A Python toolbox for large scale Calcium Imaging data Analysis and behavioral analysis.

CaImAn implements a set of essential methods required in the analysis pipeline of large scale calcium imaging data. Fast and scalable algorithms are implemented for motion correction, source extraction, spike deconvolution, and component registration across multiple days. It is suitable for both two-photon and one-photon fluorescence microscopy data, and can be run in both batch and online modes. CaImAn also contains some routines for the analysis of behavior from video cameras. A list of features as well as relevant references can be found here.

A paper explaining most of the implementation details and benchmarking can be found here.

@article{giovannucci2018caiman,
  title={CaImAn: An open source tool for scalable Calcium Imaging data Analysis},
  author={Giovannucci, Andrea and Friedrich, Johannes and Gunn, Pat and Kalfon, Jeremie and Koay, Sue Ann and Taxidis, Jiannis and Najafi, Farzaneh and Gauthier, Jeffrey L and Zhou, Pengcheng and Tank, David W and Chklovskii, Dmitri B and Pnevmatikakis, Eftychios A},
  journal={bioRxiv},
  pages={339564},
  year={2018},
  publisher={Cold Spring Harbor Laboratory}
}

We recently refactored the code to simplify the parameter setting and usage of the various algorithms. The code now is based revolves around the following objects:

• params: A single object containing a set of dictionaries with the parameters used in all the algorithms. It can be set and changed easily and is passed into all the algorithms.
• MotionCorrect: An object for motion correction which can be used for both rigid and piece-wise rigid motion correction.
• cnmf: An object for running the CaImAn batch algorithm either in patches or not, suitable for both two-photon (CNMF) and one-photon (CNMF-E) data.
• online_cnmf: An object for running the CaImAn online (OnACID) algorithm on two-photon data with or without motion correction.
• estimates: A single object that stores the results of the algorithms (CaImAn batch, CaImAn online) in a unified way that also contains plotting methods. For an interpretation of the various entries of the estimates object see here.

To see examples of how these methods are used, please consult the demos. While the cnmf methods can also be called in the old way by passing all the parameters when initializing the cnmf object, we recommend using the params object. Similarly, to run the CaImAn online algorithm it is recommended to pass a params object inside the online_cnmf object. Older scripts should be usable with the latest version of the code except for online analysis where the cnmf object will need to be replaced with an online_cnmf object. The results should be read from estimates, i.e., cnm.estimates.C as opposed to cnm.C.

In May 2018, the way CaImAn is installed changed; we now register the package with Python's package management facilities rather than rely on people working out of the source tree. If you have an older install, these are things you should be aware of:

• You should not set PYTHONPATH to the CaImAn source directory any more. If you did this before (in your dotfiles or elsewhere) you should remove that.
• Unless you're installing with pip install -e (documented below), you should no longer work out of your checkout directory. The new install mode expects you to use caimanmanager (also documented below) to manage the demos and the place in which you'll be running code. An installed version of caimanmanager will be added to your path and should not be run out of the checkout directory.

In July 2018, Python 2.x support was removed; Python 3.6 or higher is required for CaImAn.

If you want to upgrade CaImAn (and have already used the pip installer to install it) follow the instructions given in the wiki.

Also, if you want to install new packages into your conda environment for CaImAn, it is important that you not mix conda-forge and the defaults channel; we recommend only using conda-forge. To ensure you're not mixing channels, perform the install (inside your environment) as follows:

conda install -c conda-forge --override-channels NEW_PACKAGE_NAME

You will notice that any packages installed this way will mention, in their listing, that they're from conda-forge, with none of them having a blank origin. If you fail to do this, differences between how packages are built in conda-forge versus the default conda channels may mean that some packages (e.g. OpenCV) stop working despite showing as installed.

• Download and install Anaconda or Miniconda (Python 3.6 version recommended) http://docs.continuum.io/anaconda/install

git clone https://github.com/flatironinstitute/CaImAn
cd CaImAn/
conda env create -f environment.yml -n caiman
source activate caiman
pip install .

If you want to develop code then replace the last command with

pip install -e .

