[phenixbb] geometry weight and rmsd bonds - angles

Pavel Afonine pafonine at lbl.gov
Thu Aug 12 08:01:53 PDT 2010


  Hi Lionel,

> I'm using phenix.refine for the first time and have some problem to 
> get it reach reasonable rmsd bonds (0.012-0.015) and angles (1.5-1.8).

interesting... How you know this is reasonable? It may be reasonable if 
the resolution of your data is say around 2A or so (and the data is 
relatively complete). It is too loose if the resolution of your data is 
~3-4A and lower (at this resolution I would expect them approaching 
zero). It might be higher for higher resolution data.

> I set the bonds rmsd_max and angles rmsd_max to these values and try 
> different values for:
> wxc_scale (from default 0.5 downto 0.015),  wxu_scale (from default 1 
> down to 0.03), wc (from default 1 up to 8), with and without 
> optimising wxc wxu.
> Whatever the parameters, phenix never reached final rmsd better than 
> 0.022 and 2.2.
> I refine the structure with refmac so far, but need to refine the 
> occupancies (so phenix).
> Am I doing something wrong? Should I just use phenix to refine only 
> the occupancy (I am actually doing it)?
> Any help or comment are welcome.

Run phenix.refine using "optimize_wxc=true" option, which will optimize 
the target weight by choosing the one that gives the lowest Rfree.

> Some information:
> space group: P21, 2 molecules (180 aa) / au, 510 waters, about 20% 
> alternate conformations,
> ligand covalently link to the protein (correct dictionary and link 
> parameters for refmac or phenix -add link to cif),
> resolution: 1.32, refine xyz for all, Baniso for all (refmac), ncs not 
> used, R 13.6 Rfree 17.7 fom 0.913 rmsd 0.012 1.56 (last refmac)

At 1.32A I would expect phenix.refine resulting in something like 0.022 
and 2.2.

I presume you use a proper refinement strategy and model parametrization 
given this resolution:

- use H atoms;
- keep updating solvent (ordered_solvent=true);
- refine all non-solvent and non-H atoms as anisotropic;
- ... etc.

Pavel.





More information about the phenixbb mailing list