nextPYP
Particle refinement fails at merge step #66
-
|
My particle refinement split CPU step finished successfully. Now I am getting the following error in the merge step. The error is from a restarted run. Initially I ran out of walltime in the merge step. Any suggestions? p.s. Another thing I noticed is that when I rerun the pre-processing (not changing any parameters), in an attempt to complete just the merge step, the refinement (Frealign refine3d) starts running again. |
Beta Was this translation helpful? Give feedback.
Answered by
tofushan
Feb 16, 2024
Replies: 5 comments 7 replies
-
|
The error message indicates that you may have an empty reconstruction, this can sometimes happen if you set the criteria to include particles too stringently and no particles end up making the cut. For the second part, it looks like you are talking about the pre-processing and particle refinement blocks. If you change the parameters or re-run the pre-processing block, all downstream blocks will become out-of-date and will be automatically marked for execution. You can deselect specific blocks using the |
Beta Was this translation helpful? Give feedback.
0 replies
-
|
I haven't set any particle exclusion criteria. Fraction of particles under Reconstruction tab is 1. I checked folder Regarding my second comment, I meant particle refinement, not preprocessing. When I try to rerun the particle refinement run, just to try and complete the failed merge part, it starts running again the actual refinement using csp. |
Beta Was this translation helpful? Give feedback.
3 replies
-
|
Could you share one of your map1 or map2 file with us? We mean the voxels of reconstructions are all 0s, but the size of files is not necessarily 0. We've skipped merging intermediate reconstructions that are empty already, but it looks like all your intermediate reconstructions are empty. We need to check if that is the case. It would be great if you provide more information such as number of particles/projections and some statistics from one of your refinement panel. |
Beta Was this translation helpful? Give feedback.
-
|
I tried to check the MRC files but they cannot be recognised by IMOD. How could I check the values myself? What is the datatype and size of the header in these files? |
Beta Was this translation helpful? Give feedback.
-
|
They can only be read by cisTEM |
Beta Was this translation helpful? Give feedback.
-
|
I rerun the job with "verbose" on. I don't see that "skipping..." message. The full error is here: Edit: this error happens in the project where IMOD patch tracking was run on a different machine and the PKL files moved to the current machine. So perhaps some important files are missing? |
Beta Was this translation helpful? Give feedback.
1 reply
-
|
Thanks for providing the information. I'm sure that this is nothing to do with the empty reconstructions. I think it is fine with your operations too because PKL contains initial metadata that should not affect particle refinement. I'm wondering if there is an edge cases that we missed to consider.
|
Beta Was this translation helpful? Give feedback.
-
|
Thank you @tofushan. Here are the parameters I am using:
|
Beta Was this translation helpful? Give feedback.
2 replies
-
|
It is very likely that you hit the edge case we are missing... Could you try to change the particle radius to something smaller such as 290? This should fix the problem, and we will fix that in the next release. Thanks! |
Beta Was this translation helpful? Give feedback.
Answer selected by
abartesaghi
-
|
This issue is addressed in v0.6.3. |
Beta Was this translation helpful? Give feedback.
-
|
I am running with Particle radius: 280. Is it normal with my parameters (using 10 CPUs) that the merge steps is still running after 12 hours? One of the refinements had failed with this error message. I wonder if one failed sub-job could make merge to run forever? Update: This worked finally. I made the pre-processing block runnable, deleted all files in the refinement block, and then rerun both blocks. Merge took 20 min and completed without errors. |
Beta Was this translation helpful? Give feedback.
1 reply
-
|
You are right. The Merge job would wait until all the Split jobs successfully complete before merging all the reconstructions and metadata. Glad to see you can finally get it to work! |
Beta Was this translation helpful? Give feedback.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
It is very likely that you hit the edge case we are missing... Could you try to change the particle radius to something smaller such as 290? This should fix the problem, and we will fix that in the next release. Thanks!