Glia

A package to support neuroscientists in analyzing multi-electrode array recordings. Designed with the retina in mind.

Some fun visualizations:

retina2pixel saliency map of the impact of each pixel on a given MEA channel.

pixel2retina "forward" saliency map of the impact of each MEA channel on a pixel selected by the mouse.

Installation

> pip install git+https://github.com/tbenst/glia.git

We install in development mode so can update with a simple git pull

Documentation

Access the glia command line tool documentation with "glia -h." All sub commands also have documentation, eg "glia analyze -h".

Typical command order:

# create .stim file & make plots for "integrity" check
> glia analyze R1_E1_AMES_120min_celltyping integrity
# create .frames file
> glia process R1_E1_AMES_120min_celltyping
# plot receptive fields
> glia analyze R1_E1_AMES_120min_celltyping sta
# create .npz/.h5 file for machine learning
> glia analyze R1_E1_AMES_120min_celltyping convert
# run classification & plot results
> glia classify R1_E1_AMES_120min_celltyping

For Docker 1.17 (or later):

sudo docker run --network=eyecandy_default -v $(pwd):/data tbenst/glia analyze -e "http://eyecandy:3000" /data/R2_E1 convert

How to analyze data:

1. Convert *.mcd files into *.voltages

2. Pull files from MEA computer onto local machine

3. Open docker and go to folder with data e.g. docker run -v /c/Users/sandt/Desktop/160913:/data tbenst/mcd -> this will automatically start the conversion process in the folder Wait until all files are converted, i.e. the terminal says: process finished!

4. Find out header length

   Open new terminal: chdir /Documents/Github/ run: docker run --rm -v /c/Users/Administrator/OneDrive/jupyter-notebooks:/notebooks --link eyecandy_web_1:eyecandy -p 8888:8888 tbenst/jupyter-neuro go to Chrome and type: localhost:8888 go to “get header offset” type in folder with *.voltages file run script (last line will spit out header length)

5. spike sorting

6. Load data for spike sorting Open Plexon: File -> import data -> import binary file Open file location 60 channels Sampling frequency (usually 25000, information can also be found in header) Header length (see point 3) Press ok

7. Filter data and detect spikes Open “waveforms” Filter continuous data Butterworth 4th order, 330 Hz For all channels Detect spikes: Open “waveforms” -> detect spikes Threshold -3.8 -> for all channels

8. Spike sorting Open “sort” Perform automatic sorting E-M: 15

when finished: visually check the sorted data and invalidate noise spikes save waveformes as: *.txt file; all units in one file. delimiter ,

select: channel (raw), unit, timestamp

Dev notes

sudo docker run -it -v $(pwd):/data tbenst/glia:acuity analyze -v -p 4 -e http://localhost:3000 /data/R1_E1_AMES_50min_acuity.txt integrity solid --wedge bar --by acuity acuity