Floating FSC after 3D Variance using symmetry expansion


I am running 3D variance analysis of the S-trimer. I have first done symmetry expansion from C3. I use cluster mode in 3D Var Display to to cluster the different up/down states of the S-trimer RBD subunits. When I refine the particles belonging to different clusters (or their combinations) with NU refinement, sometimes, but not always, the FSC is “floating” i.e. the tail never drops cleanly to zero.

Now, as the symmetry expansion triplicates each particle, the selected cluster can contain the same particle several times (the original and its two symmetry copies). Initially they all have different orientations (here 0, 120 and 240 degrees) so the symmetry expansion alonw will not cause issues. However, the refinement (which is global) may rotate the particles to the same orientation and this happens especially when the cluster has no obvious breaks from symmetry (in some clusters in the S trimer case the conformation can be truly C3 symmetric, i.e. all RBDs down). This in turn, I suspect, is causing the spurious correlation in FSC.

Have others considered this in their data processing workflows and found solutions? Running just a local refinement, or no refinement, just a reconstruction keeping angles fixed, come to mind. Ideally, one could run NU refinement in local mode.

Best wishes,

I just realised that Local Ref New job can be used for this with the original mask.