Extract Viewing Direction Distribution

3D Refinement
#1

Hello,

We would like to analyze the Viewing Direction Distribution in more detail. Is there a way to extract this information from one of the .cs files generated by Cryosparc?

#2

Hi @Pielhaas,

CryoSPARC will soon support automatic generation of BILD files that help display the viewing direction distribution, see this thread for more discussion on how BILD files display the viewing direction distribution.

The raw 3D pose data in CryoSPARC that are used to generate these files (along with the viewing direction plot) are contained within the alignments3D/pose variables, which store poses in an axis-angle representation in radians. This could certainly be extracted and worked with in a scripting environment via CryoSPARC Tools. The viewing direction specifically is a function of the pose, and extracting it would require converting the axis-angle vector to a rotation matrix and taking it’s last column. I could provide a quick code snippet to use built-in CryoSPARC functions to do this if desired – please let me know!

Hope this helps,
Michael

2 Likes
#3

Yes if you have pyem installed you have star2bild.py which creates the BILD output for any star file. You could also use it to get out the distribution itself to do your own analysis.

1 Like
#4

Dear Michael,

Thank you for your reply and offer to help. Instead of converting the cs files to star files to bild files and then extract the numerical values, I think it would be more efficient if one could calculate the angular viewing directions directly from the cs files. Given that there are several ways to convert an axis-angle vector to a rotation matrix, I would very much appreciate if you could provide a code snippet that allows to determine the viewing directions.

Best regards, Marcel

#5

Hi @Pielhaas,

Please see this thread for a similar script which you could use as a template for how to load particles into a python script using CryoSPARC Tools. For the code snippet:

# [...]
from cryosparc_compute import geometry
import numpy as n

particles_dset = job.load_output('particles') # particles from desired job
poses = particles_dset['alignments3D/pose']

def get_viewdir(poses):
    """ Get the viewing direction from 3D particle poses """
    Rs = geometry.expmaps(poses)
    viewdirs = Rs[...,2]
    return viewdirs

def get_azimuth_elevation(viewdirs):
    """ Get the azimuth and elevation angles from a viewing direction vector """
    viewdirs_azimuth = n.arctan2( viewdirs[:, 1], viewdirs[:, 0])
    viewdirs_elevation = n.arcsin( viewdirs[:, 2])
    return viewdirs_azimuth, viewdirs_elevation

viewdirs = get_viewdir(poses)
az, el = get_azimuth_elevation(viewdirs)

viewdirs is a 3D vector pointing along the projection direction.

Best,
Michael

#6

Thank you @mmclean for the snippet.

I have cryosparc tools 4.2.0 installed but am not able to get it to work. The line :

from cryosparc_compute import geometry

causes an error: No module named ‘cryosparc_compute’

Analogous the line in the linked script,

from cryosparc_tools.cryosparc.tools import CryoSPARC

yields a similar error.

Any ideas of what might be wrong?