pfc-2 LFP data: sampling rate and recording length

Hi,

Firstly, thank you for uploading the pfc-2 dataset! I have a couple of questions concerning the data organization of LFPs in the dataset:

- it's stated in the documentation that rat EE contains 104 channels, while rat FF and GG contain 101 channels. However, decoding the .eeg files give me 52 channels for rat EE and FF, and 101 channels for rat GG. Is this particular to the downsampled .eeg data? If so, what are their corresponding shank/electrode locations? Based on the LFP itself, it seems that for the 52-channel data, channels 1-32 are in pfc, while channels 33-48 are in hpc (prominent theta).

- the session length information is at times inconsistent with the data length? For example: FF.167 .eeg file has 32.9633 minutes of data (assuming sampling rate of 1250), while the pdf indicates that the recording was 33.94 min. Visually examining the LFP power spectrum shows that the line noise is (correctly) at 60Hz when using sample rate of 1250Hz. If the recording was simply shortened, then this is no cause for concern. However, if samples were dropped, or a different sampling rate was used altogether, it would be problematic when trying to match the spike times in the _behavioral.mat files. 

- it appears that rat GG (in particular, session GG.149) has a sampling rate of 625, not 1250. This is inferred both by calculating the matching session length (indicated in the pdf) and by the location of the first line spectra (120Hz instead of 60Hz). Was this dataset (and others for rat GG) downsampled differently?

Thanks for your time!

Richard Gao

PhD Student @ UCSanDiego

Dear Richard,

1) I am sure that EE and FF should contain 104 channels (64ch PFC, 32ch CA1, 8ch for behavior signals) and GG should contain 101 channels (64ch PFC, 32ch CA1, 5ch for behavior signals), if the files are not broken.

How did you decode it? For checking .eeg files, Neuroscope is a good free software, which is avaiable at:

http://neurosuite.sourceforge.net/

2) Yes, the session length information in PDF is apporoximate one, and the length of the .dat (20kHz) or .eeg (1250Hz) files are more accurate.

3) GG has a sampling rate of 1250Hz for .eeg file. Again, maybe your decoding method would not perfectly fit with our data style.

Best,

Shigeyoshi Fujisawa

Hi Richard, 

I just found out that the code in the hc-2 dataset really helps for your question, you can convert  .eeg file directly through the LoadBinary.m included in the crcns-hc2-scripts.zip in  https://portal.nersc.gov/project/crcns/download/index.php. set up the right parameter and check the document and there you go!

Thanks for your reply, Dr. Fujisawa, and apologies for the late response - I thought I'd get a notification from the forum once a reply is submitted. 

And thanks Sitan, it looks like the code for hc2 does indeed parse the data correctly!