LocaliZoom links (nonlinear)

ingvildeb commented 3 years ago

I have three datasets for which I would like to have LocaliZoom links for the nonlinear registration.

The JSON files of the first two datasets can be found here: Z:\HBP_Curation\QuickNII_projects\Projects\Incomplete_projects\2019_Dopamine_receptors_developing_mice\D1R\P70\Nonlinear_JSONS and the corresponding Navigator blocks are:

1362_4644_6646 (C58) 1362_4624_6626 (C60) 1362_5195_7438 (C8) 1362_5166_7408 (E15) 1362_5189_7430 (E16)

1362_4844_6906 (C16) 1362_4645_6647 (C17) 1362_4845_6907 (C19) 1362_4926_7068 (C12) 1362_4665_6667 (C22)

And JSON files for the third dataset can be found here: Z:\HBP_Curation\QuickNII_projects\Projects\Incomplete_projects\2019_Dopamine_receptors_developing_mice\D2R\P70\Nonlinear_JSONS with corresponding Navigator blocks:

1362_5191_7433 (D122) 1362_5193_7436 (D139) 1362_5198_7448 (D143) 1362_5190_7432 (D120)

Thanks!

Ingvild

Majpuc commented 3 years ago

Very good Ingvild! Do you have the same file names for the Tiff files in N3 and on CSCS? The renaming before uploading to CSCS creates troubles..

ingvildeb commented 3 years ago

Hi Maja,

The file naming convention is the same as in CSCS, but of course the navigator tiffs have the prefix that is automatically generated in that system. However, we might have renamed some files (changing serial order) after they were uploaded to N3, but in these cases the names have also been changed for the corresponding file(s) in CSCS. So all in all, file names should be the same, but with prefix in N3.

For one of the datasets (the second one listed above, C16-C22), however, we have actually deleted some images in N3 of very damaged sections. Because we had already released the data to EBRAINS, we did not delete these files in CSCS, so this I can see creating trouble. I wasn't aware that you had to fetch files from CSCS to create the LocaliZoom links, or is this perhaps a test case for a new procedure?

Anyway, I would suggest going forwards with the first and the last dataset listed above. For these, the file naming convention is the same as in CSCS, but of course the navigator tiffs have the prefix that is automatically generated in that system. However, we might have renamed some files (changing serial order) after they were uploaded to N3, but in these cases the names have also been changed for the corresponding file(s) in CSCS. So all in all, file names should be the same, but with prefix in N3. Are any of these things creating a problem?

eapapp commented 3 years ago

Hi All,

I added a new sheet to the DZ Buckets table to keep track of file naming in Navigator vs on CSCS. I started with adding those series that Dmitri has already downloaded. I can't find the non-linear anchoring JSON for the C14 series though, @ingvildeb do you know if there is one?

If the order (serial no.) of the files is the same on CSCS, in Navigator, and in the anchoring JSON, that's a good starting point. I think Gergely has arrived to the conclusion that the best way forward is to re-generate the JSON files with the CSCS file names instead of Navigator file names where the two are different. This is not a problem as long as we can follow what should be renamed to what (hence the above sheet).

For one of the datasets (the second one listed above, C16-C22), however, we have actually deleted some images in N3 of very damaged sections. Because we had already released the data to EBRAINS, we did not delete these files in CSCS, so this I can see creating trouble.

This will not cause any problems unless the deleted sections are listed in the anchoring JSON (I guess this is not the case).

I wasn't aware that you had to fetch files from CSCS to create the LocaliZoom links, or is this perhaps a test case for a new procedure?

Yes, this is the new way of creating LocaliZoom links. The plan is to store all the image pyramids in the image service (on CSCS but in a hidden place), and make LocaliZoom fetch image tiles directly from CSCS instead of Navigator. This is essentially what we (SP5) have promised towards the EU and needs to be fully implemented as soon as possible.

ingvildeb commented 3 years ago

Hi Eszter,

Sorry, no, there is no non-linear anchoring for C14. I have only done it on a subset that I'm going to use for analysis (the ones listed in my entry above). Serial numbers should be the same in all three instances. And you are right, the deleted sections are not listed in the JSON file.

Best, Ingvild

Tevemadar commented 3 years ago

&view: toggles view-only mode: markers and annotation controls are hidden, filmstrip starts with fully transparent overlay
&opacity=20: sets overlay opacity to 20% (or anything else), instead of starting with outlines (100%) They can be appended to the base link, https://www.nesys.uio.no/NonLinTest/filmstripzoom.html?atlas=300100&series=6907&opacity=20&view

Majpuc commented 2 years ago

status 03.09.2021. Issue linked to issue 120. In progress.

Majpuc commented 2 years ago

Finished on oct 18th 2021

Neural-Systems-at-UIO / LocaliZoom

LocaliZoom links (nonlinear) #20