Generating a sphere for 3D refinement

[Jackie Tan]

Hi Dr Ludtke,

I wanted to ask if it was possible to generate a sphere using EMAN2 for 3D refinement, i.e. running e2refineeasy.py on a sphere with particles I've picked.

So far, what was done is that we used heavy filtering but that does not generate a perfect sphere. My question is whether there was an easier way to generate it. Thanks!

Cheers,
Jack

[Steven Ludtke]

Hi. Not quite clear on what you're after?  You want a perfect sphere in a 3-D box? 

for example, this should replace whatever is in volume.hdf (any file format is ok) with a sphere of radius 30 pixels

e2proc3d.py volume.hdf sphere.hdf --process testimage.circlesphere:fill=1:radius=30

This will give you a list of the available 2-D and 3-D "test objects":

e2help.py processor testimage -v 1

Note that, of course, it is impossible to represent a perfect sphere in cartesian coordinates, and this method will produce sharp 'jaggies' at the edges. An alternative scheme would be to make an image of all 1.0 then apply a soft edged spherical mask to it:

e2help.py processors mask -v 1

--
--
----------------------------------------------------------------------------------------------
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.

[Jackie Tan]

Hi Dr Ludtke,

Thank you for getting back to me so quickly, and I appreciate the advice! What you said was precisely what I needed, actually. It generated a nice sphere, which I used for refinement.

I'd hate to bother you again, but I wanted to ask about validating negative stain data. More specifically, is it useful to validate a negative stain model using tilt-pair validation? I tried to perform this validation but I have been having difficulting getting the right answer, i.e. the correct tilt pair angle and tilt axis. Is it because of the fact that it is negative stain data, or is my data inherently problematic?

Cheers,
Jackie

You received this message because you are subscribed to a topic in the Google Groups "EMAN2" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/eman2/_dn7tZwT8ik/unsubscribe.
To unsubscribe from this group and all its topics, send an email to eman2+un...@googlegroups.com.

[Steven Ludtke]

Hi Jackie. Tilt pair validation is, in fact, mainly aimed at insuring that you have the correct structure at low resolution, when the features of the map are not yet sufficient to judge whether the map is likely correct. That is, once you achieve 6-7 Å resolution, it is highly unlikely that you could achieve something that looks like a protein if you have the incorrect quaternary structure. While tilt-validation is never a bad thing to do, at high resolution with an obviously nice looking structure, I would not complain if it weren't performed. If you are at low resolution, however, such as the 20-30 Å negative stain will typically provide, then tilt validation is critical to demonstrate that you have correctly assembled the 2-D projections into the correct 3-D envelope.  Successful tilt validation is rarely trivial to achieve, however. If you look at published examples, like our own:

Murray, S. C., Flanagan, J., Popova, O. B., Chiu, W., Ludtke, S. J. & Serysheva, I. I. (2013) Validation of cryo-EM structure of IP₃R1 channel. Structure. 21, 900-909. PMC3696195

You will see that the results don't necessarily have to be extremely good to claim success. If you have the wrong quaternary structure, the tilt validation results will be extremely poor. If you are getting a cluster in your results, but it isn't in the correct place, it implies that your structure is probably correct, but that you may have a problem with your definition of the experimental geometry.

hope that helps.

cheers

[Paul Penczek]

Hi I would add to it that in case of stain data it makes much more sense to solve the structure using the Random Conical Tilt and do not bother with any validation at all. Pawel Penczek Sent from Yahoo Mail for iPhone

---

From: Steven Ludtke <slud...@gmail.com>;
To: <em...@googlegroups.com>;
Subject: Re: [EMAN2] Generating a sphere for 3D refinement
Sent: Thu, Oct 16, 2014 11:48:59 AM

[Steven Ludtke]

I'd have to disagree with that point. RCT either involves A) relying on preferred orientation, meaning you will always have a missing cone or B) doing multiple RCT reconstructions for different 'base' orientations of your particle then aligning and averaging them in 3-D to clear up the missing cone.   Admittedly if you have a very strongly preferred orientation you are somewhat screwed no matter what you do, so in that context you might as well do RCT. 

Experimentally, RCT is MUCH more challenging to do, for very little, if any, gain. 

- You have to work at very high tilt, which isn't as bad in stain as in cryo, but is still not trivial (to get good results)

- The RCT processing involves multiple, somewhat confusing, steps to get a single reconstruction

- You also have to deal with some 3-D alignment/averaging for the final step, which has its own set of associated problems.

Tilt validation on the other hand only requires a small handful of tilt-pairs, and they are generally at very small tilts. MUCH easier.

I'm not saying that RCT isn't useful or can't be made to work. There are clearly cases where it is very useful, but I still view it as a method of last resort.

