[3dem] Issue with RelionRefine3D

[Guillaume Gaullier]

This has happened to me a lot with one of my datasets, although always during class2D jobs (I am assuming it didn’t happen to me during class3D because at that stage I had already discarded the problematic particles). As I understand it, it is more likely to happen if your autopicking picked a lot of "junk". Using stricter settings for autopicking might help.

The way I got rid of this error was to do a particle sorting right after extraction, and then select only the good particles. It can be difficult to tell the difference between good and bad particles, but after sorting if you look at the bad end you might be able to visually recognize obvious autopicking junk even with the very low contrast of the raw micrographs. Doing this consistently prevented the error from happening, but I am not very happy with this solution because I am sure I also threw away valid particles in the process.

I wish relion would be able to automatically flag these problematic particles, ignore them during the current classification, and at the end provide a file listing them so the user can discard them from the dataset for subsequent jobs (this would require the "Select" job type to be able to do subtractions between two star files: subtract the flagged particles from the initial set of particles). That would be much more sensible and much less frustrating than the current behavior of crashing the job (I wasted so many hours of computer time because of this error…), and would make class2D an even better way of "cleaning" a dataset.

Guillaume


> On Mar 26, 2018, at 15:17, Vishaka Santosh <vs4p at virginia.edu> wrote:
> 
> Hey, I know this issue has been asked recently, but I was wondering if anyone had any idea as to how to resolve this.  I received this error when attempting to run 3D auto-refine in Relion: 
> 
> Dumped data: error_dump_pdf_orientation, error_dump_pdf_orientation and error_dump_unsorted.
> slave 3 encountered error: ERROR: 
> No orientation was found as better than any other.
> 
> A particle image was compared to the reference and resulted in all-zero
> weights (for all orientations). This should not happen, unless your data
> has very special characteristics. This has historically happened for some 
> lower-precision calculations, but multiple fallbacks have since been 
> implemented. Please report this error to the relion developers at 
> 
>              github.com/3dem/relion/issues <http://github.com/3dem/relion/issues>  
>  
> File: /programs/x86_64-linux/relion/2.1_cu8.0/src/gpu_utils/cuda_ml_optimiser.cu <http://cuda_ml_optimiser.cu/> line: 1710
> 
> Does anyone have any idea as to what might cause this and any fixes?
> 
> -- 
> Vishaka Santosh
> UVA Class of 2011, BS in Chemistry with a specialization in Biochemistry
> VCU Class of 2013, MS in Biochemistry 
> _______________________________________________
> 3dem mailing list
> 3dem at ncmir.ucsd.edu
> https://mail.ncmir.ucsd.edu/mailman/listinfo/3dem

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.ncmir.ucsd.edu/pipermail/3dem/attachments/20180327/a02fe659/attachment.html>