Performance issues: To make the package work efficiently under any configuration ALWAYS run these commands before starting spyder (this is for Linux and OSX but environment variables can be set on Windows as well):

export MKL_NUM_THREADS=1
export OPENBLAS_NUM_THREADS=1

On Windows, replace the word "export" with the word "set" to achieve the same effect.

Once CaImAn is installed, you may want to get a working directory with code samples and datasets; pip installed a caimanmanager.py command that manages this. If you have not installed Caiman before, you can do

caimanmanager.py install

or

python caimanmanager.py install --inplace

if you used "pip install -e ."

This will place that directory under your home directory in a directory called caiman_data. If you have, some of the demos or datafiles may have changed since your last install, to follow API changes. You can check to see if they have by doing caimanmanager.py check. If they have not, you may keep using them. If they have, we recommend moving your old caiman data directory out of the way (or just remove them if you have no precious data) and doing a new data install as per above.

If you prefer to manage this information somewhere else, the CAIMAN_DATA environment variable can be set to customise it. The caimanmanager tool and other libraries will respect that.

• Increase the maximum size of your pagefile to 64G or more (http://www.tomshardware.com/faq/id-2864547/manage-virtual-memory-pagefile-windows.html ) - The Windows memmap interface is sensitive to the maximum setting and leaving it at the default can cause errors when processing larger datasets
• Download and install Anaconda (Python 3.6 recommended) http://docs.continuum.io/anaconda/install. We recommend telling conda to modify your PATH variable (it is a checkbox during Anaconda install, off by default)
• Use Conda to install git (With "conda install git") - use of another commandline git is acceptable, but may lead to issues depending on default settings
• Microsoft Build Tools for Visual Studio 2017 https://www.visualstudio.com/downloads/#build-tools-for-visual-studio-2017. Check the "Build Tools" box, and in the detailed view on the right check the "C/C++ CLI Tools" component too. The specifics of this occasionally change as Microsoft changes its products and website; you may need to go off-script.
• As of this writing, there is an upstream versioning issue between scikit and numpy on Windows. If you get a runtime error that skimage/tifffile is compiled against API version 0xc but you have 0xb, you will need to manually upgrade numpy in your conda environment (which may force other version changes).

Use the following menu item to launch a anaconda-enabled command prompt: start>programs>anaconda3>anaconda prompt

git clone  https://github.com/flatironinstitute/CaImAn
cd CaImAn
conda env create -f environment.yml -n caiman
conda install -n caiman vs2017_win-64
activate caiman
pip install . (OR pip install -e . if you want to develop code)
copy caimanmanager.py ..
conda install numba
cd ..

Then run caimanmanager as described above to make a data directory.

Things that can make your life easier on windows (but which take advanced Conda knowledge):

• Finding the keras_activate.bat script for your environment and change the KERAS_BACKEND env var from theano to tensorflow
• Remove the vs2015_compiler_vars.bat from your environment setup

Alternative environments:

• Using GPU

• Installation on Linux (Windows and MacOS are problematic with anaconda at the moment)
  • create a new environment (suggested for safety) and follow the instructions for the calcium imaging installation
  • Install spams, as explained here. Installation is not straightforward and it might take some trials to get it right

• Notebooks: The notebooks provide a simple and friendly way to get into CaImAn and understand its main characteristics. They are located in the demos/notebooks. To launch one of the jupyter notebooks:

       source activate CaImAn
       jupyter notebook --NotebookApp.iopub_data_rate_limit=1.0e10

  and select the notebook from within Jupyter's browser. The argument --NotebookApp.iopub_data_rate_limit=1.0e10 will prevent any memory issues while plotting on a notebook.

• demo files are also found in the demos/general subfolder. We suggest trying demo_pipeline.py first as it contains most of the tasks required by calcium imaging. For behavior use demo_behavior.py

• If you want to directly launch the python files, your python console still must be in the CaImAn directory.

• All diffs must be tested before asking for a pull request. Call python caimanmanager.py test from outside of your CaImAn folder to look for errors (you need to pass the path to the caimanmanager.py file).

