Two bugs in use

63 views
Skip to first unread message

Hao Jiang

unread,
Mar 12, 2022, 7:30:31 AM3/12/22
to SpyKING CIRCUS
Dear Pierre, 

I've got two problems in using the spike sorter.

1. when executing the code, sometimes I got the following warning:
  inotify_add_watch(/home/hjiang/.config/ibuss/60f8b434984145d1ba368b12ddce7c24-unix-18) failed: (No space left on device)
-------------------------  Informations  -------------------------
| MUA results not found!
------------------------------------------------------------------  

whether it is a crucial bug? 


2. The GUI was forced to shut down when I click the bottom on it, and then I got the error in figure 1&2.
the dataset I use is https://zenodo.org/record/1205233#.WrTFtXXwaV4,and the parameter files is shown as follow

 [data]
file_format    = mcs_raw_binary        # Can be raw_binary, openephys, hdf5, ... See >> spyking-circus help -i for more info
stream_mode    = None                  # None by default. Can be multi-files, or anything depending to the file format
mapping        = mea_256.prb              #!! AUTOMATICALLY EDITED: DO NOT MODIFY !!
sampling_rate  = 20000
data_dtype     = uint16
suffix         =                       # Suffix to add to generated files
global_tmp     = True                  # should be False if local /tmp/ has enough space (better for clusters)
overwrite      = True                  # Filter or remove artefacts on site (if write access is possible). Data are duplicated otherwise

[detection]
radius         = auto       # Radius [in um] (if auto, read from the prb file)
N_t            = 5          # Width of the templates [in ms]
spike_thresh   = 6          # Threshold for spike detection
peaks          = negative   # Can be negative (default), positive or both
matched-filter = False      # If True, we perform spike detection with matched filters
matched_thresh = 5          # Threshold for detection if matched filter is True
alignment      = True       # Realign the waveforms by oversampling

[filtering]
cut_off        = 100, auto  # Min and Max (auto=nyquist) cut off frequencies for the band pass butterworth filter [Hz]
filter         = True       # If True, then a low-pass filtering is performed
remove_median  = True              #!! AUTOMATICALLY EDITED: DO NOT MODIFY !!  

[triggers]
trig_file      =            # External stimuli to be considered as putative artefacts [in trig units] (see documentation) 
trig_windows   =            # The time windows of those external stimuli [in trig units]
trig_unit      = ms         # The unit in which times are expressed: can be either ms or timestep
clean_artefact = False      # If True, external artefacts induced by triggers will be suppressed from data
make_plots     =            # Generate sanity plots of the averaged artefacts [Nothing or None if no plots]

[whitening]
chunk_size     = 30         # Size of the data chunks [in s]
safety_time    = 1          # Temporal zone around which templates are isolated [in ms, or auto]
temporal       = False      # Perform temporal whitening
spatial        = True              #!! AUTOMATICALLY EDITED: DO NOT MODIFY !!
max_elts       = 10000      # Max number of events per electrode (should be compatible with nb_elts)
nb_elts        = 0.8        # Fraction of max_elts that should be obtained per electrode [0-1]
output_dim     = 5          # Can be in percent of variance explain, or num of dimensions for PCA on waveforms

[clustering]
extraction     = median-raw # Can be either median-raw (default), median-pca, mean-pca, mean-raw
safety_space   = True       # If True, we exclude spikes in the vicinity of a selected spikes
safety_time    = auto       # Temporal zone around which templates are isolated [in ms, or auto]
max_elts       = 10000      # Max number of events per electrode (should be compatible with nb_elts)
nb_elts        = 0.8        # Fraction of max_elts that should be obtained per electrode [0-1]
nclus_min      = 0.002      # Min number of elements in a cluster (given in percentage) [0-1]
max_clusters   = 10         # Maximal number of clusters for every electrodes
nb_repeats     = 3          # Number of passes used for the clustering
smart_search   = True       # Activate the smart search mode
smart_select   = False      # Experimental: activate the smart selection of centroids (max_clusters is ignored)
sim_same_elec  = 3          # Distance within clusters under which they are re-merged
cc_merge       = 0.975      # If CC between two templates is higher, they are merged
dispersion     = (5, 5)     # Min and Max dispersion allowed for amplitudes [in MAD]
noise_thr      = 0.8        # Minimal amplitudes are such than amp*min(templates) < noise_thr*threshold in [0-1]
remove_mixture = True       # At the end of the clustering, we remove mixtures of templates
make_plots     =            # Generate sanity plots of the clustering [Nothing or None if no plots]

[fitting]
chunk_size     = 1          # Size of chunks used during fitting [in second]
gpu_only       = False      # Use GPU for computation of b's AND fitting [not optimized yet]
amp_limits     = (0.3, 5)   # Amplitudes for the templates during spike detection [if not auto]
amp_auto       = True       # True if amplitudes are adjusted automatically for every templates
max_chunk      = inf        # Fit only up to max_chunk   
collect_all    = False      # If True, one garbage template per electrode is created, to store unfitted spikes

[merging]
cc_overlap     = 0.5        # Only templates with CC higher than cc_overlap may be merged
cc_bin         = 2          # Bin size for computing CC [in ms]
correct_lag    = True       # If spikes are aligned when merging. May be better for phy usage  


Thanks for your support !
Best Regards, 
H.Jiang
figure2.png
figure1.png
Reply all
Reply to author
Forward
0 new messages