I second Bernhard’s frustration with this, and have dealt with this many
times. I note that SHELX uses an occupancy-based clash criterion that
will not generate a clash for atoms whose occupancies sum to <1.1. It
seems to me that this could be a more generally adopted criterion.
Furthermore, it would be nice to move away from strictly altloc-based
clash criteria for correlated disorder and always use occupancies for this
purpose, since occupancies report information in the data and altlocs are
essentially a notation issue. I had mentioned this when comments were
solicited for the PDB validation task force, and perhaps more frustrated
voices will help advance the cause.
Best regards,
Mark
Mark A. Wilson
Associate Professor
Department of Biochemistry/Redox Biology Center
University of Nebraska
N118 Beadle Center
1901 Vine Street
Lincoln, NE 68588
(402) 472-3626
[log in to unmask]
On 2/1/16, 3:47 AM, "CCP4 bulletin board on behalf of Bernhard Rupp
(Hofkristallrat a.D.)" <[log in to unmask] on behalf of
[log in to unmask]> wrote:
>Dear Developers/PDBe,
>
>I just finished an otherwise unremarkable structure to the point of my
>personal level
>of tedium, which includes a few correlated partial group occupancies*).
>
>(a)
>Upon clash checking via coot/probe, the only 5 significant red porcupines
>involve those
>
>partial occupancies, say a side chain at 0.56 and a close water at 0.44,
>the idea that the water is only there when the side chain is in its
>alternate position.
>
>(b)
>running the coordinates through
>http://wwpdb-validation.wwpdb.org/validservice/
>lists exactly and only these 5 as worst ahem ‘errors’ **)
>
>I think it should be reasonably easy to fix the program so that clashes
>between atoms
>
>with sum(occ) =< 1 are not reported as errors. This *IS* relevant,
>because remembering the
>
>‘lively discussion’ there was the opinion that the PDB reports should
>suffice for a reviewer to
>judge the quality of a structure model***). As these ‘errors’ also show
>up in the PDB report, I
>see no reason why a reviewer should not simply tell me to clean up that
>mess (which one really
>should – these spikey porcupines usually DO mean something is wrong &
>fixable).
>
>
>(c)
>looking at the next non-severe pinkish clashes, I notice that most
>involve methyl groups,
>and if I eyeball them they are precisely staggered @60 deg around the
>H3C-X torsion. But when 2 Leu
>for example are involved, I doubt that they face off at precisely that
>torsion angle, because
>energetically this is a periodic potential. I think each methyl group
>torsion would give a little
>
>bit, and voila, pink gone.
>In this case, I worry less about the pink, but how is this treated when
>riding Hs are added in
>
>refinement, for the purpose of improved VdW restraints? Is this torsion
>potential used (as
>in MD) or to we assume the ‘hard’ 60 deg stagger or nothing conformation?
>Because if the
>torsion gives, also the C position might change slightly, ultimately
>resulting in a more realistic
>
>model. Has anyone played with that?
>
>Thx, BR
>
>*) which is best done at the very very end, because if you renumber
>residues or sort waters,
>the extra occupancy keyword file needs to be changed as well; a process
>whose tediousness
>
>is better not described. Indicative of errors in this file is a refmac
>infanticide message
>child killed: SIGABRT
>
>**) or is there a more current version of the validation server already
>online?
>
>***) Ceterum censeo coordinatae densitaeque providati eunt.
>
>--------------------------------------------------------------------------
>--------------
>CVMO
>Vista, CA 92084
>001 (925) 209-7429
>[log in to unmask]
>[log in to unmask]
>http://www.ruppweb.org/
>-----------------------------------------------------------------------
>
>
>
|