Installing MNE-Python

Installing Python

MNE-Python runs within Python, and depends on several other Python packages. Starting with version 0.18, MNE-Python only supports Python version 3.5 or higher. We strongly recommend the Anaconda distribution of Python, which comes with more than 250 scientific packages pre-bundled, and includes the conda command line tool for installing new packages and managing different package sets (“environments”) for different projects.

To get started, follow the installation instructions for Anaconda.

Warning

If you have the PYTHONPATH or PYTHONHOME environment variables set, you may run into difficulty using Anaconda. See the Anaconda troubleshooting guide for more information. Note that it is easy to switch between conda-managed Python installations and the system Python installation using the conda activate and conda deactivate commands, so you may find that after adopting Anaconda it is possible (indeed, preferable) to leave PYTHONPATH and PYTHONHOME permanently unset.

When you are done, if you type the following commands in a command shell, you should see outputs similar to the following (assuming you installed conda to /home/user/anaconda3):

$ conda --version && python --version
conda 4.6.2
Python 3.6.7 :: Anaconda, Inc.
$ which python
/home/user/anaconda3/bin/python
$ which pip
/home/user/anaconda3/bin/pip

If you are on a Windows command prompt:

Most of our instructions start with $, which indicates that the commands are designed to be run from a bash command shell.

Windows command prompts do not expose the same command-line tools as bash shells, so commands like which will not work. You can test your installation in Windows cmd.exe shells with where instead:

> where python
C:\Users\user\anaconda3\python.exe
> where pip
C:\Users\user\anaconda3\Scripts\pip.exe

If you see something like:

conda: command not found

It means that your PATH variable (what the system uses to find programs) is not set properly. In a correct installation, doing:

$ echo $PATH
...:/home/user/anaconda3/bin:...

Will show the Anaconda binary path (above) somewhere in the output (probably at or near the beginning), but the command not found error suggests that it is missing.

On Linux or macOS, the installer should have put something like the following in your ~/.bashrc or ~/.bash_profile (or your .zprofile if you’re using macOS Catalina or later, where the default shell is zsh):

# >>> conda initialize >>>
# !! Contents within this block are managed by 'conda init' !!
__conda_setup= ...
...
# <<< conda initialize <<<

If this is missing, it is possible that you are not on the same shell that was used during the installation. You can verify which shell you are on by using the command:

$ echo $SHELL

If you do not find this line in the configuration file for the shell you are using (bash, zsh, tcsh, etc.), try running:

conda init

in your command shell. If your shell is not cmd.exe (Windows) or bash (Linux, macOS) you will need to pass the name of the shell to the conda init command. See conda init --help for more info and supported shells.

You can also consult the Anaconda documentation and search for Anaconda install tips (Stack Overflow results are often helpful) to fix these or other problems when conda does not work.

Installing MNE-Python and its dependencies

Once you have Python/Anaconda installed, you have a few choices for how to install MNE-Python.

For sensor-level analysis

If you only need 2D plotting capabilities with MNE-Python (i.e., most EEG/ERP or other sensor-level analyses), you can install all you need by running pip install mne in a terminal window (on Windows, use the “Anaconda Prompt” from the Start menu, or the “CMD.exe prompt” from within the Anaconda Navigator GUI). This will install MNE-Python into the “base” conda environment, which should be active by default and should already have the necessary dependencies (numpy, scipy, and matplotlib).

For 3D plotting and source analysis

If you need MNE-Python’s 3D plotting capabilities (e.g., plotting estimated source activity on a cortical surface) it is a good idea to install MNE-Python into its own virtual environment, so that the extra dependencies needed for 3D plotting stay in sync (i.e., they only get updated to versions that are compatible with MNE-Python). See the detailed instructions below for your operating system.

Download the MNE-Python environment file (done here with curl) and use it to create a new environment (named mne by default):

$ curl --remote-name https://raw.githubusercontent.com/mne-tools/mne-python/master/environment.yml
$ conda env update --file environment.yml

Installing mayavi needs OpenGL support. On debian-like systems this means installing libopengl0, i.e., sudo apt install libopengl0.

Update the base conda environment to include the nb_conda_kernels package, so you can use MNE-Python in Jupyter Notebooks launched from the Anaconda GUI. Then download the MNE-Python environment file (done here with curl) and use it to create a new environment (named mne by default):

$ conda install --name base nb_conda_kernels
$ curl --remote-name https://raw.githubusercontent.com/mne-tools/mne-python/master/environment.yml
$ conda env update --file environment.yml
  • Download the environment file

  • Open an Anaconda command prompt

  • Run conda install --name base nb_conda_kernels

  • cd to the directory where you downloaded the file

  • Run conda env update --file environment.yml

If you prefer an isolated Anaconda environment, instead of usingconda env update to modify your “base” environment, you can create a new dedicated environment (here called “mne”) with conda env create --name mne --file environment.yml.

