Demo spikeinterface notebook instructions not clear

[taningh86]

Hi,
I use spikeinterface to preprocess my raw spikeGLX data and use Kilosort3 to sort them. As you may already know, Spikeinterface creates a sorting analyzer folder where various extracted features, such as amplitudes, waveforms, correlograms, noise_levels, unit_locations etc are stored. However, I don't understand why creation of analyzer folder is repeated in the demo spikeinterface tutorial, when SI already makes it by default (or required by its 101.0 version). I am not sure what the ##Note, only applicable when you don't have the raw waveforms yet## is alluding to. What is raw waveforms here? Is it a waveform that the SI sorting analyzer creates or you are alluding to waveforms from kilosort? Kilosort don't create waveforms for each unit by default, it just gives out templates.npy, spike_templates.npy, ampltudes.npy etc, which one can later use to get waveforms. This is exactly what SI does and stores the waveform in the sorting analyzer folder. Therefore, I don't understand all the steps in your demo spikeinterface tutorial before actually running the unitmatch pipeline. They seem unnecessary to me and I think one should be able to just use the extracted features (waveforms, amplitudes etc.) in the SI sorting analyzer folder directly to start the Unitmatch pipeline.
Could you please let me know if this is the case or otherwise.

Thanks

[taningh86]

Sorry, I forgot to mention that the split recording for crossvalidation that Unitmatch requires is the only reason I can think why a new analyzer folder is created, since it requires two waveforms for each unit. If thats the case then I get it.
I just want to provide a suggestion to avoid readers/users of Unitmatch getting confused with the language. As far as I know spikeinterface or kilosort never creates two templates/waveforms for each unit. In the readme section this specific statement - "Each of these paths (typically a Kilosort output folder) should contain a subfolder called 'RawWaveforms' in which for every unit (/cluster) there is a NPY file containing two average waveforms per recording site (spikewidth X nRecordingSites X 2)" creates confusion and makes one think that kilosort makes two average waveforms by default, which is not the case. Modifying this line to something where it is clear that it is a necessity specifically required by UnitMatch and requires preprocessing of raw data would make things lot clear.

[EnnyvanBeest]

yes you are indeed answering your own question already. UnitMatch requires two waveforms per unit. Some people prefer to create this a different way than how we suggest it. So you can use our preprocessing as suggested, or extract these two average waveforms however you see fit before going to unitmatch.

I've updated the readme with the sentence "You are welcome to use our preprocessing tools to generate this from your raw data.", hope that clarifies things.

[taningh86]

Thank you! That will clear the confusion. Appreciate it!!