Re: [modeller_usage] Refinement step questions

[Modeller Caretaker <modeller-care@ucsf.edu>]

On 1/19/14 11:36 AM, gbanoczi AT mail.bme.hu wrote:
>   I am currently doing a homology modeling project, in which salt
> bridges are important. Therefore I want to make the most out of the
> refinement protocol

Typically modifying the refinement protocol doesn't buy you much. We 
generally recommend you build more models if you want better sampling.

> My two questions are:
> the molecular dynamics refinement returns the final structure from each
> simulation, but I want it to return the ones with minimal energy
> (md_return='MINIMAL'), how could I specify it in my script while still
> using the standard automodel protocols (e.g. with md_level =
> refine.slow).

You would have to define your own refinement function to do this. The 
simplest way would be to copy refine.py from modlib/modeller/automodel/ 
to myrefine.py, change FINAL to MINIMAL, then "import myrefine" and 
"a.md_level = myrefine.slow" in your script.

> My other question is that how could I also specify in my
> script to create a CHARMM trajectory of the automodel MD refinement
> steps(I tried it with automodel.get_refine_actions(), but I failed to
> define it properly because of my mentioned lack of python knowledge)?

Hard to say what you did wrong, since it isn't in your script, but 
something like the following should work:

from modeller.optimizers import actions

class mymodel(automodel):
...
    def get_refine_actions(self):
       # Get default set of actions
       act = automodel.get_refine_actions(self)
       # Add our own
       c = actions.charmm_trajectory(5, filename='traj.dcd')
       act.append(c)
       return act

You'll need a matching PSF of course:
http://salilab.org/modeller/9.12/manual/node267.html

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