In either “Refinement (New)” or Global CTF refinement, if spherical aberration is deselected, it is still refined - same for tetrafoil. It’s only if both are deselected that neither are refined. Is this a bug?
Also the spherical aberration refinement seems unstable sometimes - or at least it refines to values that are quite a way away from the expected value (2.7mm) - even though the data is fairly high resolution (2Å apoferritin). So I’m not sure whether to believe the refined value, which is why I’d like to try turning it off.
Hi @olibclarke, cryoSPARC logs the messages you see when either “Fit Spherical” or “Fit Tetrafoil” Aberration is enabled.
Can you confirm that below those lines you see this?
Not updating spherical aberration values for group.
This would indicate that spherical aberration was not refined.
Hi @nfrasser - so it refines the parameter but just doesn’t save it for the next round? Are both parameters refined independently, they are not interdependent?
@olibclarke the spherical aberration and tetrafoil Zernike polynomials are orthogonal, so their estimated values are independent of eachother and correction of one does not affect the other (until e.g. one correction causes the alignments to change in a subsequent iteration of refinement)
when the even-Zernike code is engaged it estimates all the even polynomial coefficients regardless of what was asked for, but only the parts that were asked to be refined will have their values updated (at least this is the expected behaviour)
Got it - thanks @apunjani!
If they are independent - perhaps it would be worth making plots showing their contributions independently? I find it hard to assess particularly for spherical aberration whether the refinement is “working” - even for very high resolution data (<2Å), for some datasets it refines to values that I think are probably wrong (2.9-3.0mm), but I don’t have any way of really assessing that, because the plots for the even zernike include both components.