kumar wrote:
I am trying to refine phases (using Refmac) from SAD+MR recombination (from
Phaser). Since the SAD phases are tainted with MR model bias, I am not sure
if those phases (of MR+SAD) should be used for refinement. Can any one
explain this?
If you have HL coefficients of phi/fom including both SAD and MR
information, then using those in phased refinement will tend to
constrain your model to the MR solution, because they already contain
phase information from the MR model.
However, if you have CCP4 6.1.1 you have a better alternative. Throw
away your HL coefficients, but provide refmac with the SAD data (F+/F-)
to refine against. That way you get the benefit of the SAD data without
the bias from the MR-derived phasing.
At the top line of the refmac task window, just under 'Job title', pull
down the right hand menu and select 'SAD data directly'. Then give your
F+/F-. The SAD data will help to pull your model away from the MIR
solution. The resulting FWT/PHWT maps will also help you to remove bias
if you use them in rebuilding.
Your model will need to contain the anomalous scatterers of course - if
you are using SeMet that should be no problem. Otherwise you will need a
difference Fourier to find them.
Kevin
