POWELL MIN： MAXIMUM CHANGE DOES NOT ALTER FUNCTION

Submitted by Lingyun on Thu, 2014-07-03 23:38

Forums

Frealign

Hi，Niko, I am new in using frealign,so I have confused by its output. When running frealign, it appeared that "Best score for particle 1 at Rmin/Rmax 200.0 10.0:*** POWELL MIN： MAXIMUM CHANGE DOES NOT ALTER FUNCTION ". I am wondering what does this mean about "POWELL MIN： MAXIMUM CHANGE DOES NOT ALTER FUNCTION"? Why for some particles it appears several times some never? Thanks!

Lingyun

Alexis Thu, 2014-07-03 23:45

Hi Lingyun,

"Powell min" refers to the Powell minimizer which is used by Frealign to optimize the shift and orientation of each particle relative to the 3D model. The warning you are getting means that for these particles, the scoring function (correlation between model and particle image) always gave the same value, no matter what shift parameters or Euler angles were tried by the minimizer.

When this happens, it can be a sign that there is something "funny" about these particles. Make a note of which particles give this warning and have a look at them using a graphical display program - perhaps you will notice something is wrong with them.

Hope this helps,
Alexis

Lingyun Fri, 2014-07-04 09:06

Yes, thank you. I have heard from someone else that when doing helical reconstruction, frealign could not do global search, ie MODE=4, is that right? "POWELL MIN： MAXIMUM CHANGE DOES NOT ALTER FUNCTION" appeared when I was doing global search for helical particles. So that means for helical particles I should give the correct parameters and then did local search and refinement, even if in frealign_v9.08? But I saw the script mult_refine, when only film, defocus1, defocus2 and astig angle were given, this script could generate a full parameter file, then should we run mode 4? Thanks!

Best,
Lingyun

niko Sat, 2014-07-05 17:12

Frealign should be able to do a Mode 4 search with helical filaments. However, keep in mind that the THETA angle for most filaments will be close to 90 or 270 degrees, and the PSI angles for segments from the same filament should be close to each other. Frealign is not really optimized for helical reconstruction, and doing a global search is done for each segment without making sure they end up with a uniform polarity. You should consider using Frealix, our software that is designed for helical filaments.