Skip to content

Recommended pre processing

Julie Fabre edited this page Sep 24, 2023 · 4 revisions

To maximize the quality of your data, we suggest:

  • minimizing the noise and artefacts on your data.

    • To minimize noise and artefacts while recording, we recommend either:
      • recording with either both a ground wire and reference wire going from the probe to the mouse,
      • or with a ground wire and the internal reference (if using probes other than the 3A - the internal reference on these doesn't work well).
    • To remove artefacts from your recorded data, either:
      • temporally align your channels with each other, before feeding this data into kilosort (which then common-average references the data, removing artefacts). Indeed, A to DC conversation on the probe base is done in batches of channels - ie some channel batches will end being acquired slightly before/after others. Because of this, it is best to first align all channels before performing common average referencing. This command line tool will temporally align the channels.
      • or use pyKilosort, where this channel aligning and common-average referencing are already built-in.
  • fine-tuning kilosort's spike detection parameters. We recommend looking at your raw data and spikes detected by kilosort, to assess whether most spikes are being detected. If not, consider lowering kilosort's detection thresholds.