Reintroducing subtracted volume

Dear all,
I used the ‘Particle Subtraction’ job to remove a domain which was throwing off my refinement and got a decent map of the rest of the protein. Now that I have the correct Euler angles on the remaining part of the protein , I want to reintroduce the domain and run a reconstruction job. For that I have exported my particles from before subtraction and after subtraction refinement and converted both to star files. My plan is to merge them into a new star file with the paths from the former and alignment information from the latter and then import that to cryosparc.

Does that sound like a good way to do it? I am now trying to learn how to write scripts so that I could write one to merge the star files in the manner I mentioned above.
Is there any other way that I can reintroduce the subtracted domain?


This? Homogenious refinement of full particles after 3D classification with particle subtraction - #2 by olibclarke

1 Like

Thanks a lot for your response.
I tried that but am getting the following error:
[CPU: 390.4 MB] Traceback (most recent call last):
File “cryosparc_worker/cryosparc_compute/”, line 85, in
File “cryosparc_worker/cryosparc_compute/jobs/refine/”, line 1076, in
File “cryosparc_worker/cryosparc_compute/engine/”, line 2959, in cryosparc_compute.engine.newengine.reconstruct
File “cryosparc_worker/cryosparc_compute/engine/”, line 29, in cryosparc_compute.engine.cuda_core.initialize
pycuda._driver.LogicError: cuInit failed: forward compatibility was attempted on non supported HW

Would you happen to know why this error would take place?


Sounds like computation infrastructure issue. Are you up to date? Also, could you try 3D classification with 1 or 2 classes? Reconstruction historically not ideal result although I know you want to retail eulers

Hi Sagnik,

This could indicate an inconsistent (CUDA toolkit, nvidia driver and/or pycuda) system state, similar to the the situation described in this thread:

If you would like some help identifying the problem, please post: