Requirements and Installation

Requirements

R2014a or later

Toolboxes: Image processing, Signal processing, Parallel computing, Statistics and machine learning

Note that MATLAB 2017a or greater is required to use NVIDIA GTX 1080 series GPUs which require CUDA v8.0.

Compute Capability 3.5 or greater is required (Kepler, Maxwell, Pascal or newer series).

GPU memory of 4 GB or greater is recommended.

Note that GTX 1080 series requires CUDA 8.0 which is supported by MATLAB 2017a or later.

~1/4 of the maximum raw recording file size

This is optional if you want to run other spike sorting software (such as Kilosort) through JRCLUST interface

Disable NVIDIA Watchdog timer Run Nsight Monitor > Nsight Monitor options (bottom right) > set "WDDM TDR enabled" to false.
(optional to compile Kilosort) Set the system path to the Microsoft C compiler (cl.exe) by editing "PATH" variable in "Environment Variables" in Windows. cl.exe is found in "C:\Program Files (x86)\Microsoft Visual Studio 12.0\vc\bin\amd64" for Visual Studio 2013 (v12)
Download the JRCLUST from GitHub. Git must be installed. Run
- git clone https://github.com/jamesjun/JRCLUST.git my_destination
- Replace "my_destination" with a desired installation directory.
You can update JRCLUST from GitHub by running
- jrc git-pull
For Mac and Linux, you needs to recompile the CUDA code by running
- jrc compile

Download sample data by running

jrc download sample
Create a new parameter file

jrc makeprm sample.bin sample.prb
This creates "sample_sample.prm" file.
Display the raw traces

jrc traces sample_sample.prm

or you can also run

jrc traces

You can omit sample_sample.prm since JRCLUST remembers the working parameter file.

Run "jrc clear" to clear the cached memory in JRCLUST.
Run spike sorting

jrc spikesort
Manual verification/sorting

jrc manual
You can clear the memory and delete all output files for a given parameter file by running

jrc clear sample_sample.prm
To test the full suite, run

jrc unit-test