Re: [modeller_usage] Re: model assessment method

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

Mensur Dlakic wrote:
> I am interested in the last part of this statement, which is calling 
> external programs. Could you provide a working example of how to call 
> ProsaII and evaluate each model? As it is, I use ProsaII at the end of 
> each Modeller run, but it would be nice if GA341 and ProsaII scores 
> could be obtained in parallel for easier comparison.

I don't use ProsaII here, so can't give you a working script, but you'd 
write a function such as:

import os

def prosaII(mdl):
    """Returns the ProsaII score of the given model."""

    # Write the current model to a PDB file, for ProsaII
    mdl.write(file='prosaII-input.pdb')

    # (insert Python code here to generate a prosaII input file)

    # Run prosaII with the generated input file and PDB
    os.system('prosaII prosaII-input.script')

    # (insert Python code here to parse the prosaII output, and
    # put the score into the Python variable 'pscore')

    # delete temporary files
    os.delete('prosaII-input.pdb')
    os.delete('prosaII-input.script')

    # Return the score to Modeller
    return ('ProsaII score', pscore)

Then you would simply pass this 'prosaII' function to 'assess_methods' 
rather than (or as well as) the normal assess.GA341 or assess.DOPE.

The rest is left as an exercise for the reader. ;) I suggest that the 
completed script be put into the Modeller wiki at 
http://salilab.org/modeller/wiki/ so everybody can use it.

> could be obtained in parallel for easier comparison. Also, is there any 
> way to get perl scripts for stand-alone pG calculations? Some time ago 
> Andrej (or Roberto) told me that those scripts were too unrefined to be 
> shared, but maybe that has changed?

I don't know if these scripts are still available or usable - sorry.

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