next up previous contents index
Next: RSR_map Up: The RSR commands Previous: The RSR commands

RSR_setup

  This is a menu that enables the user to change some parameters and options relating to the real space refinement. There are 2 possible choices for the refinement method, the CONV method, which tries to maximize the convolution product between observed and model density, and the DIFF method, which tries to minimize the difference density. Although the convolution product method biases the fit to place single atoms on grid points, it is less sensitive to errors produced by scaling the model to the observed density. It can therefore be used when the fit to the observed density is poor. The effect of grid biasing is most severe when fitting single atoms and is reduced by fitting residue fragments.
Autoscaling
This is an option to turn on automatic scaling of model to observed density. It is effective in most of the minor menus in RSR. It may sometimes be advantageous to switch off the autoscale option and use the manual scaling in RSR_scale instead.

Autosubtraction
The user is permitted to switch off the subtraction of model density from observed. This is not recommended, but may be used with the CONV refinement method when the scaling of densities is inaccurate.

Nscan
Set the number of search scans in the refinement procedure. Typical value is 5.

Stepsize
Set the size of the Nscan steps in Ångstrøms/degrees. Typical values are 1.0, 0.5, 0.25, 0.2, 0.05.

Form_factors
Defines the values A(o) and C, used to transform a B factor to an atomic radius A, using the relationship from [2]:
B = 4pi * ((C*A)^2 - A(o)^2)
Typical values are A(o) = 0.90; C = 1.04 for high resolution maps (1.7--1.9Å) and C = 0.82 for medium resolution (2.4Å). See the discussion on page gif.

Irad
Specify how many grid points over which the grid sums are formed. Should be of magnitude 3 Ångstrøms.

These parameters are also used by the RS_fit command.



next up previous contents index
Next: RSR_map Up: The RSR commands Previous: The RSR commands