rmbeadthr not behaving as expected

[Colin Gauvin]

I'm having some issues reconstructing a tomogram and removing fiducials from it.I'm using the --rmbeadthr option. My understanding from the tooltip is that it's density value threshold, and I'm assuming that density with a value above that is removed? But presumably there's also some sort of spatial component that's considered?

Anyways, in my tilt series, the values range from 0-255, and in one random tilt, for instance, the 32 pixel area average of a gold bead is 60. If I reconstruct the tomogram with any --rmbeadthr value however, the beads don't remove. Not at 10, not at 5, and not at 1.

Am I misinterpreting the --rmbeadthr option? What would explain this behavior?

Best,

Colin Gauvin

Cryo-EM Assistant Facility Manager

Montana State University

[Steve Ludtke]

Hi Colin,

Are you using a current snapshot version of EMAN2 or an earlier release?  (e2version.py full output would be useful)

After your tilt series are pre-processed before reconstruction the range will no longer be 0-1, they will be "normalized" (mean or edgemean 0, std 1). With current versions of EMAN2, this normalization doesn't impact file size, an 8 bit representation is still used for storage, but in HDF format the images are appropriately rescaled. (https://authors.elsevier.com/a/1fJ0I3kGCFFw16)

If you are trying to reconstruct tomograms without preprocessing (import tilt series) you are likely to have a lot of issues...

-----------------------------------------
Steven Ludtke, slud...@gmail.com 

--
--
----------------------------------------------------------------------------------------------
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/f0296353-36d7-4e06-81ce-9bf6a7c5db5en%40googlegroups.com.

[Muyuan Chen]

—rmbeadthr uses the threshold of the tomogram instead of the tilt series. So if you pull out a temporary tomo_xx.hdf file, and find a threshold that only displays gold fiducial, make it positive and use that number. 

Saying that, I have not used this option for a while, largely because I somehow only get fiducial less tilt series lately. So if this does not fix your issue, let me know and I can take a deeper look. 

Muyuan

[Colin Gauvin]

I think I didn't hit reply all last time, so my last exchange isn't showing above.

But basically, if I go to the tomorecon_00 folder and open tomo_03.hdf in the 3D viewer, and lower the threshold until just the fiducials are visible, the threshold is -20. If I set --rmbeadthr=-20, nothing happens, fiducials are still present in the reconstructed tomogram.

I'm guessing this has to do either normalization or scaling? My tilt-series directly out of SerialEM opened in the 2D viewer shows values from -40 to +470. The "area average" of a fiducial is ~60. However, the area average of blank (ice) background is ~90. A portion of my virus is ~85. So fiducials have a lower value than sample which has a lower value than background. This seems to make sense to me - assuming the integer numbers represent counts, and counts don't happen where there's fiducials or other things scattering the electrons. Once I import that tilt-series into EMAN2, the fiducial is 55, the sample is ~85, and the background is ~90. This still makes sense, nothing seems to have happened to the data at this point. The tomogram and aligned tilt-series however set the density for most of the values to negative values, including the fiducials, and a good chunk of the sample value. The background is still ~ -1 to +1.

[Muyuan Chen]

As I mentioned last time, use the absolute value , I.e. 20 instead of -20 for that option.

On Jul 18, 2022, at 9:24 AM, Colin Gauvin <ccgau...@gmail.com> wrote:



To view this discussion on the web visit https://groups.google.com/d/msgid/eman2/2d7e30ba-cb82-41e9-81a7-3b9c3226b64dn%40googlegroups.com.

[Colin Gauvin]

Ah, so it was a reading comprehension issue on my end. Sorry about that - I do see now you said that. Setting it to higher is mitigating the gold beads now. I also understand now why you want to increase the number to lower the threshold to remove the beads, since it's actually a negative number.

Thanks for your help.

[Colin Gauvin]

Hmm...as an aside, in case anyone searches and arrives on this thread, I'm getting dramatically better bead removal with patch tracking enabled than with it disabled.

[Muyuan Chen]

I noticed this lately too. I am guessing you are using relatively large beads and high mag (like 10nm gold and 1 apix), so that the default landmark based alignment does not perform as well. Maybe setting a larger --bxsz helps too. Anyway, --patchtrack should be quite stable at the latest version and enabling it shouldn't hurt. 

To view this discussion on the web visit https://groups.google.com/d/msgid/eman2/5018f789-3797-43d7-896b-bb84c2f7781en%40googlegroups.com.

[Colin Gauvin]

Yes - I'm using 10 nm gold and 1Å pixel size. I did try increasing the box size from 32 to 128 and my loss went from 0.8 nm to 0.5 nm, and all but a single gold bead was removed. It did take ~4x as long unfortunately, but I understand that comes with the territory. I will try 64 next. I think this may have been my problem from the beginning, actually.

[Ludtke, Steven J.]

The next step is to exclude the beads earlier in the process (ie - in the experiment) :^)

We haven't used fiducials for about 5 years now

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

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 Technical Cores                   and Molecular Biology

Academic Director, CryoEM Core

Co-Director CIBR Center

To view this discussion on the web visit https://groups.google.com/d/msgid/eman2/acc15aa1-635d-40da-89d6-cc005e72d407n%40googlegroups.com.