Cell type classification error

[DaniMagyar]

Hi Peter,

I hope you're doing well.

I record units from the BLA using the same Buzsaki64 probe each day. In some of my recordings, all detected neurons are classified as narrow-spikers, but on the following day, the classification appears to be correct again.

I'm trying to identify the cause of this issue but haven't been able to resolve it yet. I was wondering if you might have any insights into what could be causing this.

I would really appreciate any thoughts you may have.

Best regards,
Daniel

[petersenpeter]

Hi Daniel

Could you share a screenshot from a session where your units are classified correctly? The classification is hard-coded and depends on the actual width of the waveforms. In your screenshot, all waveforms are narrow. Are you changing your hardware filter between sessions? Which data acquisition system are you using?

/ Peter

[DaniMagyar]

Thank you for your response. I attached all my recordings. On the upper left plot I plotted the animalID against the trough-to-peak. As you can see there are 5 recordings where all neurons were classified as narrow spikers. I used the same probe during the entire experiments in all animals and did not change anything in the code. I use it with default settings. I use a CambridngeNeurotech P1 probe with 2x Intan 32 channel headstages and and intan RDH USB interface board. I record the data in OpenEphys, sort it with Kilosort 2.5 and manually curate in Phy.

[DaniMagyar]

I checked my raw data files and it seems like OpenEphys saved already filtered data in these sessions. However I don't understand how, because I use the "Record Node" immediately after the "Acquisition Board" in all my experiments to save the raw data without any modification. These recordings look like they were saved after bandpass filtering since the LFP is completely missing. Even my settings.xml files show that the Recording Node followed the Acquisition Board and bandpass filtering happened later in the signal chain. I don't understand what happened, but thanks for the hardware filter idea, now I have an explanation why I see only narrow waveforms.

[petersenpeter]

Sorry to hear that, but it’s good that you identified the issue. In some versions, Open Ephys saves the LFP data as a separate file; perhaps you’re lucky and the LFP data still exists.