Two new files on DANDI.

[rcpeene]

Hey all, we've completed the first working version of our internal processing and packaging pipeline. We've uploaded two experimental sessions 803496-2025-07-02-15-33-26 and 776270-2025-07-02-11-50-06. Links can also be found on the experiments page on the website.

Some notes about these:

• The new structure of the NWB files includes F0, dF, and dFF for both DMDs, and an improved stim table with stimulus labels. Also includes the ROI footprints, imaging plane and device metadata, and average projection images.
• Although 776270 had a calcium indicator, this is not yet packaged in the file.
• The rig software is still going through some changes and has some kinks so there were small inconsistencies in our harp timestamps across sessions. We investigated and implemented code to manage this in our processing pipeline. As a result, there is a nonzero chance that the timing information could be incorrect. I'd expect this to become immediately obvious to anyone doing analysis. If this is the case please submit an issue.

We plan to upload several more sessions soon.

[maierav]

Thanks so much!

I took a brief look at both files, and there seems to be an issue with sub-776270: The dFF timestamps run backward (~−0.1 ms steps), with occasional big forward jumps (~30.6 s). This makes alignment challenging.

There also are long periods with NaNs (see figure below). That should be less of a problem, but I thought I'd mention since I did not spot mention of recording breaks in the session notes.

I also checked sub-803496, and the timestamps of that file are monotonic - so that's good. The responses look a bit unusual, though. This could be due to alignment issues with the timestamps, and I am still looking into that. But it could also be due to the fact that these are interneurons and very strongly orientation tuned. I posted my code in an extra thread.

[jeromelecoq]

We are fixing a typo in the packaging code that could have caused this timestamps issue.

[rcpeene]

Sorry about the delay. The problem with sub-776270 has to do with how we are generating timestamps from our experimental trial structure. I recorded the problem and the solution in this github issue. Please let me know if there are additional problems.

[rcpeene]

I also supect that 803496 had a similar problem so I've corrected it and reuploaded.