SpikeInterface 0.96.0 release notes

21st October 2022

Main change:
  • Change setup.py to pyproject.toml

  • Enable loading waveform extractor without recording and with pre-existing sorting

  • Implement NpzFolder for saving Sorting object

  • Implement in-memory waveform extensions

  • Improve NEO event handling

  • Modify pyks for IBL version

  • Implement sparse PCA

  • Extend comparison to multi-segment sortings

  • Various improvements to widgets modules

core:
  • Enable loading waveform extractor without recording and with pre-existing sorting (#929 - #990)

  • Make split_sorting dumpable (#934)

  • Fix recording propagation in run_sorter_by_property (#937)

  • Add ‘use_relative_path’ param to WaveformExtractor.select_units() (#956)

  • Check chunk_size VS number of samples in random chunks (#959)

  • Missing values for dtype object should be None (#960)

  • UnitsAggregationSorting property inheritance (#961)

  • Implement NpzFolder for saving Sorting object (#966)

  • Remove scipy dependency from core (#980)

  • Implement in-memory waveform extensions (#981 - #1003)

  • Simplify and improve check_json() function (#1014 - #1019)

extractors:
  • Add NWBRecordingExtractor in extractorlist (#982)

  • Fix deprecation warning for HDsortSortingExtractor (#991)

  • Improve NEO event handling (#995 - #1001)

  • Don’t load phy spike_templates if not necessary (#998)

  • Use proper annotation key instead of ‘k’ (#1017)

sorters:
  • Improve run sorter in docker and singularity containers (#938)

  • Modify pyks for IBL version (#1011)

  • Better handling of SI dev install in containers (#1013 - #1015)

  • Improvements to Circus2 pipeline (#971)

postprocessing:
  • Fix border effect for compute_spike_amplitudes (#962)

  • Implement sparse PCA (#975)

  • Protect against PCA projection on unfitted models (#983)

  • Protect solve monopolar triangulation from optimization errors (#1012)

qualitymetrics:
  • Catch errors in PC metrics (#941)

comparison:
  • Implement hybrid recordings (#944)

  • Extend comparison to multi-segment sortings (#963 - #984)

exporters:
  • Allow export_to_phy to make parent directories (#948)

widgets:
  • Sortingview backend: display in jupyter (#942)

  • Improve timeseries widgets (#945 - #1008)

  • Use segment_index=0 as default in plot_amplitudes (#964)

  • Hide diagonal in similarity plot (#968)

  • Fix labels and unit selection in widgets (#977)

  • Fix spikes_on_traces widget (#989)

  • Fix spike amplitudes and widgets tests (#996)

  • Port plot_unit_summary to the new widgets API (#999)

  • Fix order_channel_by_depth with multiple layers (#1000)

  • Fix show_legend flag in locations plots (#1005)

  • Check backend kwargs validity (#1002)

  • Sortingview metrics and unit tables fixes (#1019)

  • Add sortingview curation widget (#1023)

package:
  • Change setup.py to pyproject.toml (#986)

CI:
  • Fix ephy_testing_data cache (#993)

docs:
  • Update overview doc diagram (#940)

  • Fix doc index.html image (#1007)

  • Use correct test badge and add Twitter (remove unused Gitter) (#1016)