I remember that I asked a question before which probably is related to this one.  The ideal bond distances between the riding hydrogen and its attached atom are different in phenix.reduce and mono library.  That is why I got a huge RMSD bond value after phenix.reduce. It has been fixed after my question.  phenix.refine now put all bond distance back to what are defined in mono library.  However, a new problem rises from this fix.  When using MolProbity to get a new set of riding hydrogen atoms, it will give a pretty bad clash score.  If I would guess, the bond distance between the riding hydrogen and its attached atom is longer in phenix.reduce or MolProbity than the value in mono library.

I believe this is a general problem and a "standard" bond distance should be used in different programs, especially those programs in one package.  

-- Jianghai

PS.  Pavel, I think you can test this one using any pdb file.  If you still need my file, I will be happy to send it to you.



On Aug 6, 2008, at 3:03 PM, Pavel Afonine wrote:

Hi Jianghai,

thanks for finding this: you find exotic problems, as always! Could you please send us the files (after phenix.refine and after phenix.reduce) so we can have a close look at this particular problem ? I can't promise to look at it immediately, but it definitely will go into the list. May be developers at Duke volunteer looking at this problem.

Thanks!
Pavel.


On 8/6/2008 8:17 AM, Jianghai Zhu wrote: 
Yes, I think the phenix group should figure that out :)

-- Jianghai






On Aug 6, 2008, at 5:12 AM, Pavel Afonine wrote:

  
Hi Jianghai,

phenix.refine idealizes the geometry of hydrogen atoms using the  
Monomer
Library definitions for ideal values (if riding model is used). I
presume that phenix.reduce uses some other values when adding H atoms.
So it is not too surprising that the geometries of H atoms are  
different
after phenix.reduce and phenix.refine.

What is interesting though is why this has a significant impact on the
clash scores.

Pavel.


On 8/5/2008 1:38 PM, Jianghai Zhu wrote:
    
Hi,

I used phenix.reduce to add riding hydrogens.  After phenix.refine,
the structure got a good score from MolProbity.  However, if I
stripped the riding hydrogens from the structure and let MolProbity  
to
add hydrogens before the analysis, I got a much worse score,
especially the clash score.  Does phenix.refine refine the hydrogen
positions?

-- Jianghai






_______________________________________________
phenixbb mailing list
[email protected]
http://www.phenix-online.org/mailman/listinfo/phenixbb

      
_______________________________________________
phenixbb mailing list
[email protected]
http://www.phenix-online.org/mailman/listinfo/phenixbb

    
_______________________________________________
phenixbb mailing list
[email protected]
http://www.phenix-online.org/mailman/listinfo/phenixbb
_______________________________________________
phenixbb mailing list
[email protected]
http://www.phenix-online.org/mailman/listinfo/phenixbb