New notebook: Neuropixel: Accessing Local Field potential. Visualizing LFP responses to a trial event, simple example with latencies, plot responses in different epochs.

[jeromelecoq]

Carter will use both visual coding LFP NWB and Glow NWB files with LFP data to plot the averaged response to gratings for a number of repetition of the stimuli. There should be a response at the onset of the stimuli in the LFP data.

It is worth talking with Jake about the rationale for this analysis. And this will also allow to debug AllenInstitute/OpenScopeNWB#46

[jeromelecoq]

The Notebook will

1. pull LFP NWB
2. Extract stimulus time table
3. Choose one repeated stimulus type
4. take the Start time of these repeats
5. Use the start time to pick LFP data around temporal windows
6. Specify pre and post intervals for the windows
7. Generate an overall array with the acitivty, Time x LFP node x Repeat
8. Visualize this array with imshow or similar
9. Plot the average across repeats for one LFP node
10. Visualize the average across nodes with an image
11. Once repo is public, you could ask Hyeyoung for feedbacks so that is it useful.

It is likely there is a timing alignment bug to fix.

[rcpeene]

In my exploration of some LFP files for this task, I noticed that the rate attribute of the ecephys time series data is None. It might be useful if we could actually give this a value. Does some rate value exist in the raw data, or is this infeasible?

[Ahad-Allen]

Carter and I looked over this yesterday and will be doing a meeting about it later today, It seems like there is a time series within the LFP files that is used to map from the frame times to seconds, but we need to explore the documentation a little more before we are fully confident with this.

[jeromelecoq]

@rcpeene is looking into temporal alignment. The stimuli start 90s in approximately. The LFP recording starts 4 s in approximately.

Next steps include averaging responses.

[jeromelecoq]

@rcpeene does have the intervals of data now.
We discussed issue of non-regular time acquisition for averaging many trials.
the suggestion is to do nearest neighbor interpolation to bring back all trials onto the same time base.

[Ahad-Allen]

I will work on updating the subsampling to reflect Jake's requests.

[rcpeene]

@Ahad-Allen does this mean I should wait on LFP timestamp analysis to continue progress on this script? Interpolation may not be necessary.

[Ahad-Allen]

Hmm, that might be correct. I'll have an updated file for you either the day of today or early tomorrow to look at.

[jeromelecoq]

We discussed current state of the notebook and Carter will :

• Add some doc
• Add an image visualization
• Add the standard deviation intervals.
• Make a PR

Jerome will then review and we will share with Hyeyoung and Jake for feedbacks.

[jeromelecoq]

• We will send update LFP NWB files to Jake once Carter has approved the newer files.
• Ahad will upload the Dandi once its approved
• Ahad will write an email to Jake summarizing the change along with an introduction on the notebooks from Carter
• We need to ask him WHY we need 4 times larger NWB files because this is a big ask.

[Ahad-Allen]

@rcpeene and I looked over these files and they seemed good! The upload failed on Friday because of an error I had with the ecephys version of auto uploading. I will retrigger the upload with some changes and then email Jake.