Project

General

Profile

Making an oriented substack

Added by Chad Park over 12 years ago

Sorry for the naive and possibly off-topic question. I'm still new to image processing.

I have a (hopefully) quick question about Appion. I have some nice alignments of an oligomer that seems to make a helix along a straight direction. However, the alignments are all at some arbitrary angle relative to the particle box directions. I'd like to have a substack of the aligned particles where they are aligned vertically so I can make the box size smaller. Is there an easy way to do that? I've clicked around a bunch of places - under the results of alignments and clustering runs, but I don't see any way to make that vertically-oriented aligned substack. Thanks for any and all advice!


Replies (1)

RE: Making an oriented substack - Added by Arne Moeller over 12 years ago

Hi Chad,

First of all it is not ideal to use a Stack for refinement that has already been "pre-rotated". This will influence the performance of the alignment.

Still Appion can be used to do that.

Depending on your version some may currently be not available for you.

There is a nice tool to mask helical filaments - without rotating them (strongly preferred!). The database basically knows the rotational angles that were determined during your alignment and rotates the mask respectively. Under the averaged stack image should be a function called Mask Particles With Box - this will do the job.
Another way is to use the helical picker. By selecting the ends of a filament the orientation is defined. During stack creation this information will be used to do the "pre-rotation".

If neither of these options is available to you there is one that should definitely work:
make a template of a particle, preferably a class average, that is rotated in the way you want it and align all particles against this using a reference based alignment. If your stack is smaller than 100k particles I would recommend to use Spider for this but the other options should also work.
After that classify the 2D-alignment (Spider, Kerden) - one of the resulting classes should have a much clearer signal than the initial reference (especially if you used a single image as template) use that to go through the process a second time (maybe a third).

If you feel comfortable to work outside of Appion you can also use a function like proc2d (Eman) or rotate-image (Imagic) (there are functions like this available in any processing packages) to rotate the best classaverage that you already have and upload the rotated result as template. Of course you can do this with more than one average to achieve a better alignment.

Finally there is one more thing you will have to do: You have to upload the aligned stack as new stack using the stack upload function.

Let me know if you have any further questions - or if that does not work for you...

Cheers,
Arne

    (1-1/1)