[modeller_usage] loop optimization in the presence of a ligand

["Thijs Beuming" <>]

Hi,

I am trying to optimize a loop in close proximity to a ligand. For  
some reason, the intial models have the ligand present, while the  
loop optimized models do not. Hence, there are major clashes between  
some of the calculated loops and the ligand. Is it possible to  
optimize loops in the presence of the ligand?

My input script:

# Homology modeling by the automodel class
from modeller import *              # Load standard Modeller classes
from modeller.automodel import *    # Load the automodel class

log.verbose()    # request verbose output
env = environ()  # create a new MODELLER environment to build this  
model in

# directories for input atom files
env.io.atom_files_directory = 'MODELLER'

class mymodel(loopmodel):
    def special_restraints(self, aln):
        rsr = self.restraints
        rsr.add(atom_ids=('CA:189', 'CA:372'),
                restraint_parameters=(3, 1, 1, 27, 2, 2, 0, 10.0, 0.1))
        rsr.add(atom_ids=('CA:189', 'CA:393'),
                restraint_parameters=(3, 1, 1, 27, 2, 2, 0, 10.0, 0.1))
        rsr.add(atom_ids=('CA:372', 'CA:393'),
                restraint_parameters=(3, 1, 1, 27, 2, 2, 0, 10.0, 0.1))

   def special_patches(self, aln):
        self.patch(residue_type='DISU', residues=(self.residues['176'],
                                                  self.residues 
['185']))
    def select_loop_atoms(self):
        return selection(self.residue_range('378:', '383:'))

a = mymodel(env,
              alnfile  = 'MODELLER/alignment.pir',      # alignment  
filename
              knowns   = 'template',              # codes of the  
templates
              sequence = 'target')               # code of the target

a.starting_model= 1                 # index of the first model
a.ending_model  = 1                # index of the last model

a.loop.starting_model = 1           # First loop model
a.loop.ending_model   = 2           # Last loop model

a.make()                            # do the actual homology modeling

Thanks,

Thijs Beuming