FAQ: MClust 3.5

Last update: 26 December 2011.

How do I load the “.t” files that MClust generates into matlab?

Use the LoadSpikes function supplied with MClust.  It can read the .t
files.  It will return a cell-array (matlab data structure) of "ts"
objects.  A "ts" object is simply a list of time stamps.  You can get
to it using the "data" function.

Try

fc = Findfiles('*.t');
S = LoadSpikes(fc);
c = Data(S{1});
plot(c,1);

Errors while calculating features on single-electrode data.

PCA and FFT do work with single electrodes. But you need to make sure you have your channel validity set correctly.  Otherwise, it will try to PCA or FFT an empty channel, which produces a singular matrix, and an error.

Errors while calculating auto-clustering.

Check to make sure that your file names do not have spaces (use “_” instead).  Also check to make sure that your file names all start with letters not numbers (use “TT2345.ntt” not “2345.ntt”).

BubbleClust and Klustakwik are too slow.

MClust requires a lot of memory and processing power.  We recommend a minimal configuration of 512MB memory and a 600MHz processor.  

Errors “Cannot find path specified” when running batch processing of auto-clustering.

MClust-3.0 had a bug which entailed missing quotation marks around certain variables so that if you had a space in the directory structure it would choke.  This has been fixed in current versions (MClust >3.01).  You can also make sure that there are no spaces in the complete directory path to your files.

Undefined function “pushdir”.

When you add the MClust folder to your path, you need to "add with subfolders".  Pushdir.m is a matlab function existing in a subfolder.