I was running a Homogeneous refinement with a mask in cryosparc2 and was getting good results. However, using the same input files, when I run the Non-uniform refinement, the job crashes. This happened multiple times, giving the same error (please see the attached below). I have no clue why it’s crashing as I have used only default values. Could you kindly suggest?
Hi @AbhinavKV, can you please have a look at this post - the user encountered the same error and was able to change a masking parameter to fix. Thanks!
As suggested by Angel, I tried running the non-uniform refinement job again with the mask, but keeping the masking to static rather than dynamic. However, the job still crashed. I am not sure what is the exact problem. Interestingly, when I run the same job without a mask, the program runs beautifully and results in a dramatic improvement in my results as compared to the same homogeneous refinement job. But as I am sure that masking out one extremely flexible component improves my result, it is imperative to run the non-uniform refinement with the mask as it would definitely improve the results even further. Kindly find the latest error log attached with this post. Could you kindly suggest what is the right thing to do.
Hi Abhinav - how did you run non uniform refinement without a mask? In my hands when I do that (e.g. setting the mask type to ‘null’) it fails with an error.
Yes, my job runs quite successfully without a mask. But i need to run it with a mask as the results will be even better i think.
No, I mean how do you run it without a mask, that is, what do your inputs look like. If you just run without providing a mask, cryoSPARC will create one, and at least in my hands if I set the mask type to “null” it fails with an error.
Yes, I didn’t give a mask, but cryosparc did use its created mask in the refinement.
Could you please report if you manage to run the NU Refinement without any manual or cryosparc own dynamic / static mask ? I am having the same issue if I set to “null” or “0”: see below:
File "cryosparc2_worker/cryosparc2_compute/run.py", line 82, in cryosparc2_compute.run.main
File "cryosparc2_worker/cryosparc2_compute/jobs/nonuniform_refine/run.py", line 164, in cryosparc2_compute.jobs.nonuniform_refine.run.run_non_uni_refine
AssertionError: Bad mask type
Nope, I cannot run without a mask
There is still a bug that stops the job from running without any mask at all, but you should be able to get the same result by changing the dynamic mask parameters to be very wide. E.g. you can set
dynamic mask near(A) to 50A and
dynamic mask far(A) to 100A. This will be equivalent to applying no mask at all.
Note however that it’s unlikely that running without any masking at all will give very good results!