0.95.0.rst

SpikeInterface 0.95.0 release notes

2nd September 2022

Main changes:

• Add BaseSnippets object to handle waveform cutouts (#764)
• Sacrificed toolkit in favor of: preprocessing, postprocessing, qualitymetrics, and curation modules (#745)
• Major refactoring widget module to allow for multiple backends (#686 / #786 / ) * matplotlib * ipywidgets * sortingview
• First implementations of (experimental) sortingcomponents-based sorters: * spykingcircus2 (#787) * tridesclous2 (#800)

core:

• Add ZARR storage_options argument to enable cloud read/write (#622)
• Add seed for sampling waveforms (#629)
• Allow for unconnected contacts when setting probe (#646)
• Refactor nested recursive functions (#651 / #774)
• Switch to packaging (#697)
• Add BinaryFolderRecording and is_binary_compatible() (#726)
• Fixed SharedMemory buffer (#741)
• Improve handling of unitless NEO extractors (#758)
• Add mp_context arg to control multiprocessing (#781)
• Option to automatically cast unsigned dtypes in get_traces (#798)
• Add mode="memory" for WaveformExtractor (#832)
• Patch memory leak when saving to Zarr (#915)
• Add return_scaled in write_to_h5 format (#908)
• Improved NEO wrapper to handle multi-blocks and return number of blocks and stream info (#909)
• Update OpenEphys and SpikeGLX for NEO and ProbeInterface releases (#916)

extractors:

• Fix 1D kilosort/phy spiketrains (#617)
• Propagate NEO annotations to SI objects (#627)
• Add AlphaOmega support (#637)
• Improve BIDS reader (#647)
• Add ros3` support for NWB (#658)
• Fix neo kwargs propagation (#723)
• Load ADC sample shifts for Open Ephys and SpikeGLX (Neuropixels) (#734 / #759)
• Load gain for MEArec (#791)
• Added functions to get extractors from name (#835)
• Update nwb recording extractor to use offsets (#846)
• Add EDFRecordingExtractor, NeuralynxSortingExtractor, BlackrockSortngExtractor (#853 / #854 / #856)
• Fix IBL cbin kwargs (#873)
• Add snippets extractors: MClust (#879) - WaveClus (#764)
• Fix Phy reading when unit groups are excluded (#876)

sorters:

• Allow to propagate local SI installation to docker/singularity container (#630)
• Refactor MATLAB-based sorters to allow compiled versions (#631 / #660 / #649 / #684)
• Fix docker mechanisms for Windows (#635)
• Remove unused mdaio from IronClust (#638)
• Add wave_length param to kilosort (#655)
• Better handling of GPU requirements (#674)
• Add default container map (#685)
• Add extra_requirements mechanism to install required packages in container (#689)
• Unify handling of job_kwargs across sorters (#706)
• Optional motion correction for KS2.5 and KS3 (#709)
• Dump npz sorting in container (#779)
• Add tmpdir parameter to MS4 (#827)
• Add SLURM launcher (#828)
• Add WaveClus from snippets implementation (#764)
• Extend docker map to HerdingSpikes and Combinato (#891 / #889)
• Fixes to pykilosort (#890)

preprocessing:

• Allow artifact removal of single sample when ms_before=ms_after=None (#618)
• Add operator to CMR kwargs (https://github.com/SpikeInterface/spikeinterface/pull/634/files)
• Add DeepInterpolationRecording (#661)
• Add zscore preprocessing (#679)
• Enable specification of dtype for whitening (#881)

postprocessing:

• Fix template metrics return values (#620)
• Add sparsity option for template metrics (#629)
• Added SpikeLocationsCalculator as WaveformExtension (#735)
• Improve efficiency and verbosity of PCA (#739)
• Add ISI histograms (#760)
• Refactor ISI histograms, CrossCorrelograms, TemplateSimilarity, and UnitLocations extensions (#778)
• Fix fancy indexing for arrays (#816)

qualitymetrics:

• Improved tests (#692)
• Add sparsity to quality metrics computation (#739)

curation:

• Add CurationSortingExtractor class (#612)
• Add deduplication of spike trains (#776 / #817)

comparison:

• Fix multicomparison duplicate removal (#620)
• Add option to remove sorters folder in GTStudy (#629)
• Better internal handling of comparison objects in multicomparison (#732)

widgets:

• Improved similarity widget (#665)
• Allow selection of waveforms to plot (https://github.com/SpikeInterface/spikeinterface/pull/672/files)
• Add Tiled plot (#693 / #821)
• Refactor multiple widgets (#762 / #820)
• Multi-layer timeseries (#783)
• Add sorting_summary widget for sortingview (#795)
• Fix generation of docstrings for plot_ functions (#920)
• Unify generation of 2d axes in MPL (#921)

sortingcomponents:

• Component-specific benchmarks (#667 / #678 / #839)
• Restructure matching module (#668)
• Refactor motion correction (#669)
• Add numba isocut implementation (#688)
• Add nearest neighbor NN triaging (https://github.com/SpikeInterface/spikeinterface/pull/694/files)
• Add compute features from peaks (#754)
• Implement "peak pipeline" (#845)

docs:

• Documentation updates for quality_metrics (#666)
• Automatic generation of function from classes (#673)
• Add intersphinx for neo readers (#677)
• Improve API documentation (#691)
• Improve documentation of containerized sorters (#765 / #777)
• Expose classes in the API docs (#863)

CI:

• Add NWB ros3 test
• Add code coverage (#675)
• Create automatic issue when a MATLAB file is changes -- might require docker image update (#756)