[3dem] Relion Polishing with EER-movies containing varying frames and fractions

[Takanori Nakane]

Hi,

 > Therefore, each subset has a different amount of EER fractions ( in the range of 600 – 1200 fractions).

Did you run 10 motion correction jobs with different EER groupings and
then join micrograph STAR files?

 > Is this a bug in relion?

Probably yes. The program should read EER grouping for every movie but
somehow this is not working.

 > How can I overcome this?
 >
 > Is the only way to first make movies with 50 frames from each subset and not work on the original EERs anymore? I processed fine to about 4 Angstroem
 > without doing so, is it really necessary just for polishing to now make 50 frame movies, or work on each subset separately? (As far as I understand
 > relion, it does not have a GUI-function to generate 50 frame movies from EER data, but it has a command-line function to do so, am I right?

Run Polish with the same run_data.star and postprocess.star but with
each micrograph STAR files BEFORE joining.
Polish with process only particles on the given micrographs.
Thus, after a chain of 10 Polish jobs, all particles will be polished.

Best regards,

Takanori Nakane

On 2023/04/03 21:56, Klink, Björn wrote:
> Hi there
> 
> I encountered a weird problem with Relion 4 polishing in a dataset with 10 subsets of data from a Falcon 4i camera (EER movies).
> 
> Each subset was measured with the same total exposure of 50 e-/A^2, but the exposure time per micrograph to achieve that was significantly different 
>   in some subsets.
> 
> Therefore, each subset has a different amount of EER fractions ( in the range of 600 – 1200 fractions).
> 
> Since in relion one can only choose how many EER fractions are combined to one frame, the amount of frames after EER fraction combination varies from 
> 49 to 51.
> 
> I figured out that with the –find-shortest option, a polishing job on all the datasets combined would find out that 49 is the amount of frames that is 
> present in all micrographs, and I assume on datasets which are not in the EER format, everything would work out fine from here on.
> 
> However, since the amount of fractions of the original EERs is still different, I found that the program crashes with the error message like:
> 
> EERRenderer::renderFrames(frame_start = 967, frame_end = 987),  NFrames = 918
> 
> EERRenderer::renderFrames(frame_start = 904, frame_end = 924),  NFrames = 918
> 
> n: 
> /Applic.HPC/Easybuild/zen3/2021a/build/RELION/4.0/foss-2021a-beta-543db0a-CUDA-11.3.1/relion-543db0a84a6735cc975be93fbafc0ffe683ebc57/src/renderEER.cpp, line 315
> 
> ERROR:
> 
> Invalid frame range was requested.
> 
> I found that when I limit the amount of fractions to sth like 20, relion would not crash (maybe because then the highest amount of fractions in one of 
> the datasets that contributed to the first 20 frames is not lower than the total amount of fractions in another subset). However, if 30 frames from 
> one dataset are made from e.g. 750 fractions, while in the other subset the total 50 frames only were made from 690 fractions (because of higher 
> exposure per second), the program would crash with above error.
> 
> Is this a bug in relion?
> 
> How can I overcome this?
> 
> Is the only way to first make movies with 50 frames from each subset and not work on the original EERs anymore? I processed fine to about 4 Angstroem 
> without doing so, is it really necessary just for polishing to now make 50 frame movies, or work on each subset separately? (As far as I understand 
> relion, it does not have a GUI-function to generate 50 frame movies from EER data, but it has a command-line function to do so, am I right?
> 
> Best wishes
> 
> Björn
> 
> 
> _______________________________________________
> 3dem mailing list
> 3dem at ncmir.ucsd.edu
> https://mail.ncmir.ucsd.edu/mailman/listinfo/3dem