[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: model_loop problem


the loop script works fine either for modeling one or multiple loops

In your top file a rarely used option was switched on: 

this command selects not only the loop segment but its spherical
environment, too. This approach does not work, because restraints are
derived from the initial structure, however during loop optimization the
selected environment of the loop is changing as its conformation is
changing.   Consequently  there will be a disagreement between the
selected set of atoms and  those atoms for which the restraints were
initially selected. Some atoms will not be restrained and they fly away.

This option will be removed from the future loop routine, if you use
older versions, please keep this line commented. You should select only
the loop segment or several loop segments.

>From a practical point of view, selecting the environment of the loop is
a not a good approach, because it means that the loop and its 
environment is 'ab initio' optimized, which is usually a too complicate
task for any existing  program, including this one. If you check our
paper (in fiser et al. Prot. Sci 2000, 9, 1753) we have discussed it, on
average 20-30 residues are  selected on average if you have an 8 residue
long loop and a 6 A cutoff. Not to mention multiple loop situation. That
is not any more a loop modeling problem, but rather ab initio modeling
of a small protein.

Our program is designed to optimize a loop(s) (efficiently upto 10-12
residues) with 'ab initio' restraints only, in a given protein

Nevertheless, you can select additional atoms/residues from the
environment but not simultaneously "during" the optimization protocol.

Please use a very simple separate top script, that selects with  SET
the list of residues/atoms from a spherical environment of your loop.
Next, you need to enter these regions to the loop script, as multiple
segments (even if it is only one residue long) and run the script.

Of course, the underlying practical problem surfaces here as well: in
principal you do not have a loop conformation because that is what you
are looking for. So selecting its environment prior to the optimization
is not trivial.  This may be more properly defined by selecting the
environment of the loop spanning residues or some other reasonable

best wishes,

 modelingGilles Labesse wrote:
> Hi,
> I have tried in vain to get model_loop to work properly using either
> modeller6v1 (directly using the model_loop script in
> examples/tutorial-model)
> or  modeller6v0 (when asking for modelling two loops at once).
> The modelled structures are 'blasted up' with numerous atoms far away
> from
> the core structure. However, asking for modelling one loop at once
> using modeller6v0 is find.
> The option SET DO_LOOP in __model.top is failling similarly.
> A slight detail PDB_EXT is applied to both sequence.BL00...,
> sequence.DL00... and sequence.IL00... files
> That's too bad as the new loop procedure looks promising.
> Thanks,
> Gilles
> --
> Dr Gilles Labesse
> Centre de Biochimie Structurale         Tel : +33 (0)4 67 04 38 47
> INSERM U554 - CNRS UMR5048 - UM1        Fax : +33 (0)4 67 52 96 23
> 15, Av. Charles Flahault                
> 34060 MONTPELLIER Cedex - FRANCE   http://www.cbs.cnrs.fr/~labesse/

Andras Fiser, PhD            # phone: (212) 327 7216
The Rockefeller University   # fax:   (212) 327 7540 
Box 270, 1230 York Avenue    # e-mail:
New York, NY 10021-6399, USA # http://salilab.org/~andras