Enter real space and Fourier mask radii for ab initio

Hi everyone,

I am working with an icosahedral virus dataset. As the box size is 888 A and the dataset is quite big, I ran out of memory when trying to classify the dataset in CryoSPARC. However, I classified a binned dataset in Relion and imported the most promising (unbinned) class into CryoSPARC.

Ab initio runs with all settings as default, except for Symmetry (I1) and Number of threads (1) yield a problem: On the inside of the virus, the refinement looks reasonable (it contains a membrane that looks fine). However the capsid looks like a ball made up from density slices.

I would like to run another ab initio reconstruction with masking the inside of the virus, thus only considering the capsid. Can I do this by adjusting the window radii in preprocessing or is this soft mask referring to something else?

Furthermore I would like to apply Fourier radii (lowpass and highpass if possible). Does the initial structure lowpass in Ab-Initio mean that the filter is only applied in the beginning of ab initio model calculation and then gradually reduced or will it remain constant throughout the model calculation? And do I influence this with the Fourier radius step or does this refer to something else?

And what is the Window structures in real space option about? Could it be helpful for me?

I would be very grateful for some help and explanations!


Is there really nobody who can help me?

Hi Anne, sorry for the long delay in replying to your message.

Here are some hopefully useful responses:

  1. Masking (in real space) is currently not supported in ab-initio reconstruction, though your use-case appears to be a scenario where it could be quite useful.
  2. Window radii in preprocessing refers to the soft spherical window that’s applied to the real-space images before they are used in the reconstruction. The window only masks out the corners of the images - the inner and outer radii specify the soft edge of the mask. So you can’t use that unfortunately to mask out the inner region of the images.
  3. The initial structure lowpass in ab-initio reconstruction is a bit confusing, sorry. That parameter is only used if an initial model is specified, which would mean that the experiment is not actually ab-initio. The other parameters called initial resolution and maximum resolution determine the gradual increase in Fourier radius during the reconstruction, and Fourier radius step determines the rate of increase.
  4. Window structures in real space applies a soft spherical mask to the structures during reconstruction. This is on by default and helps to keep the particle centered in the box during ab-initio reconstruction.

Hope that helps somewhat,

Hi Ali,

thanks a lot for your reply and explanations! I understand a lot more now!

Best wishes,