Preprocessing

Both preprocessing and EA detection and classification work as a command line tool. The basic command structure is like this:

> python <path/to/codefile.py> <path/to/paramfile.yml>

Load and resample

The goal of this is to transfer obtain a resampled version of the raw data in hdf5 format. To date the loading supports .smr and .edf files.

Just run:

> python PEACOC/runthrough/rawToResampled.py run_params/my_recording_params.yml

My command line when resampling looks like this (the “(peacoc)” indicates that I work in the virtual enviroment “peacoc” ):

(peacoc) weltgeischt@heulsuse:~/PEACOC_tutorial> python PEACOC/runthrough/rawToResampled.py run_params/my_recording_params.yml

Depending on what you put in your specific parameter file, loading and resampling either just run through automatically, which is nice when processing several recordings in a loop, or you can select the channel interactively.

Automatic mode

For this you need to know the channel you want to access and enter it directly in the parameter file:

Alternative: Interactive mode

In case you do not know the channel name and want to select a channel interactively, channel should be set to interactive in the specific parameter file:

In interactive mode you then type the name of the channel you want to select and exit with d:

You now have created a resampled .hdf5 at /my_results/data/my_recording/ and a log-file of the analysis (which is nice for handing in, in case your analysis is not working properly):

Note

You can skip this loading and resampling routine and use your own resampled hdf5 file. Take care, however, to adhere to the format given in PEACOC_tutorial/my_results/data/my_recording/my_recording__raw500.h5

Assign polarity

Polarity refers to the direction of the spike component in EA. To interactively determine and set the polarity of your recording, run the following command:

> python PEACOC/runthrough/polarityCheck.py run_params/my_recording_params.yml

Two windows will pop up: A LFP trace of the whole recording…

… and an amplitude distribution, with checkboxes:

In this example you can see a shoulder at negative amplitudes, this strongly suggests, that the polarity of this example recording is negative. To be sure you could also zoom around in the LFP trace that just popped up. As you can see, the polarity indeed appears to be negative (the spike component goes down).

By marking a checkbox in the upper right corner of the window with the amplitude distribution, you select a polarity. Clicking Done (bottom right) ends the whole procedure, and a simple .txt file is created at PEACOC_tutorial/my_results/data/my_recording/my_recording__polarity.txt:

Note

If you know the polarity of your recordings anyway, you can yourself create a file my_recording__polarity.txt and do not need to follow the interactive routine to determine the polarity. If the file my_recording__polarity.txt is not present, later analyses will assume default polarity (negative).

Detect artifacts

To run the semi-automatic artifact detection, execute this:

> python PEACOC/runthrough/artifactCheck.py run_params/my_recording_params.yml

A window displaying the whole extent of the recording session will pop up. In it single events the algorithm detected as potential saturation artifacts are marked by red dots and potential longer stretches of artifacts are marked by purple lines:

Accepting suggested artifacts: Zoom around to have a look whether you want to accept any of the proposed artifacts. Accepting an artifact means that this stretch of data (plus a margin for the saturation artifact) will be masked for further analyses. To accept an artifact, double left click on its marker. Once you accepted the artifact its marker will turn yellow.

Rejecting suggested artifacts: double right click to reject a suggested artifact. Once rejected, the marker of the artifact will turn blue:

Warning

Only artifacts marked in yellow, i.e. accepted artifacts, will be saved as artifacts. The functionality of rejecting artifacts is just there to better keep track of which artifacts you accepted. By default, all potential artifacts the algorithm highlights will not be masked in later analyses unless you explicitly accept them.

Note

Make sure to release the zoom tool (by clicking on it), once you try to accept/reject artifacts. Otherwise selection will not work.

Adding artifacts yourself: To add a single artifact event double middle click at the position where you have identified it. A yellow dot will appear. For adding longer stretches of artifacts double left click at the position where you think the artifact starts and then single left click where you think the artifact ends. This artifact stretch will be indicated by a yellow line. We typically exclude large stretches of data in that way when they appear to be peppered by artifacts.

Important

For adding your own artifacts, always click below y=0. This serves to keep self-identified and automatically suggested artifacts separate.

Once you are finished with artifact hunting, click on the blue Done button. Again a simple .txt file is created at PEACOC_tutorial/my_results/data/my_recording/my_recording__artifacts.txt:

Below the header (artstart,artstop) the start and end points of large artifact stretches are indicated (yellow lines in the GUI, see above). Below the header saturation artifacts time points of single artifact events are given (yellow dots in the GUI). Don’t worry if some artifact stretches overlap in time (as shown in this example) - the tool can resolve this automatically.

Note

You can edit the file my_recording__artifacts.txt by hand or altogether avoid the interactive routine described here and create such a file yourself. If the file my_recording__artifacts.txt is not present, later analyses will assume that there are no artifacts.