This filtering is affecting stimulus detection on non-neural channels, on at least some rigs
Branch: https://github.com/JaneliaSciComp/spoke/tree/bugfix_FilterNonNeural

PSTH XData is in scan numbers and also the XLim is autoscaled, rather than being set to the same lims as the raster plots

Currently the channel grid dimensions are set to near-square via heuristics in ziniCreateGrids().

Idea is to make this either flexible in Spoke or auto-configured to match the type of probe (Whisper or Neuropixels) and its physical channel layout.

Clearest way to reproduce it: use RMS threshold with zero threshold, particularly unfiltered.

Sometimes the values don't appear to be zero, but this seems related to #6 .

Is the mean-subtraction on the display (as currently applied) the right thing to do for this case?

• Reductions in value sometimes gives "cell array index" error
• Switching from 1 to 2 sometimes has no effect

Note: this is an issue ported from prior repo

NOTE: This might already work. Should test/understand case where stimNumDisplayRange(2) is finite (as opposed to Inf).

User: "I would like to be able to fix the number of trials to be visualized in the raster window and that it updates online keeping  the total number of visualized trials the same (e.g. if I want to see 100 trials, when it gets to trial 101, I want to see from trial 2 to 101, then from 3 to 102 etc ). "

Note: this is an issue ported from prior repo

User: Would be better to plot the spikes in the raster plot as '.' rather than 'o'

Refresh rate 0.3:
To investigate

Refresh rate 3:

Can be reproduced with low threshold (e.g. 0) that leads to frequent spike detections in Raster mode. In such cases, processing can stall even with slow refreshRate (0.3 Hz) and RefreshPeriodMaxWaveformRate clamping to small (<10) spikes per refresh period.

The current spike rate limiter (refreshPeriodMaxSpikeRate) is not useful for raster mode because the spikes are limited from the start of the refresh period which occurs at arbitrary phase relative to the stimulus

One possible implementation might be a settable spike refractory period

Note: This is a 2016 issue ported from prior repo

User: "I would like to see the units and the values of the axes of all plots (waveform, psth and raster). Since all the plots have the same scale, it can be plot in one of them."

Note: this is an issue ported from prior repo

Baseline support for handling data from imec probes. No increase in data throughput, so processing pipeline should be applied to a channel subset (128 or less), i.e. equivalent to Whisper data processing rates.

SpikeGLX MATLAB interface functions used by Spoke:

• SpikeGL
• Close
• FetchNI
• GetParams
• GetSaveChansNI
• GetScanCountNI
• IsRunning

Note: this is an issue ported from a prior repo

User: "At the moment, in the raster plot, if one trial does not have spikes, is skipped.I would be better that an empty line is plotted instead. "

Here , there are two sets of accumulations within 3 seconds, even though the refresh rate is 3.3 seconds.

Currently channels are currently displayed in channel number order within a near-square grid determined by heuristics in ziniCreateGrids().

Some users want this displayed in alternative layout and/or ordering, e.g. to match the physical layout of pads on a probe.

Such flexible channel orders/layouts might be specified within Spoke itself; or it might "inherit" from the channel maps (probe location, display order) specified in SpikeGLX.

Note: old issue ported from prior repo

This is expected to be particularly important for stimulus-triggered waveform mode use cases. Less so for known raster mode use cases.

Note: This is a 2016 issue ported from prior repo

User: "When you plot the waveforms, it would be better if the last waveform is in a different color to distinguish it. That is first time it appears is in red then it turns black. Put as an option to select how many should be in different color."

Look for opportunities to reuse, rather than re-create, graphic elements in the display grids

Depending on the settings, the first trigger in raster mode can lead to a queued-up warning. This usually only happens the first time, and doesn't happen afterwards.

Early adopters are requesting a restoration of alpha version feature allowing color-encoding/sorting of different stimulus type events in Raster mode plots.
Branch: https://github.com/JaneliaSciComp/spoke/tree/feature_restoreStimEventClassification

Note: this is a 2016 issue ported from prior repo

User: "We would like to see waveform plot window at the same time of raster  plot window. At the moment it seems that you can visualize only one at the time. "

A secondary NI board can be used with Imec3b for the auxiliary channel inputs, e.g. the stimulus trigger inputs. This enhancement woudl support this expected common configuration.

The current Phase3b support is limited to standalone mode (i.e. without a secondary NI board) using the signle digital auxiliary bit (7'th bit) available on the Sync input that's typically used for syncing with the NI card.

Currently the spikeRefractoryPeriod defaults to the horizontalRange(2) value when switched over to raster mode.

Probably the right implementation is to add a constant specifying a minimal value for spikeRefractoryPeriod when in raster mode