I've sorted some data in KiloSort, but noticed that I have a bad channel that KiloSort was finding a lot of clusters on, sometimes merging those clusters with real spikes that I'd like to pull out. To get around this issue, I got KiloSort to not include one of the channels using the connected variable in the channel map. In the end, KiloSort runs without any issues. However, when I try to manually correct clusters in Phy, it looks like something has gone wrong:
Namely, if I look at the waveforms of each cluster, there are no spikes, just flat noise (pic above)
However, I know some channels have spikes on them, so if I select a channel with spikes and look at the templates fit by KiloSort, it looks like they fit some spike waveforms!
BUT, if I look at the waveforms fit to these templates, there's nothing in the raw data:
It seems as if Phy is not using the right spike times or that the channel map has potentially been mixed up when excluding a channel (ie. the Phy channel numbers go from 0 to 14, but maybe they should go from 0 to 15, excluding the bad channel?).
In params.py there are 16 channels, but only 15 are displayed? Maybe this mismatch is the issue?
I've sorted some data in KiloSort, but noticed that I have a bad channel that KiloSort was finding a lot of clusters on, sometimes merging those clusters with real spikes that I'd like to pull out. To get around this issue, I got KiloSort to not include one of the channels using the
connected
variable in the channel map. In the end, KiloSort runs without any issues. However, when I try to manually correct clusters in Phy, it looks like something has gone wrong:Namely, if I look at the waveforms of each cluster, there are no spikes, just flat noise (pic above)
However, I know some channels have spikes on them, so if I select a channel with spikes and look at the templates fit by KiloSort, it looks like they fit some spike waveforms!
BUT, if I look at the waveforms fit to these templates, there's nothing in the raw data:
It seems as if Phy is not using the right spike times or that the channel map has potentially been mixed up when excluding a channel (ie. the Phy channel numbers go from 0 to 14, but maybe they should go from 0 to 15, excluding the bad channel?).
In params.py there are 16 channels, but only 15 are displayed? Maybe this mismatch is the issue?
Any suggestions?