Parameters for Cambridge Neurotech Acute 32-channel probe on kilosort4

[AmaliaLambertRinguet]

Hi there!
I'm soliciting your help to adjust the default settings on kilosort4, for a Cambridge Neurotech Acute 32-channel probe : the ASSY-37 Fb (3 shanks).

After running the spike sorting algorithm, we obtain all the .npy and the .tsv files, the diagnostics, drift-amount and spike-positions but the default parameters don't seem to suit. We changed the "nblocks" = 0, as recommanded for probes with less than 64 channels.

[jacobpennington]

What about the results didn't look right? Also, please upload kilosort4.log from the results directory.

[AmaliaLambertRinguet]

Look at the duration of the spikes (normally ranges between 0.5 and 2 ms, and actually where we record between 0.5 and 1.5 ms), seen in the waveforms in the "Temporal Features" (with time in seconds!) :

In my opinion, these are likely due to artifacts or issues with spike detection and sorting. I'm trying to adjust parameters such as "whitening_range", "highpass_cutoff", "nt0min" or "duplicate spike ms", that's why I would very much like to have the suggestions of someone who's worked with this kind of probe or faced similar issues.

Here is the kilosort4.log file :
kilosort4.log

Thanks for your help!

[jacobpennington]

That graph is not showing spikes, it's showing PC features extracted from single-channels spikes, and that looks like a normal set. I would suggest looking at the results in Phy to get a better idea of what the KS4 results look like, using the default parameters.

[AmaliaLambertRinguet]

That graph is not showing spikes, it's showing PC features extracted from single-channels spikes, and that looks like a normal set. I would suggest looking at the results in Phy to get a better idea of what the KS4 results look like, using the default parameters.

Oh okay, thank you!
Actually, I posted another issue about Phy (cortex-lab/phy#1319). I probably have a problem with Python that keeps me from visualising the results...