jacobpennington
commented
3 days ago Hello,
The y-axis on that plot is the estimated depth of each spike relative to the coordinates specified in your probe file, it is not related to clustering. I.e. if the values of probe['yc'] range from 0 to 1000, then that will be the range of possible values for that plot, and the y-coordinate of a point on the plot is the estimated center of a single detected spike prior to drift correction. The same is true for the spike position plot that gets generated at the end of sorting, except that those values will be on the x-axis for that plot and the spike positions are computed on drift-corrected data for that plot.
Kilosort4 does not have any anatomical information, so these values are always relative to the coordinates given for the probe itself.
davrubino
Hi everyone,
I need clarification regarding one of the output figures from Kilosort4, specifically the "drift_scatter.png" plot. The y-axis represents the computed depth for each cluster, but I’m unsure how to interpret this dimension. Does it indicate the anatomical depth of each cluster, or is it simply the coordinates of the recording sites associated with each cluster?
This is particularly important because I’m using Neuropixels 1.0, where the channel map is reversed compared to the conventional orientation. In this setup, the y-coordinates are referenced from the tip of the probe, meaning sites deeper in the anatomical sense have a lower y-position.