tetrode data from open ephys

[Mikkel Elle Lepperød]

Hi,

Working with data from Open-Ephys, I have some issues with spike sorting with Klusta. It's a setup with 8 tetrodes, however for testing purposes attached is a setup with 1 tetrode.

By just filtering the data and plotting with matplotlib I can find a nice spike in the dataset (at 29.814 s), so i'm pretty sure the data is sane, just misunderstanding something with Phy where the raw trace looks very flat and the waveforms are jiberish.

Could you please check if you can spikesort the attached data?

https://drive.google.com/open?id=0B40lnPhKINCJeFhLdVJZdGQwajg

Best regards

Mikkel

[Cyrille Rossant]

The raw data isn't looking right to me, could you send me the code you're using to plot the raw data?

Cyrille

--
You received this message because you are subscribed to the Google Groups "phy-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to phy-users+unsubscribe@googlegroups.com.
To post to this group, send email to phy-...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/phy-users/1603b58a-15d6-4f3a-b743-bc066c874f52%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Mikkel Elle Lepperød]

Hi Cyrille,

Thank you for looking into this. I'll share a dropbox folder with you with the data, and a notebook with the code for viewing the filtered data. Just tell me if you want the full recording with 32 channels, i'll just add it to this folder.

https://www.dropbox.com/sh/z6wq0eang2ri4rq/AAAPwnGMoKJDZlY566oqjU7ma?dl=0

Which part of the data does not seem right to you? The low frequency part?

Best

Mikkel

On Thursday, December 29, 2016 at 5:26:17 PM UTC+1, Cyrille Rossant wrote:

The raw data isn't looking right to me, could you send me the code you're using to plot the raw data?

Cyrille

2016-12-20 14:32 GMT+01:00 Mikkel Elle Lepperød <lep...@gmail.com>:

Hi,

Working with data from Open-Ephys, I have some issues with spike sorting with Klusta. It's a setup with 8 tetrodes, however for testing purposes attached is a setup with 1 tetrode.

By just filtering the data and plotting with matplotlib I can find a nice spike in the dataset (at 29.814 s), so i'm pretty sure the data is sane, just misunderstanding something with Phy where the raw trace looks very flat and the waveforms are jiberish.

Could you please check if you can spikesort the attached data?

https://drive.google.com/open?id=0B40lnPhKINCJeFhLdVJZdGQwajg

Best regards

Mikkel

--
You received this message because you are subscribed to the Google Groups "phy-users" group.

To unsubscribe from this group and stop receiving emails from it, send an email to phy-users+...@googlegroups.com.

[Cyrille Rossant]

To be clear, is the problem in the spike sorting process with klusta which does not find spikes, or in phy where the display is problematic?

To unsubscribe from this group and stop receiving emails from it, send an email to phy-users+unsubscribe@googlegroups.com.

To post to this group, send email to phy-...@googlegroups.com.

To view this discussion on the web, visit https://groups.google.com/d/msgid/phy-users/3d6f206c-472c-4b91-a7a3-14df97db2dfd%40googlegroups.com.

[Mikkel Elle Lepperød]

The problem is in Klusta - I think, which finds to many "false positives", in that recording one can find a small number of nice waveforms, say in the order of 10, but Klusta finds cells in the order of 1000. I'm to little familiar with the detection and spikesorting process in Klusta to understand what the problem is. In addition, we are testing a new setup with open ephys, so there might be some issues there as well. However with a different system (Axona) we can verify that the tetrodes picks up nice multi unit activity.