Code to extract neuronal traces from a multi-focal microscope (COSMOS: Cortical Observation by Synchronous Multifocal Optical Sampling), and then to analyze these traces.
It is recommended that you use an Anaconda environment for running this package. The following command line instructions assume Anaconda has been installed.
git clone https://github.com/izkula/cosmos-tools.git
Navigate to the cosmos-tools folder created by git and then run:
conda env create --file cosmos3requirements.yml -n cosmostools3
source activate cosmostools3
python setup.py develop
Optionally setup ipywidgets package for jupyter notebook (only required for trace_merge_script.ipynb, and sometimes causes weird problems)
conda install -c conda-forge ipywidgets
(For some reason, ipywidgets, which is necessary only for trace_merge_script.ipynb, appears to be required to be installed separately, after the environment has already been created. There have also been some issues when removing conda environments after ipywidgets has been installed, but remaking the environments appears to eventually work.)
The CNMF-E implementation we use is written in MATLAB (a python implementation is in progress, and may be ready by the time you are reading this). This package uses the version of CNMF-E from June 27, 2018 (commit # ddb865832f52c81725859df8b5e032b4acb421e9). A MATLAB license is thus required for a subset of this codebase (specifically, only for extracting sources from raw imaging data), all subsequent steps do not require MATLAB. You can download the most up-to-date version of CNMF_E to your desired install location with
git clone https://github.com/zhoupc/CNMF_E.git
Then, open MATLAB, change to the installation directory, and run
cnmfe_setup
From the top directory (in environment cosmos3) run:
pytest
or
pytest --pdb
If using Mac OSX, then because of an issue with matplotlib you instead need to do the following:
conda install python.app
then
pythonw -m pytest
Note: There is a small possibility that due to a certain quirk in loading ome.tif files in import_raw_cosmos_data.py, you may need to use python2 for that script. It is possible the library call to load the files has now been properly upgraded for python3 and this is irrelevant. If it is necessary, you can likely just call python2 within the cosmos3 environment, or start a new environment (this has not been very well tested) with
conda env create --file install_utils/cosmos2requirements.yml -n cosmos2 python=2.7
source activate cosmos2
python setup.py develop
-
Use
import_raw_cosmos_data.py, following the directions at the top of that script. This crops the dual-lenslet ROIs, extracts timing information from the synchronization LED, uses CNMF_E to extract neural sources (requires MATLAB), and enables manual atlas alignment. -
Then use the interactive jupyter notebook:
notebooks/processing_notebooks/trace_merge_script.ipynbto align and merge traces from the different lenslets, quality control the extract sources, and save out an .h5 file containing merged traces that will be used for all further analyses of the dataset. -
You can then run
notebooks/processing_notebooks/quickly_assess_merged_traces.ipynbto quickly assess the sources and play with the traces. For general analysis, though, move the merged_traces.h5 file from processedData to ~/Dropbox/cosmos_data/[date]/[session_name]/, and add the relevant information about the session to cosmos.params.trace_analyze_params.py
-
See
notebooks/processing_notebooks/intrinsic_imaging_alignment.ipynb -
After extracting phase map from step 1, you can manually overlay the phase map with the the image that contains vasculature. This alignment of PM/V1 boundary can then be used to precisely align already-imported COSMOS data, using
scripts/adjust_atlas.py
-
Process two-photon data (i.e. visual stimulus).
scripts/batch_cnmf_2p.py -
Analyzing orientation selective visual grating stimulation.
notebooks/primary_notebooks/fig_visual_stimulation_two_photon.ipynb,notebooks/primary_notebooks/fig_visual_stimulation_cosmos.ipynb,
ipynb/primary_notebooks/DEMO_trace_analyze_script.ipynb
notebooks/primary_notebooks/classification_analysis.ipynb
notebooks/primary_notebooks/task_class_assignment.ipynb
notebooks/primary_notebooks/fig_cluster_summary_with_mr2
notebooks/primary_notebooks/fig_trace_summary.ipynb
notebooks/primary_notebooks/fig_optics.ipynb
notebooks/primary_notebooks/fig_macroscope_comparison.ipynb
notebooks/primary_notebooks/fig_classification_summary.ipynb
notebooks/primary_notebooks/vgat_inhibition_analysis.ipynb
notebooks/primary_notebooks/fig_cluster_summary_SINGLE_TRIAL-11.13.19.ipynb
matlab/scripts/trace_analysis_spont.m (for estimating background and signal photons)
matlab/SNR/dof_snr_simulations.m (for plotting simulations)
Please cite this paper when you use COSMOS in your research. Thanks!
Kauvar IV*, Machado TA*, et al. Cortical observation by synchronous multifocal optical sampling reveals widespread population encoding of actions. Neuron 2020.
Copyright 2020 Isaac Kauvar and Tim Machado
To run jupyter notebooks, on a remote computer, go to top directory and run:
jupyter notebook --port=5558 --ip=* --no-browser
See jupyter documentation to learn how to set up a password so that this is secure.
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent