ben-willmore / spikemonger Goto Github PK

== Spikemonger

= Getting spikemonger

Spikemonger is available from https://github.com/beniamino38/spikemonger .
You can get it using 'git' (a version control system) or by just downloading a
.zip/.tar.gz file ('Downloads' link).


= Machine

Spikemonger needs a beefy machine with lots of RAM (8GB) and lots of disk 
space (budget 3x the amount of space consumed by your raw data). You also 
need a vaguely recent version of Matlab.


= Data files

Spikemonger works on .bwvt files (produced by BrainWare) or .f32 files 
(produced by BenWare). You will need to organise your data files into the 
structure expected by spikemonger:

<experiment_name>/<subdirectory>/raw.f32/<prefix>.N.f32

or

<experiment_name>/<subdirectory>/raw.bwvt/<prefix>.N.bwvt

where <experiment_name>, <subdirectory> can be whatever you like; <prefix> 
should start with 'P'.

All files that should be clustered together should be in the same directory. 
Spikemonger will assume that consecutively numbered files are from 
neighbouring sites.

For example, for a bilateral recording, you want:

F1002/P02.whitenoise.L/raw.f32/P02.whitenoise.1.f32
F1002/P02.whitenoise.L/raw.f32/P02.whitenoise.2.f32
...
F1002/P02.whitenoise.R/raw.f32/P02.whitenoise.17.f32
F1002/P02.whitenoise.R/raw.f32/P02.whitenoise.18.f32
...


= Running spikemonger

There are two stages: automated clustering, and manual cluster cutting. 
Normally, you will want to set the automated clustering running overnight on
all your data, and then return when it is done to cluster cut all the data.

* Automated clustering:

-- spikemonger_batch
-- edit the root_dir line at the start, then run spikemonger_batch

If you want to merge several different recordings together, follow these steps:

## for each recording directory $SOURCE_i, run spikemonger($SOURCE_i, 'pre-merge')
## make a new root directory for merging, $MERGEDIR
## create the directory $MERGEDIR/sweeps
## either copy/move/symlink the sweep subdirectories from each $SOURCE_i/sweeps 
into $MERGEDIR/sweeps
## run spikemonger($MERGEDIR, 'post-merge')

* Cluster cutting:

-- cluster_ui
platform