I have been using the channels.csv file for Neuropixel data, and was trying to compute the minimum seperation between 2 recording sites. Based on the documentation on IMEC site, the sites were spaced 20um by 16um away, and each site was 12x12um. So I was expecting the smaller distance between 2 sites to be >15um, somewhere around 25um. But going through the channels file, and using the ccf coordinate based numbers, I am getting the distances around 8.5um.
Were the recording site location not centered on the Neuropixels? When converting to ccf, at what resolution was the center of the recording site assigned?
Thanks in advance
Here is a diagram of the Neuropixels electrode locations. The inter-site spacing you cited is correct, and for this geometry we should indeed expect the minimum center-to-center distance to be about 25 microns.
However, in order to map the recording sites into the CCF, we first deform the probe using a warping function obtained by manually identifying structural landmarks between each individual brain and the CCF template brain. Even though the probe tracks are straight in the original brains, the tracks become curved after warping to the CCF. This warping can compress the distance between adjacent sites, which explains why the minimum spacing between electrodes can differ from what’s expected from the original probe geometry.