Spikesort (A MATLAB-based spike sorter for NEV files)

Maintained by the Smith Laboratory (www.smithlab.net). If you use Spikesort and find it useful (or if you have problems), please drop us an email (matt@smithlab.net).

Authors:

Ryan Kelly - wrote the original version of Spikesort in 2006 while he was a PhD student at CMU
Matthew Smith - worked with Ryan Kelly on the original version, and led development since 2011
Adam Snyder - extensive development of Spikesort from 2012-2018 while at Pitt/CMU as a postdoc
Shruti Vempati - contributed bug fixes and improvements in 2016-2017 while she was a BioE undergrad at Pitt
Xiaohan Zhong - cleaned up the code and added features in 2018 prior to the GitHub initial release

Publishing data using Spikesort:

If you use it in work you're going to publish, we would appreciate it if you would acknowledge us. You could put the software name and URL in your paper, or you could cite one or both of these references:

• Kelly RC, Smith MA, Samonds JM, Kohn A, Bonds AB, Movshon JA & Lee TS (2007) Comparison of recordings from microelectrode arrays and single electrodes in the visual cortex. Journal of Neuroscience, 27: 261-264
• Kelly RC (2010) Statistical Modeling of Spiking Activity in Large Scale Neuronal Networks, Ph.D. Thesis, Carnegie Mellon University

DISCLAIMER:

Spikesort is intended for sorting spike waveforms in NEV files (native format of Blackrock and Ripple hardware). This software modifies the NEV files and rewrites the sort codes for the channels. The authors of Spikesort take no responsibility for any errors in the software or in its use with your data. Remember that Spikesort modifies the NEV file, so you should have a backup before running this software. You should check that the sorting did what you thought it did with your NEV file. YOU ARE RESPONSIBLE FOR YOUR OWN DATA! That said, Spikesort continues to undergo development and has been used extensively. Bugs have existed in the software and will exist again, but it is relatively robust and mature.

Cloning:

This repository uses submodules. Be sure to use the following for cloning:

git clone https://github.com/SmithLabNeuro/spikesort.git --recursive

Alternatively, if cloned without the --recursive flag, once in the repository local copy, run the following to initialize the submodules:

git submodule update --init

Main File List:

genSpikeMapColor.c + mex-compiled files - take waveforms and produce an image shown in the GUI
genSpikeMapColorMatlab.m – [NOT USED] this is a slower Matlab-native code version of genSpikeMapColor.c
getSNR.m - calculate SNR (signal-to-noise ratio) - see Kelly et al 2007 (Journal of Neuroscience, 27: 261-264)
loadSort.m - load the saved sort file
manageTempFiles.m - manage temporary Spikesort files
readSampleWaveforms.m - load a subset of spikes for sorting
readWaveforms2.m - read the waveforms, times and units
readWaveforms2_timer.m - uses a timer function so that it can run in the background
readWaveforms2_timer_sparse.m - does a sparse (limited) read of waveforms for speed
rubberbandbox3.m - draw a rubber band box and return the start and end points (for selecting spikes)
rubberBandMotionFunction.m - window motion callback function
saveSort.m - save a sort to a specified location
spikesort_gui.m - define the operation callbacks
spikesort_nevscan.m - read the NEV file & get file information (called when a set of files are loaded)
spikesort_write2.m - called to write the sort codes to the NEV file
spikesort.m - spikesort main function

Utility File List:

Note: in some cases these are not standalone, and call other Spikesort functions
chanPlot.m - plot the spike waveforms for specified channel(s) with various plot options
justSNR.m - return channels list, sort codes, SNR values and spike counts (and average waveform)
moveToSortCode.m - write a list of sort codes into a NEV file
plx2nev.m - take a PLX file and convert it to NEV file with the same file name but a NEV extension
plxSortFromNEV.m - take a NEV-converted PLX file & write sort codes back into the PLX file
waves2nev.m - take waveforms, channels, sort codes, times and write them into a NEV file

Notes:

Mex Files: Spikesort uses a Mex-compiled C function (genSpikeMapColor.c) to speed up generating the waveform image for display. This function is currently compiled for several platforms. If you need to recompile, go to the Spikesort directory and type "mex genSpikeMapColor.c". Alternatively, there is a matlab- native version (genSpikeMapColorMatlab.m), but it is much slower.
NEV files: Spikesort works on NEV files, the native data format for Blackrock and Ripple hardware. If you have data from another recording setup, you can export your waveforms and convert them to a NEV using the waves2nev.m function supplied in the Spikesort directory. Also plx2nev.m converts Plexon PLX files to NEV.
Performance: Spikesort creates temporary sort files of in-progress sorts. When you first run Spikesort, it will ask you where to store these files (you can change the location in the GUI). It will create a "spikesortunits" folder that contains ch*.mat and hist*.mat files. These are the cached waveforms and the sorting history for each channel. It cleans up that directory and creates a tempsort_*.mat file on exit. The speed of Spikesort will be affected strongly by the disk access speed of the NEV file and also of this spikesortunits directory.

Usage:

For more details and a GUI tutorial, see this file.