BuG: Re: spike holes

[WeissShahaf]

Describe the issue:

Just making sure the issue isn't lost in kilosort4.
Issue #594 raised a serious concern regarding the data loading.
Can you confirm if this issue is present in KS4 or being investigated?

Thanks,
Shahaf

Reproduce the bug:

No response

Error message:

No response

Version information:

Linux win 10

Context for the issue:

Possible bug carried over from past versions of kilosort

Experiment information:

No response

[marius10p]

I have replied in #594. The issue does not seem to be present in Kilosort4 probably because the new binary file functions work very differently, but if anyone finds it, feel free to reopen that issue.

It also appears to affect different versions differently, i.e. it seems very strong in pykilosort, but less so in the Matlab versions. I could replicate it in the Matlab versions of KS 2.5 and 3, but not in 1 and 2.

[runbogao]

Hi Marius, thanks for testing the issue for all the version of KS. We are using a pipeline (https://github.com/jenniferColonell/ecephys_spike_sorting)) based on KS2.5 to do spike sorting, and all pipeline for downstream analysis are based on that. We would like to stick to KS2.5 for now. Do you have any suggestions to work around the bug (e.g. splitting batches with a small overlap)? Thank you in advance!

[jacobpennington]

Closing this one, it looks like #594 is the right place to continue this discussion.

[marius10p]

@runbogao The latest release version 2.5.2 should fix this issue. Would be great if you could test it and give me some feedback.