3D reconstruction from simulated particles from a constructed 3D mrc file

3D Reconstruction
1

Hii

Similar to what has been discussed in 3D reconstruction of simulated particles with high SNR and Reconstruction from clean projections, I tried to reconstruct a 3D model with ab-initio from simulated 2D projections. In contrast to the other two my input volume is a map with exact atomic positions. If I generate 10^6 particles, deactivate CTF to images, set min and max defocus values to 0.5 and 1000, respectively, and set the max defocus value per particle to 0, and do a 3D ab-initio reconstruction, I don’t get anything close to my input structure.

If I generate 10^6 particles, deactivate CTF to images, set min and max defocus values to 0 for both, and set the max defocus value per particle to 0, and do a 3D ab-initio reconstruction, the reconstructed volume loses much of the detailed surface information.

So, I tried a different approach and simulated 2D projections outside of cryosparc and imported them. 1025 different orientations, all at 0 defocus, with no noise. When using those particles as input for the ab-initio and setting the symmetry to T, I get a 3D volume that is not symmetrical at all.

So two questions: Could you think of a way to improve the ab initio reconstruction when I simulate 2D projections in cryosparc, such that the details of my volume get resolved? And, where does the asymmetry come from when the 2D projections were made with a T symmetrical structure and ab-initio was told that the symmetry should be T?

Thanks in advance for your help.

Best,

Muriel

2

Hi @haugm ,

Thanks for the post :slight_smile:. To better help troubleshoot, could you let us know the following:

  • Which CryoSPARC version the Simulate Particles jobs were run on?

  • The value of the Signal-to-noise parameter in each of the Simulate Particles jobs?

Also,

If I generate 10^6 particles, deactivate CTF to images, set min and max defocus values to 0 for both, and set the max defocus value per particle to 0, and do a 3D ab-initio reconstruction, the reconstructed volume loses much of the detailed surface information.

  • In this case, was Ab-Initio reconstruction run with any symmetry enforced? Are you able to share images of 1) the density used to simulate the particles, and 2) the reconstructed density from Ab-Initio?

So, I tried a different approach and simulated 2D projections outside of cryosparc and imported them. 1025 different orientations, all at 0 defocus, with no noise. When using those particles as input for the ab-initio and setting the symmetry to T, I get a 3D volume that is not symmetrical at all.

  • Have you tried this experiment with identical particles, but running the downstream Ab-Initio with C1 instead of T symmetry? Do the results look reasonable?

Best,

Michael