How to set-up Python for this hackathon
There are many ways to install Python, but we recommend installing Python and Python-packages through the Anaconda distribution, which has installers for both Windows and Mac (and Linux). Note that Mac (and Linux) probably already have Python installed, but we still recommend installing Anaconda to avoid any package/dependency issues that we may come across later.
Download Anaconda
Anyway, go to the Anaconda website and click on the download button corresponding to your operating system. Then, choose the Python 3.6 version of the installer (and choose between the 64 or 32 bit version; probably 64-bit). For Windows, after downloading the .exe, just follow the installation instructions.
For Mac, you can choose between the graphical installer or the command-line installer; choose whatever you like best.
During the install, the Anaconda installer is going to ask for a few configurations, like where to install Anaconda and such. Importantly, when the installer asks:
*“Do you wish the installer to prepend the Anaconda3 install location to PATH in your /home/
Install packages
If you installed Anaconda through the command line, close and reopen a terminal, which makes sure that the $PATH variable is reloaded and your OS finds the correct (Anaconda) version of Python. Now, we need to install the packages necessary for the tutorials. These packages encompass different functionality that we need to load, manipulate, and analyze MRI-images. We will use the following:
- numpy: the standard package for working with numeric data in Python (e.g., manipulating n-dimensional arrays like MRI-images!);
- scipy: provides some functionality for statistics (like the t-test);
- nibabel: the de facto package for loading in MRI data;
- scikit-learn: the extremely versatile (and most popular) machine learning package in Python;
- matplotlib: the most-often used plotting package in Python;
- skbold: our own machine-learning-for-fMRI package (an optional section on skbold is included in the workshop)
- niwidgets: this lets you look at image files inside jupyter notebook.
To install these packages, we’re going to use pip, Python’s package installer (which is shipped by default with Anaconda). The command pip works in a terminal/command line, so Mac/Linux users should simply open a terminal window. Windows users can open a command prompt either by opening the cmd utility (search for “cmd” in your programs) or by using the Anaconda command prompt (a Linux-style terminal emulator; search for “Anaconda” in your programs).
Once you’ve opened a terminal, navigate to the directory with materials, and run: use the following command to install all the packages at once:
$ pip install -r requirements.txt
Alternatively (if it for some reason doesn’t work), you can install the packages one by one by running:
$ pip install <package name>
So, to install nibabel, just type: pip install nibabel. Note: scikit-learn can be installed by doing: pip install sklearn. Also, niwidgets needs to be installed from the corresponding Github directory directly, which can be done through:
$ pip install git+git://github.com/janfreyberg/niwidgets
If you want to check whether all packages are installed correctly, you can run the verify_package_installs.py script located in the materials folder (run python verify_package_installs.py in a terminal). You might get a “DeprecationWarning”, but you can ignore that!