Materials for the workshop on CaImAn and Mesmerize for Calcium & Voltage Imaging. This repo currently includes installation instructions, we will add more materials as the workshop approaches.
This workshop will cover the following topics:
fastplotlibfor very fast interactive visualization of very large datasets in notebooks.- Preprocessing of calcium imaging data
- Various
CaImAnalgorithms including:- NoRMCorre motion correction
- CNMF(E) - Constrained Non-negative Matrix Factorization for source separation and extraction
- OnACID and FIOLA for online calcium imaging analysis
- VolPy for voltage imaging analysis
mesmerize-corefor parameter optimization and data organization with CaImAn.- Postprocessing
Fork and clone this repo to follow along with the demos.
- Fork this repo to your GitHub:
- Clone the fork to your local:
Run this in your terminal, anaconda promp or mambaforge prompt:
-
To run the demos launch
jupyter labfrom within your environment. -
Download the demo movie from the following link and place it within your
caiman_data/example_moviesdir.
https://caiman.flatironinstitute.org/~neuro/caiman_downloadables/Sue_2x_3000_40_-46.tif
If you are starting from scratch, start here. If you just need to update, then see the Update instructions below.
We recommend installing the mesmerize-core conda package which gives you everything that you need, including caiman. fastplotlib is installed separately but that is relatively trivial.
- Install Anaconda if you already don't have it. Click here for official instructions on the Anaconda website. A few important notes:
- On Windows and Mac if you use the graphical installer make sure you choose to install "Just Me" and not a system install.
- On Linux you can probably accept the default
PREFIXto install to your user home directory, do not perform a system install.
All the following commands are to be entered into the Anaconda prompt on Window, or in the terminal on Linux and Mac.
- Install
mambainto your base environment. Skip this step if you havemamba. This step may take 10 minutes and display several messages like "Solving environment: failed with..." but it should eventually installmamba.
conda install -c conda-forge mamba
# if conda is behaving slow, this command can sometimes help
conda clean -a- Important note: In general, sometimes conda or mamba will get stuck at a step, such as creating an environment or installing a package. Pressing
Enteron your keyboard can sometimes help it continue when it hangs like this.
- To create a new environment and install
mesmerize-coreinto it do this:
mamba create -n mescore -c conda-forge mesmerize-corecaiman is a dependency of mesmerize-core so it will automatically grab caiman too
If you already have an environment with caiman:
mamba install -n name-of-env-with-caiman mesmerize-core- Activate environment. You can only use
mesmerize-corein the environment that it's installed into.
mamba activate mescore- Install
caimanmanager
caimanmanager.py install- The
caimanmanager.pystep may cause issues, especially on Windows. Assuming your anaconda is in your user directory a workaround is to call it using the full path:
python C:\Users\your-username\anaconda3\envs\your-env-name\bin\caimanmanager.py install- If you continue to have issues with this step, please post an issue on the
CaImAngithub or post in the#caimanchannel on slack.
- Run
ipythonand verify thatcaimanandmesmerize_coreare installed (note to run ipython just enteripythonin your anaconda prompt):
# run in ipython
import caiman
import mesmerize_core
print(caiman.__version__) # should be 1.9.13, anything over 1.9.10 is mostly fine for the workshop but we recommend 1.9.13
print(mesmerize_core.__version__) # should be 0.1.0, make sure it's not the 0.1.0.b1 beta version- Install
fastplotlibfor visualization into the same environment (run this in the anaconda prompt, not ipython)
pip install fastplotlib(The previous instructions said to install it directly from github but we were able to make a pip-installable release since pygfx recently updated their release, and pygfx is the main dependency for fastplotlib)
If you don't have git installed you will need to install that first in the environment:
conda install git-
If you have C compilers installed on your system we recommend
pip install simplejpeg. This is usually easier on Linux & Mac than on Windows. If you cannot installsimplejpegdon't worry, it will just makefastplotlibslightly less fast. -
Install Vulkan drivers for your GPU (this includes GPUs that are commonly integrated within CPUs).
- Windows: Vulkan drivers should be installed by default so you shouldn't have to do anything, if you can run the first few cells of the
fastplotlibsimple.ipynbexample you're good to go! - Mac uses Metal instead of Vulkan, which should also be installed by default so you don't have to do anything!
- Linux: This is the one time where you need to do more work on Linux :D. See the
fastplotlibrepo for instructions.
- Windows: Vulkan drivers should be installed by default so you shouldn't have to do anything, if you can run the first few cells of the
-
Finally we recommend trying to run the simplest demo notebook for each library:
fastplotlib: https://github.com/kushalkolar/fastplotlib/blob/master/examples/simple.ipynbcaiman: https://github.com/flatironinstitute/CaImAn/blob/master/demos/notebooks/demo_pipeline.ipynbmesmerize-core: https://github.com/nel-lab/mesmerize-core/blob/master/notebooks/mcorr_cnmf.ipynb
You can either clone each repo to try out the demo notebooks or download the repo as a zip file:
Note when new releases are available for mesmerize-core, run the following from within your virtual environment to update: mamba update -c conda-forge mesmerize-core. To update fastplotlib, use the following command (from within your activate virtual environment): pip install --upgrade fastplotlib