Using JRCLUST

Table of contents

• Tutorial
  • Installing JRCLUST
  • Setting up your config file
  • Displaying your probe
  • Showing raw traces
  • The preview GUI
  • Detecting spikes
  • Clustering spikes
  • Curating your clustering
• Migrating from JRCLUST v3
  • What’s different
• Input and output files
  • Input files
  • Output files

Getting started

You’ve got a recording file and you’ve got MATLAB. Now what? If you’ve used an older version of JRCLUST, you might want to check out the migrating page to see what’s changed from the old JRCLUST. If you’re completely new to JRCLUST, start with the tutorial.

System requirements

Software Requirements

JRCLUST 4 was built and tested with MATLAB R2017b on Microsoft Windows 10. If you are running another operating system or an earlier version of MATLAB, JRCLUST may not work as expected. Post an issue on our issues page with the error message and we’ll get it fixed.

You will also need the CUDA Toolkit. See this link to determine which CUDA Toolkit version you will need for your version of MATLAB. It’s best to keep your GPU driver updated as well.

In addition, the following toolboxes are required:

• Parallel Computing
• Image Processing
• Signal Processing
• Statistics and Machine Learning

Hardware Requirements

An NVIDA GPU with compute capability 3.5 or later is strongly recommended. In addition, we also recommend you have at least 32 GiB of memory.

Usage

In general, usage goes like

jrc COMMAND REQUIRED\_ARG1 REQUIRED\_ARG2 [OPTIONAL\_ARGS ...]

where COMMAND is described below.

Documentation and help

• jrc help: Display a help menu.
• jrc version: Display the version number.
• jrc about: Display program information.

Pipeline commands

• jrc importv3 mysession_jrc.mat: Import a JRCv3 _jrc.mat file to the new format. Does not overwrite your old results.

• jrc bootstrap: Create a config file.

• jrc detect myparams.prm: Perform spike detection and feature extraction. (Replace myparams with a unique name of your choosing.) Output files:

  • myparams_res.mat: A MAT-file containing detection results
  • myparams_filt.jrc (filtered spike traces)
  • myparams_raw.jrc (raw spike traces)
  • myparams_features.jrc (computed features)
  • myparams_hist.jrc (operation history)

  See Input and output files for details.

• jrc sort myparams.prm: Cluster detected spike features. If JRCLUST can’t find any detected spikes, it will also perform the detect step. Your myparams_res.mat file will be updated with a clustering handle object.

• jrc detect-sort myparams.prm: Detect and cluster spikes. The difference between sort and detect-sort is that detect-sort will overwrite any previously-detected spikes, whereas sort will only perform spike detection if JRCLUST can’t find any previously-detected spikes.

• jrc recluster myparams.prm: Recluster spikes. You might want to do this if, e.g., you’ve updated the clustering threshold parameters but don’t want to recompute rho and delta.

• jrc manual myparams.prm: Open the manual curation GUI.

Preview commands

• jrc probe {myprobe.prb, myparams.prm}: Plot probe layout. You may specify either a config file or a probe file directly.
• jrc traces myparams.prm [File #]: Display traces from your raw recordings. If you have multiple recordings, you may specify the index of the file to display from.
• jrc activity myparams.prm: Plot firing rate as a function of time and depth.

Missing commands

Many other commands were available with previous versions of JRCLUST. These have been disabled because, as far as we know, they were not widely used. If you’re missing a command, post an issue on our issues page request and we will correct our error.

Getting help

Have questions? Check out our support channel on Gitter. If you run into any bugs or have any feature requests, create an issue on our GitHub issues page.