Auto expansion related issue

[Bindusmita Paul]

Hi, 

I have been trying to use TomoNet for S-layer particle picking. I have few queries.

In the paper, it is mentioned to generate reference using PEET. But inside the TomoNet I couldn't find any function for the generation of reference using PEET. 

My query is do I need to generate a reference outside TomoNet in PEET?

Also, I tried auto expansion with 50 particles without any reference, at the end it showed "Particle number 50. After clean 0! with 0 patches".

Can you please suggest if I am doing something wrong? or how to deal with this?

Thank you

Bindusmita

[HUI WANG]

Hi  Bindusmita,

Thanks for trying TomoNet. It seems you have 50 particles, but they were removed during the expansion. I might need to know more details to help with troubleshooting (like your S-layer configuration, your searching params etc).

Yes, TomoNet do not provide a direct way to generate a reference. You can generate it using conventional PEET method or anything else.

However, you can use "Auto Expand" by setting the round # to 1, that is equals to just trying to align and average your input seed particles using PEET.

I can give you a example on how I manage to get the S-layer particles without a reference in the beginning using TomoNet.

It takes two main Parts:

First part, the main goal is to generate a rough reference

1. I manually picked ~30-50 particles like this, with the stick represent PEET Y-axis at the S-layer six-folder axis.

2. The set the search param like this. some key elements here. 1. 180 global search on Y-axis, 30 degree for X or Z planar search. 2. low CCC to avoid particle removal 3. set refine reference yes, and a reference will be generated under the cache folder.

3. After run the Auto Exp for 1 round, I can get a initial reference like the above (maybe you can reorient and recenter for later expansion runs), and this should be the end of First Part.

You can find the reference map like this XXX_cache/round_0/unMaskedXXX_Ref#.mrc if it works.

Second Part: Auto Expand with a initial reference

1. Generate a new search params file to the second part. Key elements here: 1. local search to save more times 2. determine your transition list for expansion 3. relatively large CCC to exclude bad particles (can tune this based on specific cases) 4. you can set refine reference to Yes if you want end with a better initial reference, if you think your initial reference is good enough set No.

2. Run several rounds using this setting, then if your particles looks like this, meaning "Auto Expand" works.

Typically, I will do the second part again (from beginning) with a finalized initial reference.

If you have a reference to start with, then you can skip the first part as well.

Hope this will help you! Let me know if you have any further questions.

Best,

Hui

[Bindusmita Paul]

Hi HUI,

I have tried with your suggested parameters and it is working now.

Regarding translation list, how do we determine the translation list? I have used your translation list parameters and it worked. But is there a way to determine it?

Thank you

Bindusmita

ᐧ

--
You received this message because you are subscribed to a topic in the Google Groups "TomoNet" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/tomonet_cryoet/4ukN-M3W1PM/unsubscribe.
To unsubscribe from this group and all its topics, send an email to tomonet_cryoe...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/tomonet_cryoet/99d7af35-e463-4a0c-a062-a3a06dc68aa5n%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[HUI WANG]

Glad it works!

The [vectors] in translation list are calculated from your initial reference. Basically, for each user customized expand direction, i.e., a center to center vector, can be calculated by (x_n, y_n, z_n) - (x_0, y_0, z_0), where (x_n, y_n, z_n)   is one of your Target_center coords, and (x_0, y_0, z_0)  is the reference map center (i,e, your current unit center). In the case of a hexagonal lattice, the # of vectors should be 6

Best,

Hui