Refinement of tilted particles?

Submitted by CSmith on Tue, 2010-11-16 08:22

Forums

Frealign

Hi

Can you tell me whether Frealign would be suitable for refining a 3D model generated from negatively stained, tilted particles? The issue I see is in limiting the angles to be searched to those relevant to the tilt process.

Thanks!

niko Tue, 2010-11-16 08:36

Yes, we have used Frealign in the past to apply CTF correction and to refine single particle parameters. You can lock any of the five parameters (3 Euler angles; x, y shifts) using CARD 3 (called MASK in the example scripts). CARD 3 is usually

1 1 1 1 1

but you can set any of the ones to zero to lock the corresponding parameter. I am not sure if this does exactly what you want but if Theta (the second parameter) indicates your RCT tilt angle, you could use

1 0 1 1 1

for CARD 3 to lock it to whatever you have used in your data collection.

CSmith Tue, 2010-11-16 08:54

That's great - thanks a lot!

On a related matter - I was wondering what the reason for removing the FLIP option in CARD 1 was in the later Frealign versions? Is there a better way to test for whether the 'upsidedown' orientation has been wrongly assigned?

niko Tue, 2010-11-16 09:33

The flip option was useful for the very first project done with Frealign (complex 1) but it did not seem to make a lot of sense in most other projects. Frealign undergoes 'clean-ups' every now and again when I review the usefulness of features. The flip option was removed in one of these, together with other parameters, such as NBACKUP and PFRAC. If there are specific orientations you need to test you could write a script that generates a second parameter file with these orientations. After running Frealign a second time with the new parameters you can compare and select those parameters that gave the best phase residual.

A word about refining angles with tilted negative stain images: The stain often produces features at the edges of the particles that are stronger than in the untilted views. If this happens, angular refinement is mostly driven by these features and not the details seen within the particle structure. It is then best to refine only the x,y coordinates and lock the angles (set the first three numbers in CARD 3 to zero).