I am trying to model missing residues into my crystal structure following the method on the website: https://salilab.org/modeller/wiki/Missing_residues. I am also trying to restrain the crystal coordinates so that only missing residues move (and the rest of the original crystal coordinates remain the same) during refinement using the select_atoms method. However when I compare my final models produced by my script to my original crystal structure, I find that the side chains of several amino acids three of four amino acids away from the missing residues have been flipped. How can I ensure that only the missing residues are refined and the crystal coordinates remain the same? I am wondering if I’ve missed something important out of my script. See my script below.
from modeller import * # Load standard Modeller classes
from modeller.automodel import * # Load the AutoModel class
log.verbose() # Create log file
env = Environ()
# directories for input atom files
env.io.atom_files_directory = ['.', '../atom_files']
env.io.water = True
# Only refine/move missing residues - restrain crystal coordinates
#picks residues to be refined by loop modelling
#refines residue ranges defined simultaneously
return Selection(self.residue_range('5:A', '13:A'),
# redefine the special_patches routine to include additional disulfides
a = MyModel(env, alnfile = 'alignment.ali', # alignment file
knowns = 'xxx', # aa sequence of original template - crystal coordinates
sequence = 'xxx_fill', # aa sequence of original template with missing residues filled in
loop_assess_methods = (assess.DOPE,
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 refined model
a.loop.ending_model = 10 # Last loop refined model
a.loop.md_level = refine.fast # Loop model refinement level
a.make() # Do modelling
I am a French System. administrator and I just wanted to know it you would consider adding Modeller to the Spack package manager ?
It is a very useful and flexible package managers that is used mostly in HPC Supercomputer environments.
There are already 7000+ package recipes available and many worldwide HPC supercomputers use Spack.
I believe it would be really useful for the scientific (and bioinformatics) community to have Modeller available too.
Here are some useful links:
Main homepage: https://spack.io/
Adding a recipe to build/compile a program such as Modeller would not be too complicated, it is just that one must adapt the actual Install script. As is, the installation of modeler via your Install script does not really fit into the “usual” ways of building softwares such as make, cmake, or for python packages pypi, setuptools, etc.
Thank you in advance for considering the effort,
Ingénieur d'études - Administrateur des systèmes d'information
Inserm | Université Paris Cité | DSIMB UMR_S U1134
Site Broussais, Bâtiment Ady Steg
8 rue Maria Helena Vieira Da Silva | 75014 Paris
Tél. 01 81 72 43 31 | Mob. 07 81 70 35 15