I’m working with a 240 kDa, homohexamer that displays 3D conformational change in 3DVA with a 20 A high pass filter. Analysis of data sets with the apo and ligand bound structures show that there is more variability in the apo data set, as you might expect.
The issue is that this is not reflected in the refined maps. Both the ligand bound and apo maps overlap and show both in a closed conformation after localized refinement where I am running symmetry expansion and centering on a subunit by placing the fulcrum point at the center of my mask.
Is this because the protein exists in the closed conformation most often, even in the apo data set? Or is it because localized refinement is aligning the rest of the protein to the area selected for refinement refinement. Is there a way to validate 3D variability in the apo data set?
if the closed state is more populated in both datasets, you will see the map converge to that in refinement, since the extra signal from the more populated state will dominate reconstruction and alignment, and at typical thresholds at which you will visualize the map, you won’t see the shadow of density from the other state.
performing local refinement selecting one subunit will also allow both open- and closed-state particles to align to just that one subunit, minimizing the differences in conformation that would otherwise be visible across the rest of the complex. This is a subtle point, but would be especially important if the main difference between the open- and closed- conformation is between subunits rather than within each subunit.
Validation procedures for 3DVA are something we are working on, but you can for example take “clusters” of particles (or use the “intermediate” reconstruction mode in 3D variability display) to separate particles along the variability components that you see. Then, you can take particles from just one end of the spectrum and subject them to homogeneous refinement where you will get a gold-standard FSC estimate and all the usual validation methods are available.