• Andrea Giovannucci, Flatiron Institute, Simons Foundation
• Eftychios A. Pnevmatikakis, Flatiron Institute, Simons Foundation
• Johannes Friedrich, Flatiron Institute, Simons Foundation
• Pat Gunn, Flatiron Institute, Simons Foundation
• Erick, Cobos, Baylor College of Medicine
• Valentina Staneva, University of Washington
• Ben Deverett, Princeton University
• Jérémie Kalfon, University of Kent, ECE paris
• Mike Schachter, Inscopix
• Brandon Brown, UCSF

A complete list of contributors can be found here.

The following references provide the theoretical background and original code for the included methods.

If you use this code please cite the corresponding papers where original methods appeared (see References below), as well as:

[1] Giovannucci A., Friedrich J., Gunn P., Kalfon J., Koay S.A., Taxidis J., Najafi F., Gauthier J.L., Zhou P., Tank D.W., Chklovskii D.B., Pnevmatikakis E.A. (2018). CaImAn: An open source tool for scalable Calcium Imaging data Analysis. bioarXiv preprint. [paper]

[2] Pnevmatikakis, E.A., Soudry, D., Gao, Y., Machado, T., Merel, J., ... & Paninski, L. (2016). Simultaneous denoising, deconvolution, and demixing of calcium imaging data. Neuron 89(2):285-299, [paper], [Github repository].

[3] Pnevmatikakis, E.A., Gao, Y., Soudry, D., Pfau, D., Lacefield, C., ... & Paninski, L. (2014). A structured matrix factorization framework for large scale calcium imaging data analysis. arXiv preprint arXiv:1409.2903. [paper].

[4] Zhou, P., Resendez, S. L., Stuber, G. D., Kass, R. E., & Paninski, L. (2016). Efficient and accurate extraction of in vivo calcium signals from microendoscopic video data. arXiv preprint arXiv:1605.07266. [paper], [Github repository].

[5] Friedrich J. and Paninski L. Fast active set methods for online spike inference from calcium imaging. NIPS, 29:1984-1992, 2016. [paper], [Github repository].

[6] Giovannucci, A., Friedrich J., Kaufman M., Churchland A., Chklovskii D., Paninski L., & Pnevmatikakis E.A. (2017). OnACID: Online analysis of calcium imaging data in real data. NIPS 2017, pp. 2378-2388. [paper]

[7] Pnevmatikakis, E.A., and Giovannucci A. (2017). NoRMCorre: An online algorithm for piecewise rigid motion correction of calcium imaging data. Journal of Neuroscience Methods, 291:83-92 [paper], [Github repository].

[8] Giovannucci, A., Pnevmatikakis, E. A., Deverett, B., Pereira, T., Fondriest, J., Brady, M. J., ... & Masip, D. (2017). Automated gesture tracking in head-fixed mice. Journal of Neuroscience Methods, 300:184-195. [paper].

[9] Tepper, M., Giovannucci, A., and Pnevmatikakis, E (2018). Anscombe meets Hough: Noise variance stabilization via parametric model estimation. In ICASSP, 2018. [paper]. [Github repository]

The implementation of this package is developed in parallel with a MATLAB toobox, which can be found here.

Some tools that are currently available in Matlab but have been ported to CaImAn are

• MCMC spike inference
• Group LASSO initialization and spatial CNMF

A list of dependencies can be found in the environment file.

Documentation of the code can be found here. Moreover, our wiki page covers some aspects of the code. A list of known issues can be found here. If you still encounter problems please open an issue.

Please use the gitter chat room for questions and comments and create an issue for any bugs you might encounter.

Special thanks to the following people for letting us use their datasets for our various demo files:

• Weijian Yang, Darcy Peterka, Rafael Yuste, Columbia University
• Sue Ann Koay, David Tank, Princeton University
• Manolis Froudarakis, Jake Reimers, Andreas Tolias, Baylor College of Medicine

This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version.

This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.

You should have received a copy of the GNU General Public License along with this program. If not, see http://www.gnu.org/licenses/.