Different motion corr outcome when applying the gain reference to the same dataset using dev31 vs. dev34

[Siyu Chen]

Hi all,

We recently updated our warp2 version from dev31 to dev34. When I used dev34 to process the same Falcon4 dataset, I realized that I need to apply '--gain_flip_y' to the motion_ctf function, otherwise I see strong gain artifact after motion correction. However, I never have to worry about this flag when using dev31.

My question is whether dev31 was able to successfully apply the gain and if there is anything changed between these two builds, as I have used dev31 to process several datasets and I would like to make sure that they are well corrected and don't need to be re-processed. I never see apparent error messages applying the .gain file into the WarpTool commands in dev31 or dev34 and I can see the correct link in setting files.

In the long term, I also wonder where is the best place for a user to check what has been changed between different builds.

Thank you so much for your help!

Best,

Siyu

[teg...@gmail.com]

Hi Siyu,

There were significant changes to the EER rendering code that got merged just before a different pull request (AreTomo3 wrapper) necessitated the new dev34 version tag.

I've been using that code internally for several months now and thought it was stable, but our F4i's gain pattern is almost invisible, so I wouldn't have noticed if something flipped. I'll look into it.

I agree release notes would be great to have. I need to check if anyone has figured out an automated LLM-based solution.

Cheers,

Dimitry

[Siyu Chen]

Hi Dimitry,

Thanks for checking on this issue. I saw that in the latest release, the EER gain reference code has been modified. I wonder whether that addressed the discrepancy I saw in dev31 vs. dev34, and how I could help to check if the latest version works well.

Thanks again for your help!

Best,

Siyu