Virtual laboratory / Brain regions and the whole brain
High-density recording and spike sorting
Recording of many neurons with a silicon probe and sorting of their spikes, checked against the known true assignment.
Guided experiment: Recording many neurons at once with a silicon probe.
High-density recording and spike sorting
Silicon probes such as Neuropixels carry hundreds of recording sites along a thin shank and record the spikes of many neurons at once. The spike of each neuron appears on several neighbouring sites, and spike sorting assigns every detected spike to the neuron that fired it. In the first report of Neuropixels probes, two probes recorded more than 700 well-isolated neurons at the same time in an awake mouse (Jun et al., 2017).
Design the experiment
Questions to investigate
- Lower the detection threshold. How many of the extra events are noise?
- Switch to the sparse site spacing. Which neurons can no longer be told apart?
- Choose too few or too many clusters. What do merging and splitting errors look like?
- Use refractory-period violations to judge cluster quality without the true answer.
In a real laboratory
The probe is inserted into an anaesthetised or head-fixed animal, or implanted chronically. Recordings are carefully referenced and grounded, corrected for slow movement of the brain relative to the probe, sorted by automated software and checked by hand. Because the true source of each spike is unknown, sorting quality is judged from measures such as refractory-period violations and cluster isolation. All procedures require approval by an Institutional Animal Ethics Committee (in India, under the CCSEA and the Prevention of Cruelty to Animals Act 1960; elsewhere, the equivalent national body).
How the simulation works
Neurons at different depths fire as Poisson processes with a refractory period. Each neuron has its own spike waveform, whose amplitude falls with distance from each recording site. Two neurons always lie close together with similar spikes, so that only the spread of their spikes across sites separates them. Gaussian noise is added, and noise common to all sites can be removed by common-average referencing. Spikes are detected by a threshold on the most negative channel, described by principal components of the waveforms on nearby channels and grouped by k-means clustering. Probe drift and changes in spike shape during bursts are not simulated.
- Jun, J. J., Steinmetz, N. A., Siegle, J. H., et al. (2017). Fully integrated silicon probes for high-density recording of neural activity. Nature, 551(7679), 232–236. https://doi.org/10.1038/nature24636
About the simulations
Each simulator is a simplified version of a published model of cells, signals, instruments or behaviour, and runs in the web browser. The simulations show how the techniques work and how design choices change the results; they do not replace training in a laboratory.
All data are simulated and generated afresh on each run, so results vary between runs, as in real experiments. Nothing is stored or sent: designs and results are lost when the page is closed, unless the protocol is copied or the data file is downloaded.
Last updated