During the Global CTF Refinement job, CryoSPARC reports two different values for beamtilt one in mrad and in one angstrom (but only writes out the angstrom values into the particles file…). I am trying to understand what actually the angstrom value is referring to. The mrad value I can understand, as it is (presumably), the residual tilt of the beam from parallel. Is the angstrom value a movement of the image on the back focal plane?
I have also noticed that in my last job, for two groups on the second iteration, CryoSPARC did not report a tilt value in mrad, only in angstrom. Can this happen if the number of particles in those groups is too low?
Hi @thewanderingbeam! I’ll try explain what’s going on here with a bit of extra detail in case other users are curious!
Why Å?
Formally, what Global CTF Refinement is doing is correcting error in the CTF fit using higher-order Zernike polynomials. These polynomials describe phase shifts applied to the CTF, which we convert to lengths using the wavelength of an electron. In CryoSPARC’s convention, all CTF length parameters use Å, so particles get their third-order Zernike parameters stored with units of Å. This explains your observation that there are two ways of reporting the beam tilt, and why the particles only have the Å value stored, but not your questions about the mrad values.
Where does the mrad value come from?
Beam tilt is the most likely (but not the only!) source of a third-order error in the CTF fit, so we call the third-order error the “beam tilt”. If we assume that all third-order error comes from beam tilt, the beam’s tilt is proportional to the third-order term divided by the spherical aberration. This is the value we report in mrad.
Why is mrad only sometimes reported?
This presents a problem when the Cs value is low, either because the scope is Cs corrected or the Cs has been refined to a low value. In these cases, the reported tilt would be extremely high, when in fact the third-order error is likely coming from a non-beam-tilt source. For this reason, we do not report the tilt in mrad when the Cs value is below 0.1.
It may be that some of your exposure groups had their Cs parameter set below this threshold in the first iteration, which would resuilt in the tilt mrad not being reported in the second.
I hope that makes it all clearer, and please let me know if you’re left with any questions!