Difference between revisions of "Software Setup"

Revision as of 00:35, 11 June 2016

We rely on a number of different software packages to analyze MRI and MEG data. Here is a growing tally of what we use and where to get it. Most of these tools reside on github and you should definitely use git to clone the repositories rather than download a snapshot of the code.

Setting up a user account

To add a new user with sudo privileges:

sudo useradd -m -G bde,sudo -s /bin/bash userid

To set user password.

sudo passwd

Have new user enter desired password.

Git

All git directories should be maintained together.

mkdir ~/git

Yeatman Lab Tools

cd ~/git
git clone https://github.com/yeatmanlab/BrainTools.git

Vistasoft

MATLAB based toolbox, from Brian Wandell's lab at Stanford, that contains many functions we rely on for analyzing diffusion MRI and functional MRI data

cd ~/git
git clone https://github.com/vistalab/vistasoft.git

Automated Fiber Quantification

cd ~/git
git clone https://github.com/jyeatman/AFQ.git

Setting up Matlab for AFQ

AFQ requires Matlab r2014a, SPM8, Vistasoft, and AFQ. If you wish to use a different version of Matlab for your core coding, do not create a /usr/local/bin file during installation. Instead, after installing r2014a to the default location.

cd /usr/local/bin
sudo ln -s /usr/local/MATLAB/R2014a/bin/matlab AFQ-matlab
sudo AFQ-matlab

In the Matlab terminal that opens,

cd /usr/local/MATLAB/R2014a/toolbox/local
edit startup

Paste the following code into the new file.

addpath(genpath('~/git/AFQ/'))
addpath(genpath('~/git/BrainTools/'))
addpath(genpath('~/git/vistasoft/'))
addpath(genpath('~/Documents/MATLAB/spm8/'))

Save the file

If all the requisite dependencies for AFQ have been installed according the instructions on this wiki, you can now execute

AFQ-matlab

from the terminal in order to launch Matlab r2014a with all the necessary dependencies.

Anaconda

Each user should have anaconda set up to manage their Python packages. Do this before installing nibabel and dipy. Anaconda will manage versions and dependencies for each user separately. Each install will be specific to the user that installs it. Do not install as root, as you will deny yourself privileges to your own directory. You must restart your terminal session after install before using Python or the conda command. See instructions here:

https://www.continuum.io/downloads

nibabel and DIPY

Python based toolbox for dealing with nifti images. While nibabel is on github we suggest installing using pip:

pip install nibabel
pip install dipy

Neurodebian repo

The neurodebian project is an incredible resource making it easy to install most of the widely used nueroimaging software packages.

wget -O- http://neuro.debian.net/lists/trusty.us-nh.full | sudo tee /etc/apt/sources.list.d/neurodebian.sources.list
sudo apt-key adv --recv-keys --keyserver hkp://pgp.mit.edu:80 0xA5D32F012649A5A9
sudo apt-get update

FSL, ANTS and other Neurodebian packages

Once you have added the neurodebian repos you can easily install fsl and other packages

sudo apt-get install fsl-5.0-complete
sudo apt-get install fsl-5.0-eddy-nonfree
sudo apt-get install ants

FSL requires an edit to the .bashrc file

gedit ~/.bashrc

Copy the following into the file that open

# FSL setup
. /etc/fsl/5.0/fsl.sh

SPM8

Fill out the form located here in order, making sure to select SPM8. Run this script from the terminal to install.

unzip ~/Downloads/spm8.zip -d ~/Documents/MATLAB/

Psychtoolbox for Matlab

Requires Neurodebian repo to install.

sudo apt-get install matlab-psychtoolbox-3

In order to launch Matlab with Psychtoolbox, use terminal command

ptb3-matlab