app/Contents/Resources/Java/Pipeline.jar ui.gui.Main

Finally, to add the LONI modules and workflows from CAJAL to your personal LONI library, select your personal library location from LONI > Preferences > General > Personal Library Directory. More information is here. Note that you can add multiple locations by specifying symbolic links in your personal library directory.

vesicle-cnn setup¶

This material is covered in the tutorials section and the readme for each of the respective repositories.