th-1 FAQ

1) Are the spike times in filename.res.N expressed as multiples of the 20kHz sampling rate, so that time in seconds = (spike time in file) / (20e3)?

Yes

2) Do the cluster indices in the various cluster files for a given session refer to the same clusters? So for example would cluster 2 in Mouse12-120806.clu.1 refer to the same cell as cluster 2 in Mouse12-120806.clu.2?

No. Mouse12-120806.clu.1 refers to the identity of cells recorded on the first shank of the silicon probe, for the entire session. Mouse12-120806.clu.2 refers to shank #2. The number of lines in the the clu files is the same as in the res files plus on (the first value is the total number of clusters).

3) In the states files (e.g., Mouse12-120806.states.REM, Mouse12-120806.states.SWS, Mouse12-120806.Wake) are the numbers also times expressed in terms of the 20 kHz sampling rate?

No, sorry, it's a tiny difference with the hc-3 datasets. It's in seconds!

4) Finally, I gather the .whl file contains the LED coordinates in the format (x1, y1, x2, y2): 

a. Is this still at a sample rate of 39.06 Hz, so that the nth sample is at time n/39.06?

b. Is it synced with the spike times in any other way? The Methods mention that there's a random misalignment to start with. Would this affect matching

spikes to head position?

c. Do you calculate the head direction as the angle of the vector pointing from the 2nd to the 1st LED?

To address this question, the following two archive files are added to the data set:

PosFiles.tar

AngFiles.tar

The "pos" files are smoothed (i.e. interpolated) positions of the animals in centimeters (the way we used the positions for analysis), still sampled at 39.062 Hz (which is not just a magic number btw: it's EEG sampling rate – 1250 Hz - divided by 32 so that the sampling of the position is an integer number of LFP samples). The absolute time is the same as for spikes, no need to realign anything. HOWEVER, as you said, there is a tiny jitter (up to, maybe sometimes, 100 msec) between spikes and video. This jitter is constant across an entire session (it was a problem with the trigger starting data recording that we discovered once all the data were recorded...). This jitter seems to be always positive (the spike times always lag behind the video). So there is one analysis that can't be done: it's the anticipatory properties of HD cells in the thalamus that was reported by Blair and others.

The second tar file (AngFiles.tar) contains "ang" files: these are the orientation of the head in radians (between 0 and 2 pi), in the same reference frame as the (X,Y) positions given by pos files (i.e. at 0 rad, the animal points right).

In both files (pos and ang), -1 values mean that the detection failed.