Ensure adequate coverage of points specific to electron diffraction

[dagewa]

The points listed are:

• Unknown (different) center in each diffraction pattern
• Unknown azimuthal tilt axis position
• "flat" Ewald sphere
• Large uncertainty in the geometrical peak position (excitation error additional to other geometrical errors)
• Often presence of disorder / diffuse scattering in the crystals along the incident beam direction Each requires a short reply. I think the first three points are adequately covered in the manuscript, or in cited papers. I'll draft replies for each of those here. The final two points are not discussed in depth here so I am not sure what to do with those yet.

[dagewa]

I remain somewhat confused by these points. I hope it is clear that the first 3 points are well-covered by the manuscript. I am not sure what the 4th refers to. Is there an additional "excitation error" in ED that I am not aware of? The refined RMSDs are given in Table 1, and are all sub-pixel. Does that answer this point? I don't know. The disorder / diffuse scattering issue is not one we have looked at in detail.

[dagewa]

@cmaxtb @tgruene Do you have any additional opinion on the above? Thanks, D

[tgruene]

• center of diffraction pattern: as far as I know, we do not observe a beam shift with our instruments and our setup. My experience, of course, is limited to 4-5 instruments.
• the 'azimuthal tilt axis position' might refer to the rotation axis? Dials treats this as a vector with arbitrary orientation. We assume initially, it is parallel to the detector plane (z-coordinate = 0 in XDS coords), and the orientation within the plane can be read from the sum of all images. @cmaxtb : do you have an example image, or graph that shows this? From there, we are back at the refinement of the parameters, because one only needs to be correct within 5-10 degrees. The axis does not change when the detector distance is not change, so one can find the axis per distance setting from a radiation hard sample.
• the flat Ewald sphere only affects the correlation between the parameters during refinement, with the consequence, that the distance is not refined, only the cell (or vice versa, referring to reviewer 2)
• peak positions: We could cite Dimmeler et al, J Appl. Cryst, (2000), 33, 1102-1112 here, one of the older occurrences of 3D protein electron crystallography. However, I believe that profile fitting corrects for variations in the peak positions. I think the variation is the reason why XDS suggest unreasonably large REFLECTING_RANGE value, but I don't know if DIALS has the same issue?
• I have so far ignored diffuse scattering, I guess like most people to in MX and organic crystallography.

[clabbersm]

The central beam is drifting slightly, but we correct for that. No apparent beam shift, also we do stay continuously within the same settings during data collection and we do not realign intermediately. So there are no errors originating from that, appart from maybe the correct defocus(eucentric height) changing with crystal size and orientation

I have such an image, it’s only not very clear to see the rotation axis there. On the Talos the rotation axis is always the same, regardless of detector distance, it does change but the fei software will adjust the lenses such that it’s always orientated the same way. So you have to determine it once, using something radiation hard and well known. But not all microscopes are the same of course.

Yes, distance and uc are not refined simultaneously, but we can do a lattice refinement in refmac where we minimise the rmsd.

Agree with Tim, also the oscillation width per frame has some uncertainty since it was recorded manually. This also depends on how well the goniometer behaves, if it slips or does not have the same constant rotation speed through all the different angles of the stage.

Don’t have anything addional to comment on diffuse scattering, we did not specifically look for it or encountered anything extraordinary, with beam sensitive material it would be anyhow difficult I guess since eg exposure will affect the crystals strongly and radiation damage would induce defects and diffuse signal, without a stable reference it be hard to quantify any effect

[dagewa]

Reply composed directly in the draft letter. Closing the issue.