v1.7.3 (matlab) /v0.67 (python)

This release fixes and refactors python code:

• fix an indexing discrepancy in SNR calculation - thank you @m-beau!
• allow more flexibility in spike half width and baseline time calculations, esp. when using non-KS spike sorters
• uses the template peak channel also for raw waveforms
• uses the correct ACG function for ephys_properties
• some refactoring and code optimizations by @nikhilchandra and optimization of the ISI code by @FrancescoNegri - thank you both!

v1.7.3 (matlab) /v0.44 (python)

We are excited to announce the beta-release of bombcell in python! 🎉
The python version:

• Calculates up to 17 different quality metrics per unit to classify units into Good (somatic) , MUA (somatic), non-somatic and noise units
• Outputs summary plots so you can quickly see how well bombcell is working and what your data looks like
• Generates an interactive GUI for you to
  • quickly flip through cells and assess their quality metric scores
  • add manual labels
• Suggests changes to bombcell classification parameters based on your manual labels
• Calculate ephys properties for each unit, and optionally classifies striatal or cortical cell types based on these parameters

A huge thank you to @Sam-Dodgson and @m-beau for help developing this python version! Also thank you to new contributor @yunchang-z!

Additionally, this release introduces a fix to the MATLAB gui (see issue #206 )

v.1.7.2

• handles all zero waveform templates (as output-ed by Kilosort in some circumstances)
• detects extreme outliers in amplitude

v1.7.1

• Updates gettingStarted script
• Updates GUI for more robust visualization
• Fixes bug in non-somatic UpSet output, where some units were wrongly not classified as non-somatic

v1.7.0

A bunch of things because I haven't done a release in a while:

New features:

• support for post-hoc splitting/merging in phy - bombcell can now run on those new outputs. This introduced a breaking change - the loadEphys function now needs an additional input to work properly - see the gettingStarted.mlx script
• option to fit spatial decay with an exponential function - this is now the default, it performs much better
• better non-somatic metrics: we now have two ratios, the initial peak to repolarization peak ratio and the main peak to trough ratio
• introduces a new noise metric: the repolarization peak to trough ratio
• another (slightly better but slower) option to compute refractory period violations: Llobet et al. set param.hillOrLlobetMethod to 0 / false to enable.

Visualization update:

• updated GUI to be in two parts, with dynamic histograms that show where the current unit falls in the distribution. Helpful to adjust and fine-tune parameters

Fixes:

• only compute spatial decay of geometry of probe allows (ie if it is a high density probe)
• more robust waveform peak and trough detection

v1.6.0

• more robust spatial decay computation: template values are optionally (if param.normalizeSpDecay is set to 1) normalized relative to the peak value, making the spatial decay slope calculation more robust. - thanks @cbimbo for the suggestion. To update previously computed values, see the function bc_updateSpatialDecaySlope

• distance metric bug fix - thanks @chrisan97

v1.5.0

Handles kilosort4 spike sorting output.

v1.4.0

Added feature: option to fine-tune the non-somatic classification. Now, if a unit has a its largest detected peak before the trough, it is no longer automatically classified as non-somatic. The peak needs to be param.firstPeakRatio times larger than any value after the waveform's trough. Thanks @LaurenzMuessig .

Fixed a bug in classifying MUA unit. Thanks @JRegg66.

v1.3.0

New module to compute a host of ephys parameters and to classify striatal and cortical units.

v1.2.0

New option to separate non-axonal spikes into MUA and good. Thanks @cbimbo !

If param.splitGoodAndMua_NonSomatic is set to 1, non-somatic units are further separated into non-somatic multi-unit and non-somatic good unit.