[Jackie Tan]

Hi Dr Ludtke,

Ah ha! That clears things a little. I read the paper not too long ago, and found it insightful in elucidating the importance of tilt-pair validation for structures obtained via cryo-EM. I did not recall negative stain structures being mentioned, which prompted this question. Nonetheless, I agree that it's indeed a useful tool. 

That said, the protein I'm working with ranges from 180-360 kDa, and I wasn't sure if the size would affect the validation, i.e. IP₃R1 is 1.3 MDa and might have an easier time in tilt-pair validation due to its size. As for Dr Penczek's suggestion on RCT, it was one of the things we tried, but it did not yield good models unfortunately. 

One interesting thing when I observed during my validation was that there was a lack of clustering when I used 'ccc' for all fields, whereas upon changing to 'dot', I managed to obtain clusters around the centre of the circle. 

I tried looking for more information regarding the advanced options in the EMAN2 website, but to no avail (http://blake.bcm.edu/emanwiki/EMAN2/Programs/e2tiltvalidate.py). I was hoping if you had suggestions on where I could read more regarding these methods.

Thank you so much once again, I really appreciate your responses. 

Cheers,
Jackie

[Paul Penczek]

I strongly disagree. There is logical fallacy in tilt pair validation and in fact it is more straightforward to do RCT, which also can be done using low tilt and few pairs. This is of course true for stain data we consider here. Doing tilt validation AFTER the structure is done is putting cart in front of a horse. Pawel

Sent from Yahoo Mail for iPhone

---

From: Steven Ludtke <slud...@gmail.com>;
To: <em...@googlegroups.com>;
Subject: Re: [EMAN2] Generating a sphere for 3D refinement

Sent: Thu, Oct 16, 2014 1:54:41 PM

[Steven Ludtke]

I doubt we'll come to agreement on this issue, but just so observers in the newsgroup can better understand the debate:

- Can you clarify what you believe is the logical fallacy in tilt-validation?

- How do you make RCT work well with small tilt angles ?  Certainly if you have a preferred orientation (the basis for the original RCT) this won't work. However, even if you have random orientations, this would give a huge missing cone for each RCT reconstruction, making reliable 3-D alignment quite time-consuming and less reliable...

[Michael Landsberg]

For those following this discussion, I was recently directed to the following paper regarding the assessment of tilt pair validation tests, and what constitutes a "good" result. 

http://www.ncbi.nlm.nih.gov/pmc/articles/PMC4136738/

Cheers

Michael

[Jesus Gerardo Galaz Montoya]

Jackie,

Just do tomography and forget about complicated validations and the need for reliable initial models :-)

If your stain and specimen heterogeneity are limiting your resolution to ~20Å or worse anyway, you can easily get that with NS cryoET followed by single particle tomography (subtomogram averaging), provided that your data collection parameters are reasonable.

2-D imaging techniques are really not the best option for heterogeneous complexes... and your protein ranges from 180 to 360 kDa (I wonder what the gradation might be?).

Both tilt validation and RCT are an attempt to extend 2-D imaging into the 3-D realm (or rather, compensate for the deficiencies).

Why not just try full 3-D, if high resolution is not the goal?

With stain, you only need ~300 to 600 subtomograms (particles) to achieve ~20-25Å (per accurately classified conformation). 

Depending on the distribution/concentration of your specimen and the number of conformations you expect to resolve, this could amount to as little as 3 tomograms.

In fact, a collaborator is about to submit a paper where we reconstructed a ~170 kDa complex at ~20Å (gold standard resolution criteria, no symmetry, no external initial models). Granted, the complex dimerizes and binds to lipid membranes, and it was stained... but still, we also derived a ~30Å resolution average for the monomer in solution, in cryo conditions (no stain), and Rossi Irobalieva (from Wah Chiu's lab) has demonstrated single particle tomography for tiny (and flexible) ~50 kDa RNAs for which (2-D) single particle analysis yielded garbage.

To unsubscribe from this group, send email to eman2+unsubscribe@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+unsubscribe@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

-- -- ---------------------------------------------------------------------------------------------- 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+unsubscribe@googlegroups.com

For more options, visit this group at http://groups.google.com/group/eman2 --- You received this message because you are subscribed to a topic in the Google Groups "EMAN2" group. To unsubscribe from this topic, visit https://groups.google.com/d/topic/eman2/_dn7tZwT8ik/unsubscribe.

To unsubscribe from this group and all its topics, send an email to eman2+unsubscribe@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

-- -- ---------------------------------------------------------------------------------------------- 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+unsubscribe@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+unsubscribe@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

-- -- ---------------------------------------------------------------------------------------------- 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+unsubscribe@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+unsubscribe@googlegroups.com.