Per micrograph defocus refinement?

[Oliver Clarke]

Hi Sjors,

The CTF refinement module in Relion 3 is fantastic and very fast!

I notice that one can refine astigmatism per micrograph or per particle, but defocus can only be refined per particle (not per micrograph).

Would it be worth adding per micrograph defocus refinement at some point? This might be beneficial for example where most of the error in the initial defocus is systematic (rather than due to variations in z-height of the individual particles), due to having sparsely distributed large particles and hence relatively low signal at high res in the power spectra.

Or does Relion already do this internally - a global, per micrograph refinement, followed by local, per-particle refinement of defocus values?

Cheers Oli

[Takanori Nakane]

Hi Oli,

Currently, RELION refines defoci independently, without restraining neighbours like Gctf or Warp. In future, implementing a statistical model of per-particle defocus variations might be beneficial.

Or does Relion already do this internally - a global, per micrograph refinement, followed by local, per-particle refinement of defocus values?

No.

Best regards,

Takanori Nakane