Homogenous refinement vs Helical refinement


I am very new to helical refinements. In my case, the assmetric unit is around 24 kDa and there seems to be around 7 units per twist. When I am performing a homogenous refinement using the particles, I get to around 3.8 A resolution, and the map looks like the resolution estimate is correct. However, when I try to perform helical refinement with approximate symmetry parameters, I get similar resolution estimates. But the map is nowhere close to what I get from homogenous refinement. The mean-squared error plot from helical refinement has the minima close to 3 - from what I undetstand, that means the symmetry parameters are probably close to the actual values. So, I was wondering why I am not getting good quality maps from helical refinement.

Any help or insight is appreciated.

Thank you

Hi @Adwaith99,

Thanks for the question! We have some follow-up questions

  • How did you do particle picking for this dataset? Are particles picked closely along the filament?
  • How did you estimate the helical symmetry parameters?
  • Is the homogeneous refinement done with a mask around the entire filament, or just one of the ASUs?

If possible, could you post (or DM to me) the following plots?

  • FSC plots from homogeneous and helical refinements
  • The mean squared error from the helical refinement

Thank you,

Hi @mmclean,

Thank you for the response!

  • These particles were picked using Template picker. The particles are picked closely along the filament.

  • I estimated the helical symmetry parameters using the Symmetry search utilility using the good map that I got from homogenous refinement. And the values it gave made sense as well.

  • For the homogenous refinement, I did not provide any mask. So, it used the default mask that homogenous refinement generates, and it covers the entire filament.

I am also attaching the plots you requested:

  1. FSC for homogenous refinement

  2. FSC for helical refinement

  3. Mean squared error plot for the helical refinement

Thank you

Hi @Adwaith99,

Thank you for the quick response; nothing stands out to me as problematic from these plots, nor from your workflow. Both are at reasonable resolutions, the MSE plot indicates that the symmetry is reasonably locked-in (whether it is correct or not is a separate question).

I would be curious to see the discrepancy in the density maps solved from homogeneous vs. helical refinement. Do you see secondary structure in both maps? If not, it is certainly possible that the helical refinement locked into the wrong symmetry. (It could then be worth taking the good 3.4Å map from homogeneous refinement, searching it for plausible symmetry parameters, and then using these for helical refinement – although perhaps you already tried this as per your original post).