RS FS Cell Type Classification on Visual Coding Dataset

Hi everyone!
I am a student currently interested in using the visual coding dataset to complete a class project on classifying regular spiking and fast spiking cells. Sorry if my question might not be the most well informed, please excuse my very very rudimentary knowledge in this area. I am using the brain observatory MATLAB toolbox to access the data. I had a few questions about how to go about this as I’m not as familiar with all of the computations behind some of the metrics. These are a few of my questions:

  1. How was the halfwidth for the units calculated (which part of the waveform was it calculated on)?
  2. I attempted to plot the halfwidth against the spontaneous firing rate I calculated from the spontaneous period before any visual stimuli were given and it seemed like I couldn’t really see two distinguishable clusters for FS and RS cells. I also tried combinations of amplitude vs halfwidth, duration vs halfwidth, amplitude vs duration, firing rate vs halfwidth, etc. as well as looked at the distributions but it seemed like none of these allowed me to see two clear clusters or a bimodal distribution. I was wondering if perhaps anyone might have some insight into why this may be and potential ways to circumvent this issue?

Thank you so much for the help and sorry if it’s not the most serious question … any help would be greatly appreciated!

Hello! All of the waveform metrics were calculated using the mean_waveforms module in the ecephys_spike_sorting package. The calculation for waveform halfwidth can be found here.

Waveform duration is generally more informative than halfwidth for distinguishing FS and RS cells. If you plot the histogram of waveform durations for units in cortex, you should see a bimodal distribution with a dip around 0.4 ms. Note that this only works if you’ve pre-selected cortical units, as other areas don’t display the same bimodality.