Helical per-particle scale factors lost in homogeneous reconstruction?

Helical Processing
1

Hi, I am hoping this question has a trivial answer. I have a helical refinement job that gives a resolution of 2.4 angstrom resolution. To test the Ewald correction, I ran two homogeneous reconstruction jobs with the two opposite Ewald signs, inputting the particles and mask from the helical refinement jobs.

However, the resolution got much worse in both the reconstruction jobs (3.0 vs 2.9). I see that the per-particle scale factor distribution in the reconstructions is uniformly set to 1, which I suspect could be the cause of this. So my question is:

Could it be that the homogenous reconstruction somehow ignored per-particle scale factors in from my helically refined particles?

A follow-up question would be, if so, how would I fix homogeneous reconstruction to include the scale factors?

Thanks! - Chuck

2

Hi @cvsindelar,

Thanks for the post. I believe the issue is likely that symmetry information was not used during the Homogeneous reconstruction job. In order to apply helical symmetry while reconstructing with Ewald sphere correction, the particles from the parent helical refinement must first be symmetry expanded using the three helical parameters (twist, rise, and helical symmetry order) that were printed to the stream log at the end of the helical refinement job. Then, the expanded set of particles can be passed into a homogeneous reconstruction job with Ewald sphere correction enabled. Do you find that this solves the resolution discrepancy?

Best,
Michael

1 Like
3

Thanks so much Michael. Your answer is exactly right on the money. I would have gotten more clues except that I forgot to give the symmetry order in the Homogeneous reconstruction job, so it didn’t even try to apply helical symmetry (hence the low number). I confirmed that Homogeneous reconstruction fails with a “not implemented” error when I did give the symmetry order. So I did the symmetry expansion, did the Ewald tests, and got a resolution bump of 0.08 A. Modest but it’s also the first time one of my own data sets ever reached a resolution where it benefited from Ewald correction. Hooray! Thanks and cheers - Chuck

1 Like