Discussion: meetings/2025-11-25

[jeromelecoq]

My meeting slides: https://docs.google.com/presentation/d/1BYuxc7ujoE0lQcEua6_W0RykRcFWcEv5hMGpHNpD44E/edit?usp=sharing

[jeromelecoq]

Here are point of analysis we discussed today :

• Any issues with temporal alignment?

• Compare trippy, zebra vs gabors for receptive fields: Can we replace gabors with Zebra and free some recording time?

• Are oddballs responses present?

• Is the duration mismatch protocol well balanced?

• Do we sequence responses in prefrontal cortex?

Please reply to this thread if you are interested in specific aspects of this. Or mention additional thread you would be interested looking into. We proposed to use this forum to organize efforts.

[lrudelt]

Dear Jerome, Severine and Carter,

thanks a lot for the clear presentations and to the entire team for this amazing progress!

We would be very interested in comparing the receptive field mapping for the different stimuli and analysis approaches (first for the Neuropixels, this could of course later or in parallel also be applied to the SLAP2 data).
In principle, we could also easily extend our previous analyses from SLAP2 to the Neuropixels data to assess how responses are modulated by context. This might also help to detect any problems with temporal alignment.

Is there overlap with the plans or ideas of any other group?

Looking forward!

[jeromelecoq]

I have reached out to Sophie and Max (who presented Zebra). Sophie expressed interest to look into those assets. Perhaps you should discuss with her to coordinate.

[severine2305]

P3PilotsReview.pptx

[jeromelecoq]

The other discussion point is about introducing wheel into the cage. Our experience is that mice will more spontaneously (and perhaps more gracefully) walk on the wheel while head fixed when we add this to the box. We also used it as a way to provide enriched environment.

The concern raised earlier is that mice could learn the fixed world environment while walking instead of the VR coupling.
We discussed this tradeoff today. We would like to propose to put back this wheel as it will favor more walking behavior in our experience. I am hoping to get perspective on this in this thread. Is there any data that speaks to one outcome or the other? Are there controls that should be done?

[koweiss]

I'm in favor of putting back the wheel to encourage walking behavior. This may improve our ability to dissect the mismatch responses since responses will be amplified in general (not only during the oddball) due to decreased noise correlations, consistent with what others have shown
https://www.jneurosci.org/content/37/14/3764
https://www.cell.com/neuron/fulltext/S0896-6273(10)00059-0

We could then compare the responses given the difference in contexts (fixed world vs VR coupling).

[Alishamsnia]

Hi, I am Ali Shamsnia, a master's student from Najafi Lab @farznaj @najafi-laboratory . I did alignment analysis (regarding addressing the point 1 mentioned). See my pilot analysis figures:
alignment_report.pdf

Please let me know your thoughts on how to proceed and/or coordinate with others working on the same dataset.

[farznaj]

im looking at this page
https://allenneuraldynamics.github.io/openscope-community-predictive-processing/detailed-experimental-plan/#experimental-design

my understanding was that control blocks dont have oddballs, but the duration control block schematic does show a duration oddball. is that correct? but then what's the purpose of a separate duration session type?
@jeromelecoq

also what does n=40 each mean for each control block? i think i know what it means for the randomized block, perhaps 40 of each orientation. but what does it mean for the duration control block, and for the sequence control block?

[farznaj]

another question @jeromelecoq
why dont we add the "duration oddballs" to the set of 4 oddballs that are presented in the standard/sequence/closed-loop session types, and eliminate the duration session types entirely?

[farznaj]

and one more @jeromelecoq

do non-oddball sessions [the initial 2 weeks of recording] have the 4 control blocks too?

ps. im working closely with Ali and Arya, 2 masters students in my lab, on the npx data analysis, so i want to make sure we fully understand all the details. thanks for helping me better understand the details.

[jeromelecoq]

1/ The control blocks should show all trials but in shuffled context with the same incidence rate as during oddball block. So you should be able to rule out some form of long term adaptation because of the interval between oddball presentations.
2/ n=40 should be the number of repeats of each trial type. In shuffle context, it should still be relevant to build statistics.
3/ the challenge is stat power duration oddball are additional trials. In addition you need to present a single drifting grating with a fixed duration many time to set the expectation. This is not the case for sensory-motor, and sequence block. You could add to standard but that means fewer trials for each type.
4/ The non-oddball sessions do not have oddballs blocks yes. You can see the design here : https://github.com/AllenInstitute/mtrain_regimens/blob/Openscope_P3_v52/regimen.yml
And check the code to generate the stim table here : https://github.com/AllenNeuralDynamics/openscope-community-predictive-processing/blob/main/code/stimulus-control/src/Mindscope/generate_experiment_csv.py

Feel free to ask any other questions!

[Alishamsnia]

I attached the entire sessions/mice analysis here:
alignment_report_820454-2025-11-04.pdf
alignment_report_820454-2025-11-05.pdf
alignment_report_820454-2025-11-06.pdf
alignment_report_820454-2025-11-07.pdf
alignment_report_820459-2025-11-10.pdf
alignment_report_820459_2025-11-12.pdf

I also modified the notebook to handle any experiment settings as previously it was developed on just one session type. Also, I improved the comments and added some explanation. I added my alignment code to the last cell of this notebook for your reference:

intro_to_ephys_nwbs.ipynb

[Alishamsnia]

This is the latest version of the notebook:

intro_to_ephys_nwbs.ipynb

Also I wanted to mention that for the sessions: "820459_2025-11-12_14-34-33" we missed one of the probes. We don't have data from probe E.

[maierav]

Thanks so much, Ali! This is excellent work. Here is an adaptation for Google Colab:
https://colab.research.google.com/drive/1v7rjT6gDnF4rYLRptH3SyWBfrKQP1jAA?usp=sharing
I had to shorten the time series given memory (RAM) limitations.