Problem running spike sorters about spikeinterface HOT 9 CLOSED

Hi! Can you paste the code that you are running?

from spikeinterface.

Thanks for your response, sorry for not being clearer, this was the command that elicited the error:

In[25]: sorting_IC = ss.run_ironclust(recording_cache, output_folder='results_split_ironclust', grouping_property='group', parallel=True)

from spikeinterface.

Also the previous part please :)

What operating system are you using?

from spikeinterface.

Ahh. I was basically running through this tutorial to learn (https://github.com/SpikeInterface/spiketutorials/blob/master/NWB_Developer_Breakout_Session_May2020/SpikeInterface_Tutorial.ipynb), but I changed it to my data. I’m running this on a Windows machine running Windows 10 on an anaconda installation with Jupiter. The non plotting commands are pasted below:

import spikeinterface
import spikeinterface.extractors as se
import spikeinterface.toolkit as st
import spikeinterface.sorters as ss
import spikeinterface.comparison as sc
import spikeinterface.widgets as sw
import matplotlib.pyplot as plt
import numpy as np

rawdatafile = 'TY20210308_signalCheck_morning.ns6'
recording = se.BlackrockRecordingExtractor(rawdatafile)

recording_prb = recording.load_probe_file('TY_array.prb')
recording_f = st.preprocessing.bandpass_filter(recording_prb, freq_min=250, freq_max=7500)
recording_rm_noise = st.preprocessing.remove_bad_channels(recording_f, bad_channel_ids=[34, 50, 52, 67, 81])
recording_cmr = st.preprocessing.common_reference(recording_rm_noise, reference='median')
recording_cache = se.CacheRecordingExtractor(recording_cmr)
recording_cache.move_to('filtered.dat')
recording_cache.dump_to_pickle('recording.pkl')

ss.IronClustSorter.set_ironclust_path('F:/MLToolBoxes/ironclust')
Setting IRONCLUST_PATH environment variable for subprocess calls to: F:\MLToolBoxes\ironclust

ss.Kilosort2Sorter.set_kilosort2_path('F:/MLToolBoxes/Kilosort-2.0/')
ss.Kilosort3Sorter.set_kilosort3_path('F:/MLToolBoxes/Kilosort/')
ss.WaveClusSorter.set_waveclus_path('F:/MLToolBoxes/wave_clus/')
Setting KILOSORT2_PATH environment variable for subprocess calls to: F:\MLToolBoxes\Kilosort-2.0
Setting KILOSORT3_PATH environment variable for subprocess calls to: F:\MLToolBoxes\Kilosort
Setting WAVECLUS_PATH environment variable for subprocess calls to: F:\MLToolBoxes\wave_clus

sorting_IC = ss.run_ironclust(recording_cache, output_folder='results_split_ironclust', grouping_property='group', parallel=True)
print(f'Ironclust found {len(sorting_IC.get_unit_ids())} units')

(as a different issue, the only sorters that show up as installed after the above commands are:
In [24]: ss.installed_sorters()
Out[25]: ['ironclust', 'kilosort2', 'waveclus']

Im not sure why Kilosort3 doesn't install? There was no error output).

from spikeinterface.

If it matters at all, I tried duplicating this in Linux (Ubuntu 18.04) and I get the same error as above:

AttributeError: 'BlackrockRawIO' object has no attribute '__read_nsx_header_variant_a'

from spikeinterface.

Hi! What version of neo and SpikeInterface are you using? The problem seems to be related with the BlackRock input..@samuelgarcia any clue?

from spikeinterface.

I'm running neo 0.9.0 and spikeextractors 0.9.6
spikeinterface 0.12.0

On Windows, I installed it with pip, on Linux I installed it by downloading from GitHub and running python setup.py install

from spikeinterface.

This error is strange. Is is seems related to neo and joblib.
Does the same code work with no parralel ?

from spikeinterface.

Hello! On the linux install, I tried restarting the jupyter kernel and rerunning the code without the parallel variable in the argument list. This indeed worked. I verified this by restarting the jupyter kernel and running the code again with the parallel variable set to True as an argument. The same error as previously described occurred. Thanks for your help.

from spikeinterface.