SpikeInterface 0.93.0 release notes

4th February 2022

  • add WaveformExtractorExtension (PC, qualitymetrics, spike amplitudes) to automatically store and retrieve processed data waveforms folder

  • add singularity integration in run_sorter

  • add a link to the originating recording to the sorting object

  • new framework for collision benchmark

  • refactor comparison module and add TemplateComparison

  • add template_matching module (experimental)

  • add motion_correction module (experimental)

core:
  • fix timestamps when from h5py

  • allow concatenation of datasets with diff srate

  • propagate properties in Recording and Sorting OldToNew wrappers

  • add WaveformExtractorExtension.

  • fix old to new scaling

  • implement filter_units() in waveform extractor and extensions

  • make waveform cache to memory optional

  • refactor old<->new API

  • sorting have a link to the recording and so

  • sorting.get_unit_spike_train have return_times=True/False

  • add ‘missing_value’ to set_property

  • improve portability of WaveformExtractor

  • add SplitSegmentSorting

extractors:
  • fix KS keep_good_only query

  • remove __del__ from nwbextractor

  • various fix

sorters:
  • singularity integration

  • enhance container handling

  • matlab-compiled IronClust version

toolkit:
  • whiten by property

  • fix unit_localization

  • fix slice in get_chunk_with_margin

  • fix ‘single’ common referencing

  • enable n_jobs >= 2 compute_spike_amplitudes

  • fix gain/offset shape

  • fix peak sign for amplitude cutoffs

  • refactor template metrics

comparison:
  • new framework for collision benchmark

  • refactor comparison module and add TemplateComparison

widgets:
  • widgets for collision benchmark

export:
  • fix export_to_phy

  • add quality metrics and SI unit ids to phy tsv

sortingcomponents:
  • add select_peaks()

  • add template_matching module (experimental)

  • add motion_correction module (experimental)

doc:
  • various improvements

  • fix some typos