I'm not quite sure what is "wrong" here except perhaps for the idea
of "manual correction".  If you calculate the rmsd for thousands of
angles and you change four of them it is unlikely there will be a large
change in its value.  The reason programs disclose both the overall rmsd
and the individual outliers is that the two measures of quality are
mostly unrelated.

   Different programs have differing libraries of "ideal" values for
bond lengths and angles and will report slightly different rmsd values.
Why do you consider the different values reported by Buster and
MolProbity to be a problem?

   What does concern me is the idea of "manual correction" of an
outlier.  The model you deposit should have been produced by a
refinement program, without arbitrary changes made by the user.  If the
refinement program produces a model with outliers you can look at those
instances and identify problems with your model.  Maybe you can change
the rotomer of a side chain or perhaps there is a mistake in your ideal
geometry cif.  If you find a good reason why the computer is creating
that outlier and correct the problem, you can run more refinement and
get a model with fewer outliers.

   If you simply change the model in Coot to make the outlier "go away"
and declare that, edited, model to be your final model you are not doing
it right.  In most cases you will find that the refinement program would
take that model and recreate the outlier, so you haven't accomplished
anything other than hiding a shortcoming of your model.

   If you can't figure out how to get the program to produce a model
without that outlier then your model should be deposited with it, and
the ultimate users of your model can see that this region, at least, of
your model should be considered less reliable.  Tricking people into
placing too much trust in your model is not a good idea.

Dale Tronrud.


On 7/5/2018 1:11 AM, zheng zhou wrote:
> Hi all
> 
> Just finishing up a new structure at 2.4A. Buster refine gives RMS bond
> 0.008 and angle 1.13, while MolProbity gives 0.01 and 1.83 degree. I
> checked the 4 outliers from molprobity……>4sigma. After manual
> correction, warning goes off, but RMS angle only goes down to 1.82
> 
> I am using Phenix 1.13 and buster2.11.6
> 
> Could not figure out what went wrong.
> 
> Sorry for not CCP4 related questions.
> 
> Thanks 
> Joe
> 
> 
> 
> ------------------------------------------------------------------------
> 
> To unsubscribe from the CCP4BB list, click the following link:
> https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=CCP4BB&A=1
> 

########################################################################

To unsubscribe from the CCP4BB list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=CCP4BB&A=1