pmi/em.py

## \example pmi/em.py
"""This script shows how to create DENSITY representations.
and set up the Gaussian EM restraint.

These representations are useful when you are doing EM fitting with rigid bodies.

Preliminary step: you should convert your EM map to a GMM file
with the command line utility create_gmm.py (located in isd/pyext/src/create_gmm.py):
python create_gmm.py emd_1883.mrc 50 emd_1883.gmm50.txt -m emd_1883.gmm50.mrc
"""

import IMP
import IMP.atom
import IMP.algebra
import IMP.rmf
import IMP.pmi
import IMP.pmi.topology
import IMP.pmi.dof
import IMP.pmi.macros
import IMP.pmi.restraints.em
import sys

IMP.setup_from_argv(sys.argv, "Set up the EM restraint")

###################### SYSTEM SETUP #####################
# Preliminaries
mdl = IMP.Model()
seqs = IMP.pmi.topology.Sequences(IMP.pmi.get_example_path('data/1WCM.fasta'))

# Setup just one molecule
s = IMP.pmi.topology.System(mdl)
st = s.create_state()
mol = st.create_molecule("Rpn4",sequence=seqs["1WCM:D"],chain_id="D")
atomic_res = mol.add_structure(IMP.pmi.get_example_path('data/1WCM_fitted.pdb'),
                           chain_id="D",
                           offset=0)

# Below we create a GMM approximation for this moleucle
# This "DENSITY" representation is used in the GaussianEMRestraint (and others in the future)
# For structure regions we "fit" GMM components to all atom centers
mol.add_representation(atomic_res,
                       resolutions=[1,10],
                       density_residues_per_component=10, #how much to coarsen this representation
                       density_prefix="Rpn4_gmm",         # will write a .txt and .mrc file forcomponent
                       density_force_compute=False,       # set True if you want to overwrite
                       density_voxel_size=3.0)            # set to 0 if you don't care about writing the map
                                                          # if rasterizing takes too long, increase this value

# for the unstructured regions, we simply decorate each bead AS a gaussian, so no fitting is necessary
mol.add_representation(mol.get_non_atomic_residues(),
                       resolutions=[10],
                       setup_particles_as_densities=True) # just set this flag, nothing is written

hier = s.build()

# You can always check the representations with:
IMP.atom.show_with_representations(hier)

###################### RESTRAINTS #####################
output_objects = []

# To add the GaussianEMRestraint, first select all densities
densities = IMP.atom.Selection(hier,representation_type=IMP.atom.DENSITIES).get_selected_particles()
emr = IMP.pmi.restraints.em.GaussianEMRestraint(
    densities,
    target_fn=IMP.pmi.get_example_path('data/emd_1883.gmm50.txt'),  # created by user, see top of file
    slope=0.01,                           # a small number, helps drag bits into map
    scale_target_to_mass=False,           # if the model is the same size as map, usually set to True
    target_mass_scale=100000,             # manually set the mass of the target map (remove if you set above to True)
    weight=100.0)                         # the data weight
emr.add_to_model()
output_objects.append(emr)
mdl.update()
print(emr.evaluate())