search

CINPLA / expipe-plugin-cinpla

Plugins for expipe command line interface
http://expipe-plugin-cinpla.readthedocs.io/en/latest/
GNU General Public License v3.0
1 stars 3 forks source link

if one channel is grounded the relative channel group gets only one messy cluster #7

Closed alejoe91 closed 5 years ago

alejoe91 commented 7 years ago

We could duplicate one of the good channels in the channel group on the bad one

espenhgn commented 6 years ago

Wouldn't this screw up with klusta? Would it not be better to provide klusta with a modified probe file where the grounded channel is removed from the probe-file 'channel' entry? This works well when manually running klusta it seems.

Imo bad channels should not be included in further steps at all (e.g., it shouldn't be possible to use the LFP in that channel in analysis for example)

alejoe91 commented 6 years ago

I think the clustering would not be screwed up, as we are removing noisy information and just duplicating existing information. Sometimes EAP is a bad channel, but when filtered in the LFP band it might still be useful. Anyway we could discuss about modifying the probe file on the fly. It can be a good solution :)

espenhgn commented 6 years ago

Ok, maybe some middle ground would be fine. I suppose "bad" channels can be labeled as such in exdir/Neo when --ground is used in the process action, and that subsequent analysis using e.g., exana would raise some userWarning if that data is used.

alejoe91 commented 5 years ago

Now this is nicely fixed by removing the bad channels in spike_extractors. Needs to be updated for the command line interface though

alejoe91 commented 5 years ago

Fixed with SpikeInterface: bad channels are removed