I was looking to the distribution of orientations coming from cryosparc with imposing C7 symmetry and I see they are distributed all over the sphere where as in Relion we are used to see them being distributed in a small region of sphere, like a melon slice (one of the 7 slices). Could you please let me know how the symmetry is being treated in Cryosparc and if it is different from Relion?
Hi @GhonchehM this is an interesting question. In fact in cryoSPARC to make things simple, we don’t even use the symmetry constraint during alignment of particles. Typically symmetry reduces the search space (as you see in Relion) but in cryoSPARC alignment using the Branch-and-Bound algorithm is very fast, so that we don’t attempt to exploit this gain, to keep the code simpler. Symmetry is only used during reconstruction (backprojection) in the typical way (same as Relion) to incorporate information from each particle multiple times (the order of the symmetry). This is why in cryoSPARC, particle alignments are spread evenly in all asymmetric units, rather than being confined to a single one as in Relion.
I see. Thanks for your response and of course all your efforts to release such a software.
Wish you the best.
@GhonchehM FYI in pyem there is a program called
star2bild.py which generates Relion-style orientation plots for display in Chimera. It can take any star file, including ones converted from cryoSPARC, and has a
--sym argument that applies symmetry to the angles. If you want to force the angles into one part of the rotation sphere, you can give it a try.