Installing to a headless server

With pyvista: Follow the steps described in Installing MNE-Python and its dependencies but use the server environment file instead of the environment file.

With mayavi: Installing mayavi requires a running X server. If you are installing MNE-Python into a computer with no display connected to it, you can try removing mayavi from the environment.yml file before running conda env create --file environment.yml, activating the new environment, and then installing mayavi using xvfb (e.g., xvfb-run pip install mayavi). Be sure to read Mayavi’s instructions on off-screen rendering and rendering with a virtual framebuffer.

Note: if xvfb is not already installed on your server, you will need administrator privileges to install it.

Testing MNE-Python installation

To make sure MNE-Python installed itself and its dependencies correctly, type the following command in a terminal:

$ python -c "import mne; mne.sys_info()"

This should display some system information along with the versions of MNE-Python and its dependencies. Typical output looks like this:

Platform:      Linux-5.0.0-1031-gcp-x86_64-with-glibc2.2.5
Python:        3.8.1 (default, Dec 20 2019, 10:06:11)  [GCC 7.4.0]
Executable:    /home/travis/virtualenv/python3.8.1/bin/python
CPU:           x86_64: 2 cores
Memory:        7.8 GB

mne:           0.21.dev0
numpy:         1.19.0.dev0+8dfaa4a {blas=openblas, lapack=openblas}
scipy:         1.5.0.dev0+f614064
matplotlib:    3.2.1 {backend=Qt5Agg}

sklearn:       0.22.2.post1
numba:         0.49.0
nibabel:       3.1.0
cupy:          Not found
pandas:        1.0.3
dipy:          1.1.1
mayavi:        4.7.2.dev0
pyvista:       0.25.2 {pyvistaqt=0.1.0}
vtk:           9.0.0
PyQt5:         5.14.1

If you see an error like:

Traceback (most recent call last):
  File "<string>", line 1, in <module>
ModuleNotFoundError: No module named 'mne'

This suggests that your environment containing MNE-Python is not active. If you followed the setup for 3D plotting/source analysis (i.e., you installed to a new mne environment instead of the base environment) try running conda activate mne first, and try again. If this works, you might want to set your terminal to automatically activate the mne environment each time you open a terminal:

$ echo conda activate mne >> ~/.bashrc    # for bash shells
$ echo conda activate mne >> ~/.zprofile  # for zsh shells

If something else went wrong during installation and you can’t figure it out, check out the Advanced setup of MNE-Python page to see if your problem is discussed there. If not, the MNE mailing list and MNE gitter channel are good resources for troubleshooting installation problems.

Installing a Python IDE

Most users find it convenient to write and run their code in an Integrated Development Environment (IDE). Some popular choices for scientific Python development are:

  • Spyder is a free and open-source IDE developed by and for scientists who use Python. It is included by default in the base environment when you install Anaconda, and can be started from a terminal with the command spyder (or on Windows or macOS, launched from the Anaconda Navigator GUI). If you installed MNE-Python into a separate mne environment (not the base Anaconda environment), you can set up Spyder to use the mne environment automatically, by opening Spyder and navigating to Tools > Preferences > Python Interpreter > Use the following interpreter. There, paste the output of the following terminal command:

    $ conda activate mne && python -c "import sys; print(sys.executable)"
    

    It should be something like C:\Users\user\anaconda3\envs\mne\python.exe (Windows) or /Users/user/anaconda3/envs/mne/bin/python (macOS).

  • Visual Studio Code (often shortened to “VS Code” or “vscode”) is a development-focused text editor that supports many programming languages in addition to Python, includes an integrated terminal console, and has a rich ecosystem of packages to extend its capabilities. Installing Microsoft’s Python Extension is enough to get most Python users up and running. VS Code is free and open-source.

  • Atom is a text editor similar to vscode, with a package ecosystem that includes a Python IDE package as well as several packages for integrated terminals. Atom is free and open-source.

  • SublimeText is a general-purpose text editor that is fast and lightweight, and also has a rich package ecosystem. There is a package called Terminus that provides an integrated terminal console, and a (confusingly named) package called “anaconda” (found here) that provides many Python-specific features. SublimeText is free (closed-source shareware).

  • PyCharm is an IDE specifically for Python development that provides an all-in-one installation (no extension packages needed). PyCharm comes in a free “community” edition and a paid “professional” edition, and is closed-source.

Next: Installing FreeSurfer