[phenixbb] histidine flip in refinement

Dale Tronrud detBB at daletronrud.com
Wed Jul 22 11:01:51 PDT 2015


-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi,

   I'm afraid I have to disagree with Pavel's statement.  For a model
refined to X-ray data in the 2-3 A resolution range, with
stereochemical restraints, you can deduce quality information about
the distances between atoms.  The bond length and angle restraints do
a very good job at resolving the ambiguities in fitting the blobby
density.  This has been routinely done for decades.

   A proper error analysis and estimates of the standard deviation of
your measurement is a tricky proposition, I agree, but if the atoms
are at full occupancy and the density is clear my "gut-level" estimate
would put the sd around 0.2 to 0.3 A.

   If you want something more quantitative, I suggest you measure the
NH--O hydrogen bond distances in some stretches of alpha-helix and
compare your model's values to those of true atomic resolution models
of alpha-helices.  The scatter in values in your model should give a
reasonable estimate of the ability of your model to "predict" other
hydrogen bond distances.

Dale Tronrud



On 7/21/2015 9:26 PM, Pavel Afonine wrote:
> Hi Smith,
> 
> 2-3 A is not atomic resolution, so you cannot meaningfully measure 
> distances between individual atoms in your model at this
> resolution, I think (I guess it is safer to say that you can
> measure these distances, technically, but their meaning is not
> going to be straightforward to interpret).
> 
> Pavel
> 
> On 7/21/15 20:10, Smith Liu wrote:
>> Dear Pavel,
>> 
>> Related to Joel's question, suppose the resolution is about 2-3
>> A, and I have added H (should be modelled "H")for the refinement.
>> If I want to measure the H-bond length between the NE2 and H of
>> OH of Tyr, I need to measure the distance between NE2 and the
>> "modelled H" of OH of Tyr. Is this "modelled H" position reliable
>> for the length measurement of the H-bond?
>> 
>> Best regards.
>> 
>> Smith
>> 
>> 
>> 
>> 
>> 
>> At 2015-07-22 10:04:39, "Pavel Afonine" <pafonine at lbl.gov>
>> wrote:
>> 
>> Hi Joel,
>> 
>> as was suggested main.nqh_flips=False should disable this.
>> 
>> However I'm puzzled about this. NQH flips functionality is 
>> designed to flip these residues such that the clashes are 
>> minimized and plausible H-bonding is achieved. So I wonder why
>> it is not working in your case?
>> 
>> Would it be possible to send me input files (off list) so that I 
>> can reproduce this and investigate. Also please indicate HIS in 
>> question.
>> 
>> Thanks, Pavel
>> 
>> On 7/21/15 02:07, Joel Tyndall wrote:
>>> 
>>> 
>>> 
>>> Hi all,
>>> 
>>> 
>>> 
>>> We are trying to optimise a histidine interaction where NE2
>>> would ideally make a hydrogen bond with an adjacent tyrosine
>>> hydroxyl. The starting point contains the hydrogen bond.
>>> However upon refinement the ring flips (chi2 x 180 degrees) to
>>> place the CE1 adjacent to the tyrosine hydroxyl.
>>> 
>>> 
>>> 
>>> Is it possible to stop this as I see no reason why
>>> phenix.refine would want to do this
>>> 
>>> 
>>> 
>>> Regards
>>> 
>>> 
>>> 
>>> Joel
>>> 
>> 
>> 
>> 
> 
> 
> 
> _______________________________________________ phenixbb mailing
> list phenixbb at phenix-online.org 
> http://phenix-online.org/mailman/listinfo/phenixbb Unsubscribe:
> phenixbb-leave at phenix-online.org
> 
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.22 (MingW32)

iEYEARECAAYFAlWv2o8ACgkQU5C0gGfAG10nkwCggIs2Q8edzc9FwUFSYlTVXTvk
YPMAn2g5qmWw/ThzWL/ISOf9BNo1Y085
=A5WT
-----END PGP SIGNATURE-----


More information about the phenixbb mailing list