cryopros-recondismic usage (cryopros)

[julienbous]

Hi,
Could you elaborate a bit on the usage of "cryopros-recondismic" to generate the micelle. What initial volume to use (molmap generated from pdb?). What mask work the best, (tightness, softness), should the mask include only the micelle or protein+micelle? what parameters are preferred overall.
Kind regards,
Ju

[julienbous]

Hi again,
A rather tight mask around the protein component with softness about 10 Å on the molmap-generated map seems to work. Is that correct? Also, my protein has an extra membrane component that was a problem. If included in the mask, it ends up surrounded by micelles, but if the mask is restricted to the transmembrane part, the final result is chopped off in the final map. I ended up using the second approach and adding back the extra membrane component with vop maximum (Chimera). Is there a better way to do that?
Best,
Ju

[zhengdihan]

Hi Ju,
Thanks for reaching out!

What initial volume to use:
Yes, you can use a molmap generated from a PDB file as the initial volume.

What mask works best (tightness, softness):
In our case, we used a tight mask that only included the protein, excluding the micelle region. The idea is to keep the protein fixed and reconstruct only the surrounding region.

There are two functions related to micelle reconstruction:

cryopros-gen-mask \
--volume_path latent_volume.mrc \           # Path to latent volume  
--result_path ./mask.mrc \                  # Path to save the volume mask  
--threshold 0.1                             # Threshold value, chosen by the user  

Be careful when selecting the threshold value, it determines how much density to retain. Inspect the output file mask.mrc; it defines the fixed region, and only the region outside this mask will be reconstructed.

Then, to run the micelle reconstruction process:

cryopros-recondismic \
--gpu_ids 0 1 2 3 \                          # GPU IDs  
--task_name task_name \                     # Task name  
--box_size 256 \                            # Box size  
--Apix 1.0 \                                # Pixel size
--init_volume_path latent_volume.mrc \      # Path to initial volume  
--volume_scale 10 \                         # Volume scale  
--mask_path ./mask.mrc \                    # Path to volume mask  
--data_path raw_particles.mrcs \            # Path to raw particles  
--param_path param.star \                   # Path to STAR file  
--invert \                                  # Invert projections  
--dataloader_batch_size 24                  # Batch size  

Be careful with the volume_scale parameter, it controls the intensity of the protein signal within the micelle. If the micelle appears too weak, decrease this value; if it's too strong, increase it.

Regarding your case, I think your approach to generating the micelle looks correct.
Hope this helps!

Best,
Dihan

[julienbous]

Dear Dihan,
Thank you, that is helping a lot. I didn't notice you also made this script for mask generation, I used relion so far.
One last question, it seems that you do not specify any soft edge value for the binary mask. How is the soft edge value determined in your software?
Best,
Julien

[zhengdihan]

Hi Julien,

Glad that helped!
Are you asking how to determine the threshold value?
The mask is generated based on the threshold value as follows:
mask[i, j] = 1 if volume[i, j] > threshold,
mask[i, j] = 0 if volume[i, j] ≤ threshold.
This threshold is chosen by the user. You can use Chimera to inspect the volume intensities and select an appropriate value. For example, according to the formula above, a lower threshold will result in a larger mask.

Best,
Dihan

[julienbous]

Hi Dihan,
I am referring to the extend of the smooth transition between the 1.0 values inside the mask and the 0.0 values outside the mask to prevent introduction of ringing artifacts during operations operated in Fourier space.
Best,
Julien

[zhengdihan]

Hi Julien,

I see. I think the result would not differ too much, the full volume is defined as:

cryopros/cryoPROS_source_code/cryoPROS/models/network_mp.py

Line 174 in 0cc32f7

return (1 - self.mask) * self.micelle + self.volume * self.volume_scale

I believe using a soft mask is definitely feasible.

Best,
Dihan