I have a naive question about which defocus value to use during data collection. I usually set the target defocus range of -0.8 ~ -1.8 um when I collect micrographs on EPU, but when CryoSparc estimates the CTF, it tells me the estimated defocus is -0.4 ~ -1.3 um. When I collect multishots of SerialEM, the offset is even bigger. I am wondering which value I should use. Should I trust the defocus target setting of EPU or SerialEM, or should I trust CTF estimation and add back the offset to obtain the expected defocus (for example set target defocus of -1.8 ~ -2.8 um for compensation so CTF estimation is -0.8 ~ -1.8 um?)
If you see robust power spectra and clear 1D ctf plots than the estimated defocus values are much more precise than what you set during acquisition, given you determined ctf with proper cs, ht, apix etc.
Usually defocus values vary quite significantly during acquisition, which is good indeed.
No need to worry about the nominal defocus as long you can see particles.
It’s not uncommon for there to be a systematic offset in the nominal defocus, because the nominal defocus is derived from the objective magnetization via microscope calibration. As long as your true (estimated) defocus is in ~0.4 - 2.5 µm you should be happy!
(Close to focus fitting can be less accurate, and far from focus the CTF envelope is worse and the delocalization is more severe, as are any defocus-dependent effects such as magnification with a not-quite-parallel beam).