Hi,
Currently on the fly correction for the curvature of the Ewald sphere is available in NU-refine, but not in local refinement.
Would it be possible to add it to local refinement?
Obviously it will not be useful if a small local region is masked, but there are cases (e.g. pseudosymmetry) where it is desirable to perform local refinement with a global mask, and in such cases EWS correction would be be helpful.
Cheers
Oli
Noticed this yesterday evening and was going to post this morning…!
Hi all,
We’ve added this to our roadmap 
In the meantime, note that Ewald Sphere correction can be applied to any particle stack (e.g. outputs of local refinement) by using the Homogeneous Reconstruction Only job. The only difference between this job and running Ewald Sphere correction in Non-Uniform refinement is that Non-Uniform refinement will reconstruct with marginalization enabled by default, whereas Homogeneous Reconstruction Only does not do marginalization. Marginalization usually has been observed to increase the FSC curve values, so that could cause a discrepancy between the outputs of Local Refinement and Homogeneous Reconstruction Only.
When implementing Ewald Sphere correction, we observed that accounting for Ewald Sphere curvature during particle alignment doesn’t make a discernible difference, which is why we didn’t push to add it to every job type. However, it makes sense to do this given the note about marginalization, so it’s on the roadmap!
Best,
Michael
Any ETA for this feature?
Dear @daniel.s.d.larsson,
We don’t have an ETA at the moment. We are wondering if there are datasets that you have encountered where the procedure of upstream local refinement + homogeneous reconstruction only is insufficient compared to local refinement with Ewald Sphere Correction fully incorporated into local refinement (in light of the differences outlined above).
Our main rationale for keeping the initial implementation of Ewald Sphere Correction to global refinements is that in our experimentation, the types of data that typically benefit most from ESC are large, rigid, high-res structures, usually the opposite regime of the datasets that benefit most from local refinement (small and/or flexible proteins). We’re curious though if you have a dataset that might benefit from both of these corrections simultaneously – I could see some particular cases where this might be the case (e.g. sym expanded data on a large virus)
Best,
Michael
One application would be processing ribosomes. They can go to fairly high resolution (better than 2 Å) and are also quite large (ballpark 300 Å), so EWS correction helps. Ribosomes have multiple subunits, which are rotating relative to each other. For that reason, one might want to do masked (local) refinement of each subunit and this would benefit from EWS correction.
Thanks all. This use case makes good sense, and we have recorded this request.
We have also recorded a related request to allow Volume Alignment Tools to modify the defocus of particles when shifting/re-centering is done (first discussed at the beginning of this thread), as this will at least allow for partial correction of the defocus gradient present in large samples, which is related to Ewald curvature
Kind regards,
Michael