Filament question

62 views
Skip to first unread message

Bindusmita Paul

unread,
Jul 5, 2024, 9:16:56 AM7/5/24
to EMAN2
Hi Eman2 team,

I am using Eman2 to do subtomogram averaging of filaments (I have followed the Eman2 tutorial). However, my density maps appear to have wedges on one or both sides (see images attached below).

Would you be able to give me some pointers on how to determine if these "wedges" are artifacts or if this feature is present in my structure? I have done classification and masking; however, these features seem to be present in all of my averages.

Any help would be greatly appreciated.


EMAN 2.99.52 ( GITHUB: 2023-07-21 12:36 - commit: f413f05eb )

Screenshot from 2024-07-05 20-45-20.pngScreenshot from 2024-07-05 20-45-32.png Screenshot from 2024-07-05 20-45-52.png

Thank you
Bindusmita Paul
Phd student
University of Melbourne

Ludtke, Steven J.

unread,
Jul 5, 2024, 10:06:23 AM7/5/24
to em...@googlegroups.com
Symmetry is a tricky issue in single particle analysis. Say you have an object ostensibly with a C4 symmetry. If you do a reconstruction of this object without imposing any symmetry, then the alignment routine is free to assign each particle to any one of 4 different orientations. Which will it choose? In noisy data there is no such thing as true symmetry. Even if the structure were perfectly symmetric, the noise breaks the symmetry, and you will generally wind up with a structure where one side of the structure has higher density than the other side, as the algorithm iteratively aligns whatever it can find which breaks the symmetry. Imposing the symmetry eliminates that issue, but of course, also conceals any real symmetry breaking which might occur. 

Maximum likelihood methods can also partially solve the problem by putting each particle in multiple orientations without explicitly imposing symmetry. However, this is a double-edged sword. The assumption is that your probability model is sufficiently perfect that it can distinguish noise-based differences from true structure differences unambiguously. That is, if it slightly under-estimates the low resolution noise, the symmetry will be broken and if it slightly over-estimates the noise then real symmetry breaking features can be concealed. 

Of course there is no such thing as a perfectly symmetrical biomolecule. At SOME resolution the symmetry will be broken, if nothing else by dangling side-chains or thermal noise. Very often, however, there are much lower resolution differences due to the function of the object. There are various approaches used to tackle this issue, the most common of which is to perform a symmetry-imposed reconstruction, then extract the individual subunits from the original data and perform your dynamics/heterogeneity analysis on the individual subunits rather than the whole structure. Then you can try and fit the dynamics information back into a self-consistent model of what's going on in the whole assembly.

Now, this takes us to subtomogram averaging of helical filaments. In most helical filaments following a line along the top edge of the object in the image will result in a gradual rotation of orientation. From your structure it looks like your helix is almost completely straight from this perspective. I suspect if you look at the Fourier transform of the Z projection of your filament (open the control panel and select "Amp" in a 2-D view of the Z-projection, then set full contrast), you will find that the missing wedge has been aligned to basically the same orientation in all of the particles, causing your apparent symmetry breaking. While there is a possibility that there might be some real flattening of a helical object in ice, that isn't what this looks like to me.

The obvious solution would be to determine the Nstart component of the helical symmetry (equivalent to Cn symmetry) and at least impose that (cn)on your refinement. The remaining aspects of the helical symmetry would still emerge naturally that way.



---
Steven Ludtke, Ph.D. <slu...@bcm.edu>                      Baylor College of Medicine
Charles C. Bell Jr., Professor of Structural Biology        Dept. of Biochemistry 
Deputy Director, Advanced Technology Cores                  and Molecular Pharmacology
Academic Director, CryoEM Core
Co-Director CIBR Center


On Jul 5, 2024, at 6:15 AM, Bindusmita Paul <paulbin...@gmail.com> wrote:

Hi Eman2 team,

I am using Eman2 to do subtomogram averaging of filaments (I have followed the Eman2 tutorial). However, my density maps appear to have wedges on one or both sides (see images attached below).

Would you be able to give me some pointers on how to determine if these "wedges" are artifacts or if this feature is present in my structure? I have done classification and masking; however, these features seem to be present in all of my averages.

Any help would be greatly appreciated.


EMAN 2.99.52 ( GITHUB: 2023-07-21 12:36 - commit: f413f05eb )

<Screenshot from 2024-07-05 20-45-20.png><Screenshot from 2024-07-05 20-45-32.png> <Screenshot from 2024-07-05 20-45-52.png>

Thank you
Bindusmita Paul
Phd student
University of Melbourne


--
--
----------------------------------------------------------------------------------------------
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.
To view this discussion on the web visit https://groups.google.com/d/msgid/eman2/ec8a7014-0dd2-411f-a44a-faa367eb9435n%40googlegroups.com.
<Screenshot from 2024-07-05 20-45-20.png><Screenshot from 2024-07-05 20-45-32.png><Screenshot from 2024-07-05 20-45-52.png>

Bindusmita Paul

unread,
Jul 6, 2024, 5:13:48 AM7/6/24
to EMAN2
Thank you Steve for the suggestions! 

Also, one more question, is there a way to determine helical pitch and turn using Eman2? If not, could you suggest any alternatives?

Ludtke, Steven J.

unread,
Jul 6, 2024, 12:09:43 PM7/6/24
to em...@googlegroups.com
Helical objects have never been a really major focus in EMAN2, but we do have basic support for helical symmetry. Off the top of my head I can't think of any scripts we distribute which will automatically search for symmetry parameters, but it would be a pretty trivial sort of script to write. Basically you'd want a nested loop covering the two helical variables (http://eman2.org/Symmetry), and impose the "test" symmetry on the map, then compare the similarity between the symmetrized and unsymmetrized maps for each putative symmetry. There should be a peak...

---
Steven Ludtke, Ph.D. <slu...@bcm.edu>                      Baylor College of Medicine
Charles C. Bell Jr., Professor of Structural Biology        Dept. of Biochemistry 
Deputy Director, Advanced Technology Cores                  and Molecular Pharmacology
Academic Director, CryoEM Core
Co-Director CIBR Center

Bindusmita Paul

unread,
Jul 9, 2024, 12:33:11 AM7/9/24
to EMAN2
Thank you for the suggestions!
Reply all
Reply to author
Forward
0 new messages