[phenixbb] ncs restraints weight

[Pavel Afonine]

Hi,

yes, this is correct: phenix.refine offers NCS restraints only. Here are 
two parameters you may want to play with:

  ncs
  {
    coordinate_sigma=None
    b_factor_weight=None
  }

This will set up specified values for all NCS groups. Try an array of 
values to see which value works best (monitoring Rfree and overall 
stereochemistry). The defaults are 0.05 and 10, for coordinates and ADP 
correspondingly. Making coordinate_sigma smaller makes the NCS 
restraints tighter.

Alternatively, you can set set different values for each NCS group like 
in example below:

  ncs
  {
    restraint_group {
      reference=XXX
      selection=YYY
      coordinate_sigma=0.05
      b_factor_weight=10
    }
    restraint_group {
      reference=AAA
      selection=BBB
      coordinate_sigma=0.03
      b_factor_weight=5
    }
    restraint_group {
      reference=ZZZ
      selection=QQQ
      coordinate_sigma=0.01
      b_factor_weight=2
    }
  }

Be careful to not deviate too much from the default 
coordinate_sigma=0.05 since it may results in overall stereochemistry 
distortion. Anyway, like I said above, try an array of values and see 
which works best (judging by Rfree and overall geometry).

Please let me know if you have more questions.

Pavel.


Daniele de Sanctis wrote:
> Dear all,
>
> i'm refining a low resolution structure (3.2A) with four molecules in
> the asymmetric unit. I saw that phenix.refine doesn't allow
> constraints, so i'd like to change the restraints weight but i can't
> find the right keyword...
>
> thanks for your help
>
> d
>
>