Re: [modeller_usage] model refinement and assessment

[Modeller Caretaker <>]

Stephen A. Smith wrote:
> In MODELLER 7v7 one was able to turn off model
> refinement using:
> SET MD_LEVEL = 'nothing'
>
> I am curious if there is an analogous way to do this
> in 8v0?  I've looked through the manual and examples
> and have not found a solution.  Have I missed
> something simple?

You just set it to the special Python value 'None', e.g.
a = automodel(...)
a.md_level = None

See http://salilab.org/modeller/manual/node35.html

> Is there a way to implement the GA341 and DOPE methods
> of analysis during the same model building session? 

I'm not sure what you mean. If you want to use both GA341 and DOPE to 
assess automodel models, then you just set
assess_methods = (assess.GA341, assess.DOPE)
when you create the automodel object - i.e. you can pass a list of 
methods, not just a single one. See 
http://salilab.org/modeller/manual/node34.html

> I'm also interested in using these analysis methods on
> other protein models and templates.  Is there a quick
> way to read these models into MODELLER and let it
> analyze them?

You can analyze an arbitrary PDB file with something like the following:
env = environ()
m = model(env, file='foo.pdb')
score = m.assess_dope()

Replace assess_dope with assess_ga341 to use GA341 instead. (But read 
the note at http://salilab.org/modeller/manual/node153.html first; GA341 
needs to know the sequence identity between model and template, which 
arbitrary PDB files may not have.)

DOPE is an unnormalized score, so it's best used within automodel to 
rank the generated models; scores for arbitrary PDBs are not generally 
comparable (just as with the Modeller scoring function).

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