Using patch CTF estimation to filter particles by local ice thickness

CTF Estimation
1

Hello,

In several of my projects involving very small particles, I need extremely thin ice, so filtering micrographs by ice thickness is very helpful during particle picking.

However, when a micrograph contains an ice meniscus, the ice thickness varies significantly across the field of view. In those cases, filtering based on a single global ice thickness value does not work well.

Would it be possible to estimate ice thickness locally (e.g., using the patch-based CTF estimation) and restrict particle picking only to regions where the ice thickness falls within a desired range?

Thanks!

Jorge

2

Hi @jplopez

Unfortunately, as Patch CTF Estimation is implemented currently, there is no way to derive a measure of local ice thickness. Are you currently use some mechanism to select the particles in the thinner ice (maybe IceBreaker or a similar software)? If so, can you elaborate on how much of a difference you see between the thicker ice regions and the thinner ice regions. We would be interested in seeing map quality, resolution, and cFAR scores for the thin and thick ice reconstructions.

Additionally, do see a bimodal distribution in per-particle scale values that corresponds to particles in thick and thin ice?

All the best,
Kye

1 Like
3

Hi Kye,

Thanks for the reply.

We are currently working with a small membrane protein that has very little mass outside the membrane region, so alignment is quite challenging, especially when the ice is thicker. When the ice is even moderately thick, we struggle to obtain good ab initio reconstructions.

In practice we usually do not observe a clear bimodal distribution in the per-particle scale values, probably because most particles located in thicker ice are already removed during earlier cleaning steps (micrograph curation and multiple rounds of 2D classification and heterogeneous refinement). However, these aggressive cleaning steps also tend to remove a significant number of good particles.

For that reason, we think that some kind of tool that could identify and remove particles located in locally thicker ice regions earlier in the workflow might help improve the overall particle set. We have not tried IceBreaker yet, but we will definitely test it — thanks a lot for the suggestion.

Best,
Jorge

2 Likes
4

IceBreaker looks very useful - I hadn’t seen that but have wanted similar in the past. I wonder if it would be possible to hook into CS via CS tools..? It is already possible to use as an external job in Relion, so maybe..

1 Like
5

Hi all,

Thanks for your feedback and additional information @jplopez, I have noted your request!

@olibclarke, yes, you can use cs-tools in conjunction with IceBreaker in the command line with relatively little friction. The only thing outside of those two programs that needs to be done is the particles need to be converted to .star format using pyem.

Best,

Kye

1 Like