Defocus vs resolution

[ashuthe...@gmail.com]

Dear all,

I  have two general and related questions.

1) Although, resolution in single particle in cryo em depends on so many factors, is there any benchmarking for defocus vs resolution for a given microscope? Ignoring (!) a very important factor which is quality of particles, can we calculate the theoretical possible resolution for a defocus range? For instance, when I am limited by defocus, given the smaller size of my protein, is there a way to find out the best I can get from a cryo em data?

2) Having acquired a data, is it possible to determine the resolution potential of the data before the refinement is complete? As in protein crystallography, we know the resolution of the data even before we have the structure, is that possible with cryo em data? I am aware that in crystallography where you need max a couple of hundreds of images to get the statistics, in single particle cryo em it will depend on 100s of 1000 of particles.

Thanks,
Ashu

[Steve Ludtke]

See below

On Feb 13, 2017, at 11:55 AM, ashuthe...@gmail.com wrote:

Dear all,

I  have two general and related questions.

1) Although, resolution in single particle in cryo em depends on so many factors, is there any benchmarking for defocus vs resolution for a given microscope? Ignoring (!) a very important factor which is quality of particles, can we calculate the theoretical possible resolution for a defocus range? For instance, when I am limited by defocus, given the smaller size of my protein, is there a way to find out the best I can get from a cryo em data?

There are many factors:

- microscopes with LAB6 filaments rather than FEGs have a much stronger dependency between defocus and envelope function

- use of a K2 in counting mode dramatically improves SNR across the entire spectrum, extending the achievable resolution on any microscope not hard-limited by some specific parameter

- as you get further from focus, the CTF oscillations get much closer together at high resolution. There is a point at which this becomes uncorrectable, but this point depends on the box size the phase flipping is done on. This is one reason why some strategies for phase-flipping operate on the entire micrograph. Of course that won't save you if your particle is too close to the edge...

- The threshold for information recovery is highly specimen dependent. An icosahedral virus, for example, provides a strong enough per-particle alignment that fairly high resolution information may still be recoverable even when the envelope has fallen off substantially.

2) Having acquired a data, is it possible to determine the resolution potential of the data before the refinement is complete? As in protein crystallography, we know the resolution of the data even before we have the structure, is that possible with cryo em data? I am aware that in crystallography where you need max a couple of hundreds of images to get the statistics, in single particle cryo em it will depend on 100s of 1000 of particles.

Yes, to an extent, but this is an imperfect science. If you have a highly monodisperse specimen frozen over holes with minimal degraded protein and minimal scattering from the buffer, then yes. You can look at the 1-D power spectrum computed by e2evalimage.py and assess where the CTF oscillations are no longer visible (or get too close together).

However, if you have a carbon substrate, or a lot of scattering from the buffer, then you need to carefully subtract these effects before you can assess the resolution potential. Typically this is done by boxing out particles and looking at the SSNR curves in e2ctf.py (Visual CTF in projectmanager). The SSNR explicitly takes the background into account, and is a pretty accurate estimate if the defocus is fit properly (which it normally should be). Our rule of thumb in the past has been that once the SSNR falls below ~0.02, the information is not readily recoverable, but there is a bit of variability in this cutoff.

In crystallography, the fact that you have crystal scattering peaks which you know must occur in specific locations gives you excellent separation from the background, and allows easy resolution assessment. Since the data is continuous and much noisier, there simply isn't a direct equivalent for CryoEM.

 

Thanks,
Ashu

--
--
----------------------------------------------------------------------------------------------
You received this message because you are subscribed to the Google
Groups "EMAN2" group.
To post to this group, send email to em...@googlegroups.com
To unsubscribe from this group, send email to eman2+un...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/eman2

---
You received this message because you are subscribed to the Google Groups "EMAN2" group.
To unsubscribe from this group and stop receiving emails from it, send an email to eman2+un...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

----------------------------------------------------------------------------

Steven Ludtke, Ph.D.

Professor, Dept of Biochemistry and Mol. Biol.         (www.bcm.edu/biochem)

Co-Director National Center For Macromolecular Imaging        (ncmi.bcm.edu)

Co-Director CIBR Center                          (www.bcm.edu/research/cibr)

Baylor College of Medicine                             

slu...@bcm.edu

[ashuthe...@gmail.com]

Thanks Steve,

This was very informative.

[Jason K]

Ashu,

I found it helpful to play around with e2ctfsim.py to think about envelope function. This is a GUI program that graphs simulated CTF for given input parameters where you can just move the sliders around to see how the intensity will change with defocus. Of course things like noise are not accounted for in this model.

Jason

[ashuthe...@gmail.com]

Thank  you Jason,

I had been playing with this option, essentially when I was looking at the quality of individual micrographs. Here, I was looking for something which can give an average estimate of all the micrographs together (in unreasonable comparison to crystallographic data processing). Because, the data I am currently working on have been collected in a defocus range of 2-4 microns. And to get a guesstimate through visual inspection, I need to see the distribution (and occurrence) of defocus values across the entire dataset. But yes, playing with a handful of images of known defocus value will give an idea of defocus vs resolution trend with the data.

Thanks so much again!

Ashu