[phenixbb] hydrogens in riding model create steric clashes after refinement
Pavel Afonine
pafonine at lbl.gov
Thu Aug 19 15:22:20 PDT 2010
Hi Yarrow,
> I was given two sets of coordinates (my model and a pdb for Iron atoms) I
> included this new model in the next round of refinement but I did not
> include the pdb for the Iron atoms.
this is strange... Why you decided to use a worse model (an incomplete
model without Iron atoms) in refinement?
> I ran refinement with the folowing parmeter file:
>
> refinement.main.ordered_solvent = true
> #refinement.main.ordered_solvent.mode=filter_only
> refinement.refine.strategy = individual_sites+individual_adp
> #refinement.refine.adp.individual.isotropic = "chain W"
> refinement.output.prefix = CIN4_refine
> refinement.output.serial = 18
> refinement.output.write_def_file = True
I presume you refine non-H and non-water atoms as isotropic and the rest
as anisotropic.
> resolution is 1.37A with close to 400,000 unique observations (obs/param
> is still about 4)
> Stats:
> Rwork Rfree bonds angles
> start: .1444 .1723 .016 1.617
> end: .1371 .1657 .010 1.301
The numbers look good given the resolution.
> I also ran wxu and wxc optimization from the command line.
Good idea.
> I know that the defualt is "riding-model" so that the hydrogen atoms are
> not refined. However, I thought that the program would be able to avoid
> steric clashes.
Using hydrogen atoms in refinement helps to reduce the amount of clashes
but does not guarantee to eliminate them completely. I would just check
all them one by one - which is very easy using PHENIX GUI: in fact,
PHENIX GUI will give you the list of clashes (run Comprehensive
Validation tool) and then you just click on each of them and this will
automatically bring it zoomed in Coot. So, it should be a quick and easy
exercise. Then, once you are done with fixing them, just re-run the
refinement.
> There are also a handfull of hydrogens that
> exist out in space that I decided to delete.
This is weird and should not happen. I would find out how you got them.
Pavel.
More information about the phenixbb
mailing list