[modeller_usage] building in a loop

[jo hanna <>]

Hi

I have a quick question realting to this post from last October:

mathieu coincon wrote:
> I had a structure where a 15-residues loop is missing.
> So I want to create it but without moving the other atoms.
> I tried a lot of things (md_level= None, and so...) but I wasn't able to 
> obtain a model where the non-loop atoms stayed in place.
> I think that there is still an optimization I can't stop.
> After that I could try to optimize my loop with the refine_loop.py from 
> the advanced tutorial but first I need my model.
>The suggestion you received was nearly right - you should specify 
'exit_stage=2' when you call automake.make(). See 
http://salilab.org/modeller/manual/node52.html

>For example,

>a = automodel(...
>a.make(exit_stage=2)

>If you actually want to optimize the loop region, then simply select 
only the loop residues for optimization. See
http://salilab.org/modeller/manual/node28.html for optimization >using 
the standard potentials, and the second example at 
http://salilab.org/modeller/manual/node33.html for optimizing using 
Modeller's loop modeling potential.

>	Ben Webb, Modeller Caretaker

I have tried to add in resdiues missinhjg from a loop in my pdb in two ways;

1. using the exit stage method to build a model with the aurtomodel
class (where the resisues are added but in a complete straight line not
joined at both ends of the loop) and then loop optimisation to join the
added piece to the protein and optimise

and 

2. By using the loop modelling rotine directly or the refine only a
certain section to add in and then loop modelling this section.

My question is that the first method preoduces a model with a lower
objective function than the second, seemingly regardless of how many
models I build.

Would someone be able to comment on the validity of method one for building a loop in?

Thanks
Jo