Installation

Download by using

git clone https://github.com/JulesLebert/spikesorting_scripts.git

Install all the dependencies by using cd to this directory and then use

pip install -e .

All the spike sorters have to be installed separately (see https://spikeinterface.readthedocs.io/en/latest/install_sorters.html)

Running spikesorting on concatenated recordings

Edit the parameters of the spikesorting in concatenated_recordings_params.json

Navigate in scripts/ and run

python spikesorting_concatenated_NP.py json_files/concatenated_recordings_params.json

The jobs folder contains job files to run on the UCL cluster myriad (https://www.rc.ucl.ac.uk/docs/Clusters/Myriad/)

Installation of pykilosort on myriad (UCL clusters)

I highly recommand to use conda for pykilosort on myriad as cupy can be complicated to install without

If a python module is loaded, unload it by using

module unload python

And load python with conda using

module load python/miniconda3/4.10.3
source $UCL_CONDA_PATH/etc/profile.d/conda.sh

Create a new conda environment

conda create --name ibl_pykil_ss python=3.9
conda activate ibl_pykil_ss

And install dependencies

conda install cupy
pip install phylib pypandoc
# recommended
git clone --branch ibl_prod https://github.com/int-brain-lab/pykilosort

cd pykilosort
pip install -r requirements.txt
pip install -e .

And finally install the spikesorting_scripts package following the instructions under Installation

~~/!\ As of today (23/01/2023 version ibl_1.4.1) there is a typo in pykilosort that will produce an error when exporting data to phy~~

Pull request with proposed changed merged (https://github.com/int-brain-lab/pykilosort/commit/dae7d4b3d815b701c5c9f797848bd4340f338d4e)

Submit the job to myriad

Example of job running pykilosort in jobs/run_single_NP_pykilosort.sh