SUSS (Semi-aUtomated Spike Sorter)

Semi-automated spike sorting for single channel electrophysiology. Designed for sorting extracellular signals on single channels with non-stationary units (waveforms changing over time).

Example of results after running automated clustering and manual cluster selection

Usage

Spike sorting

Prepare your data into arrays of spike waveforms and spike arrival times. The sorting function takes three parameters

times: 1-dimensional array of length N_spikes, each element is the arrival time of a spike in seconds

waveforms: 2-dimensional array of shape (N_spikes, N_samples), each row is a spike waveform

Usage example:

import suss.io
from suss.sort import sort

# Load your own spike dataset
data = suss.io.read_pickle(...)

# Run the sorting algorithm
sort_result = sort(data.times, data.waveforms)

# Write the output of sorting
suss.io.save_pickle(..., sort_result)

Cluster merging (curation)

The output of sort() returns 20 to 40 putative clusters in the dataset. We provide a gui tool to assist in the visual assessment of spike clusters and convenient merging and deletion of clusters.

Run the gui (from within your python environment) and load the pickle file saved from using suss.sort:

python -m suss.gui.app

Spike detection

Not implemented (yet?). Use your favorite spike thresholding / detection method.

Installation and Dependencies

This project requires Python3.5+ (I recommend using a virtualenv.)

From your python environment

(with pip)

pip install -e git+https://github.com/theunissenlab/[email protected]

# uninstall
pip uninstall suss

This should install the package as well as its depenencies

Creating a isolated python virtual environment

Separately, install python virtualenv and make sure you have python3 installed.

git clone https://github.com/theunissenlab/suss-sorter.git
cd suss-sorter
python -m venv env
source bin/activate
pip install -r requirements.txt

If you have issues with Qt (in Linux), set QT_DEBUG_PLUGINS to 1.

export QT_DEBUG_PLUGINS=1

If you have missing libraries, try commands like

sudo apt install libxcb-xinerama0 libxcb-cursor0

Other dependencies

• ffmpeg for creating HTML5 animated plots in notebooks

• imagemagick for generating animated plots in gif format

References

Dhawale AK, Poddar R, Wolff BE, Normand VA, Kopelowitz E, Olveczky BP. Automated long-term recording and analysis of neural activity in behaving animals. eLIFE, 2017

Hill DN, Mehta SB, Kleinfeld D. Quality Metrics to Accompany Spike Sorting of Extracellular Signals. JNeurosci, 2011