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

mixup in units? #87

Closed mariapfj closed 1 month ago

mariapfj commented 1 month ago

Ref issue posted recently with removing excess spikes; was able to spikesort 137-310121-7 from mec-social but I'm encountering some really odd issues during curation.

One unit (really nice grid cell so can't simply ignore it) seems to have been merged across tetrodes, only to be revealed as belonging to separate tetrodes when clustercutting. Images shows units 135 and 136, which I split from a single unit initially. However out of nowhere the amplitude and main channel changes (from 28 to 18 for one unit, not the other), as well as waveforms (clearly showing two different tetrodes). channel_group remains the same for both (contrary to changes in ch), but if I keep curating, or undo changes, channel_group disappears (this has previously also been an issue when saving to nwb, can't remember if we implemented a solution already).

I have no idea whats going on here

image image

this is what is seen in the notebook before I start clustercutting, before it splits into two different units, which seems to correpsond more to the "noisy" unit in phy (ref a missing channel), but also the waveform corresponds a lot better to what the unit initially looks like before I start clustercutting

image image

mariapfj commented 1 month ago

can also add that they have a similarity score of 1 :)

mariapfj commented 1 month ago

for reference this is what it looks like initially, before removing anything

image

alejoe91 commented 1 month ago

So that seems to be a problem with Phy?

mariapfj commented 1 month ago

phy or kilosort i guess, but i wanted to run it by you in case it could be related to any changes made recently ?

fre. 20. sep. 2024 kl. 16:55 skrev Alessio Buccino @.***

:

So that seems to be a problem with Phy?

— Reply to this email directly, view it on GitHub https://github.com/CINPLA/expipe-plugin-cinpla/issues/87#issuecomment-2363924552, or unsubscribe https://github.com/notifications/unsubscribe-auth/ANYAOKNQIAEPOLDAYKY2Q4LZXQZNBAVCNFSM6AAAAABOSGAXEGVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDGNRTHEZDINJVGI . You are receiving this because you authored the thread.Message ID: @.***>

alejoe91 commented 1 month ago

I see! It could be that those spikes that you're cutting out because they look lilke outliers on in the feature view are indeed noisy spikes that also appear on other channels. This means that by isolating them into a separate cluster, the main signal could be associated to a different tetrode. Does it make sense?

mariapfj commented 1 month ago

could be yeah, I’ll have another look on monday to see if I can sort it out

fre. 20. sep. 2024 kl. 17:08 skrev Alessio Buccino @.***

:

I see! It could be that those spikes that you're cutting out because they look lilke outliers on in the feature view are indeed noisy spikes that also appear on other channels. This means that by isolating them into a separate cluster, the main signal could be associated to a different tetrode. Does it make sense?

— Reply to this email directly, view it on GitHub https://github.com/CINPLA/expipe-plugin-cinpla/issues/87#issuecomment-2363953050, or unsubscribe https://github.com/notifications/unsubscribe-auth/ANYAOKPWQK4JC5C4H4UP3YDZXQ3ABAVCNFSM6AAAAABOSGAXEGVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDGNRTHE2TGMBVGA . You are receiving this because you authored the thread.Message ID: @.***>