Re: [modeller_usage] initial model

[Modeller Caretaker <>]

On 08/14/2010 05:38 AM, sdh wrote:
> I want to model a multi-domain protein. Thus, I aligned manually all
> templates (corresponding to the individual domains) and used them as an
> input for Modeller. (Note that the templates overlap only over few
> residues.)
>
> However, when I compared the initial model (.ini) to the aligned
> templates, I saw that Modeller did something more than just transferring
> the coordinates from the templates. Namely, the relative orientation of
> domains is not as in my structural alignment. Is thee any way to force
> Modeller to use exactly the input coordinates?

Modeller simply transfers the coordinates from templates to the target 
in regions where there is only a single aligned template. Note that 
before it does this, however, it checks the alignment for sanity. In so 
doing, it does a pairwise structural alignment of the templates:
http://salilab.org/modeller/9v8/manual/node275.html

If you don't want it to do this, override the 
automodel.check_alignment() method, e.g.

class MyModel(automodel):
    def check_alignment(self, aln):
        pass

a = MyModel(...)

Alternatively, you can supply your own initial model in whatever 
orientation you prefer:
http://salilab.org/modeller/9v8/manual/node26.html

Note that unless you have sufficient overlap between your domains, you 
won't generate accurate inter-domain restraints, and so you won't 
necessarily get a good final model even if your initial one looks OK.

	Ben Webb, Modeller Caretaker
--
             http://www.salilab.org/modeller/
Modeller mail list: http://salilab.org/mailman/listinfo/modeller_usage