> I am trying to get representative or average structure from molecular
> dynamics trajectory. I want to use it or an ensembel of them for the
> docking. I am using Molprobity as one of the assessment tools. I know
> that it is not that straghtforward to use it for molecular models
> assessment, but it is just some indicator for the comparison. The
> strange thing that I get is that the quality of models which I get from
> clustering both of representative structures and average is very bad
> while PDB file generated from the rst file is much better. I can't
> understand what is the basis for this. I can understand this for average
We cannot understand either since we have no way to know what molProbity
uses for assessment or whether the observed differences are "significant".
If you are clustering the trajectories (and not the .rst files) then there
is no reason that the "representative" should be equivalent to one of the
.rst files. If you are clustering the .rst files directly, then the
representative should match at least one of the .rst file structures; if
it doesn't, this is an issue... (but lack of agreement could result from
naming issues, see below).
Looking at the overlap between the .rst and representative
structures supplied, they are fairly close (1.6 A).
There are differences in the naming, etc however... The restrt file has
CYX residues (whereas the rep structure has CYS) at various places. The
"rep" is in some Accelyrs PDB file format; this could be choking the
molProbicity.
The .rst file has CB hydrogens named "2HB" and "3HB" (which is not
standard AMBER format (?) whereas the rep has "HB2" and "HB3". The .rst
has "HIE", etc whereas the rep only has "HIS". Make sure when you are
doing the molProb thing with PDB files were created by the same programs
(i.e. same format, naming conventions, etc).
--tec3
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Sep 15 2010 - 15:30:03 PDT
