Kilosort
Kilosort copied to clipboard
Double Counting shifted spikes
It seems somehow, the spikes are shifted in time and double counted. You can tell by the overlay of the waveform and the peak at 0 in CCG. Not sure hoe did this happen. As only a subset of the spikes are double counted and the blue one has lower amplitude, my guess is that after subtracting the spike from the trace, a little bump is detected around the original spike.
You are describing the same issue previously discussed here: https://github.com/MouseLand/Kilosort/issues/29
There is no perfect solution, but that thread provides some workarounds.