Converting from Sparx to Frealign

Submitted by rydog on Tue, 2013-07-09 16:43

Forums

Frealign

I was trying to run some Sparx functionalities on a stack of particles that were refined in FREALIGN, and I want to convert the FREALIGN angles and translations over to Sparx.

I used the "frealign_to_sparx.py" script on the website and I'm pretty sure something's going wrong. I can see the alignment metadata as values for the xform.projection tag, but when I do a back-projection to get a volume in Sparx (sxrecons3d_n.py), it looks like pure noise (especially when compared with the FREALIGN volume.)

Do you think I'm going about this the wrong way? How would you test to see that the frealign_to_sparx.py script was working correctly?

Thanks.

niko Tue, 2013-07-09 16:46

It is difficult to say what goes wrong. Did you set the values in the script (pixel size, voltage etc) correctly?

rydog Tue, 2013-07-09 16:56

I set those values correctly, but they're mostly going unused because the CTF was applied to the stack. Only the apix values could influence the results.

Alexis Tue, 2013-07-09 16:55

My favourite way of debugging conversion issue is to start with a single image (best is a test image you generate which is asymmetric) and do the reconstruction in software 1, then in software 2 after conversion. Load the results in your favourite visualistion tool (mine is IMOD, but Chimera works too) to check the images were backprojected identically.
At the very beginning, zero all parameters except one of the Euler angles. Then re-do the test with one more the Euler angles added in. Keep going until something messes up.

Also, do things without any symmetry options at first. Some packages do not have equivalent symmetry definitions (for example, you may need additional 90 degree rotations etc).

Hope this helps
Alexis