hi Derek,
I agree with Azat, who pointed out the two extreme solution for you, and I
would like to add that using the MAKE_RESTRAINTS command and selecting its
arguments you can play with the restraint types to be used. E.g. selecting only
the stereo
chemical restraints and an arbitrary non bonded term like Lennard-Jones or
soft sphere, you will imitate a kind of CHARMM optimization. You can mix it up
gradually with additional restraints until you include all of them and you
"arrive" to comparative modelling.
Neglecting template derived restrains is kind of dangerous, as Azat also
pointed out, it means that you do not provide info about the fold of your
molecule thus during the optimization-refinement you can seriously diverge - to
anywhere.
With MAKE_RESTRAINTS you have a more continuous way to unselect restraints that
tie your model.
As a quick solution, I would SET DEVIATION to a few angstroms (different
starting conformations), set MD_LEVEL to the most thorough optimization
(refine1) and generate several models (a few dozen) and select the one with
the lowest energy. This is a reasonable refinement in a first approach and
easily can happen that it provides the best available answer.
Best wishes,
Andras
Azat Badretdinov wrote:
>
> Dear Derek!
>
> It is not very clear what do you want to do, but it looks like you do
> not want to use information in the templates. I know two ways to forget
> about the templates. First is to model the model based on the identical
> alignment with the model itself (you will be chained to original
> structure though - not a lot of room to optimize), second is to dealign
> the alignment with templates completely. Then MODELLER will use it's own
> homologueless general scoring function to minimize the structure. So far
> we know that this function works oK for short (<10 res) loops.
>
> Hope this helps.
>
> Derek Smith wrote:
> >
> > Dear Modellers
> >
> > I have a manually constructed model protein, which I would like to refine
> > using the MODELLER default schedule. I have attempted this using the protocol
> > based on question 12 in the MODELLER FAQ. However, I find that MODELLER reads
> > the model.ini file, and then extracts restraints from the templates, which
> > are then used in the optimization of the model. I would like to refine my
> > model based only upon my initial conformation. Is this possible do do, and if
> > so, how should I go about it?
>
> --
> - Dr. Azat Badretdinov
> - Molecular Simulations Inc.
> - 9685 Scranton Road
> - San Diego CA 92121-3752
> - Phone: (858) 799 5728
--
,
Andras Fiser, PhD # phone: (212) 327 7206
The Rockefeller University # fax: (212) 327 7540
Box 270, 1230 York Avenue # e-mail:
New York, NY 10021-6399, USA # http://salilab.org/~andras