Signal subtraction to improve resolution of flexible region

[Reika Watanabe]

Hello!

I would like to test whether subtracting the signal of core regions from a large complex might improve the resolution of peripheral regions for in situ subtomogram averaging.  

Has anyone tried? If so, could you tell me how you proceed?

Currently, I am using Warptools for preprocessing/2d particle extraction instead of Relion5.

Is it possible to perform signal subtraction in this current workflow? Or should I start everything using Relion5?

 Thank you very much for your kind attention in advance!

With best wishes,

Reika 

[teg...@gmail.com]

Hi Reika,

The only time I tried subtraction on tomo data was ca. 10 years ago. It didn't help. However, a lot of bugs have been fixed since then. In theory, I'd expect less benefit from subtraction compared to SPA data because refinement of tilt series data is less sensitive to structured noise (at low resolution, at least).

Both multi-body refinement (which is built around signal subtraction) and the stand-alone signal subtraction tool in Relion appear to only support 3D subtomogram particles, not the new particle tilt series format that's standard in Relion 5. So you'd have to export your particles as subtomograms. You can do that either in Warp or Relion.

One thing to keep in mind is that signal subtraction works best when the signal scale in each particle is known exactly. Signal scale determination is well-implemented in Relion for SPA particles, but more of an afterthought for tomo.

Cheers,

Dimitry

[Reika Watanabe]

Hi Dimitry, hi everyone,

I sincerely appreciate your insightful comments.

Concerning 2d particle versus 3d subtomogram, it is mentioned here. 

https://github.com/3dem/relion/issues/1295

My current (theoretical only) understanding of workflow is

1. After Relion5 3d refiment run

2. Signal subtraction tool in Relion5

3. Using new particle files, perform 3d subtomogram export using Warp tool.

4. 3d refine using signal subtracted data set.

Is my understanding correct?

Thank you very much for your kind attention!

Reika