Frealign helical refinement

Submitted by Duncan Sousa on Fri, 2011-01-07 14:28

Forums

Frealign

I had a couple Frealign questions that I'm hoping someone can help me with:

1. How does "set_polarity.exe" calculate the 'score' after adjusting a parameter file and how does it choose which polarity to select initially?

2. How does Frealign use the 'stiffness' setting and how precisely is this value used to alter the psi angle refinement. Why isn't a persistence length or max angle change used? A few of my filaments vary up to 20 degrees over their full length, but Frealign appears to take that variation and flatten it completely (for instance, using a 5 stiffness, one filament starts with a trend from ~145 to ~170 degrees, but after 5 rounds of refinement that trend is flattened to random variation from 155 to 160).

niko Fri, 2011-01-07 19:30

set_polarity calculates the score for a filament as

        # of segments within a given tolerance from average PSI
score = -------------------------------------------------------
                   # of total segment in a filament

(the tolerance is the half of the value entered for PSI angle variation).

The stiffness setting determines how strongly the PSI and THETA angles are forced to agree with the averages for these angles calculated for a filament. For both PSI and THETA a term is added to the correlation coefficient calculated between segment and model projection. The term is calculated as

SIG2*(-(STIF*D)**2/2.0/SA**2)

where SIG2 is the variance of the noise present in the images, STIF is the stiffness parameter, D is the difference between PSI (or THETA) of a particular segment and the average for the filament the segment is from, and SA**2 is the variance of the angular deviations of PSI (or THETA) from the averages calculated for each filament, averaged over the entire a data set. The terms for PSI and THETA represent restraints. If the stiffness is set to 0, the restraints are removed. Larger values for stiffness increase the restraints. For a more detailed description of the formula, please consult the appendix of our publication on the rotavirus VP7 recoated particle [1].

References


Duncan Sousa Fri, 2011-01-07 21:04

This clears up a lot of confusion on my part! I'll have to think about how the refinement behaves with the stiffness parameter when there is a known angular change trend.