si-0.101.2 - November 2024

Package versions

  • spikeinterface version: 0.101.2

  • aind-data-schema version: 1.0.0

Bug fixes

  • Fix handling of recordings with non-monotonically increasing timestamps, by resetting timestamps to start from 0.0.

  • Fix NWB LFP export in case of multiple channel groups: they are aggregated and saved as a single LFP electrical series in the NWB file.

  • Fixed a few bugs in dealing with multi-segment recordings (when concatenate option in job dispatch is enabled).

New features

  • Use SortingAnalyzer zarr backend as postprocessed output format. This removes the need of the postprocessed-sorting folder since the Sorting object is saved by the SortingAnalyzer by default.

  • Remapping of recording JSON files in preprocessed and postprocessed folders to the using the session name from the data_description file (if available). This enables to reload the preprocessed recording automatically, provided that the raw asset is attached.

  • Motion correction now uses dredge_fast as default and handles multi-segment recordings (which are concatenated before motion correction).

  • Added option to split_groups in job_dispatch to process different groups independently.