def load_gsd_topology(filename, frame=0):
""" Create an MDTraj.Topology from a GSD file
Parameters
----------
filename : path-like
Path of GSD trajectory file.
frame : int, 0
Frame of GSD file to parse topology
Returns
-------
top : mdtraj.Topology
Notes
-----
GSD files support systems with variable topologies.
For compatibility with MDTraj, only the topology from GSD frame 0 is
used to construct the MDTraj topology.
"""
import gsd.hoomd
with gsd.hoomd.open(filename, 'rb') as gsdfile:
top = Topology()
generic_chain = top.add_chain()
generic_residue = top.add_residue('A', generic_chain)
all_particle_types = gsdfile[frame].particles.types
for particle_type_id in gsdfile[frame].particles.typeid:
top.add_atom(all_particle_types[particle_type_id], virtual_site,
generic_residue)
for bond in gsdfile[frame].bonds.group:
atom1, atom2 = bond[0], bond[1]
top.add_bond(top.atom(atom1), top.atom(atom2))
return top
def mutate(self, mut_res_idx, mut_new_resname):
"""Mutate residue
Parameters
----------
mut_res_idx : int
Index of residue to mutate.
mut_new_resname : str
Three-letter code of residue to mutate to.
"""
assert (self.topology.residue(mut_res_idx).name != mut_new_resname), "mutating the residue to itself!"
# Build new topology
newTopology = Topology()
for chain in self.topology.chains:
newChain = newTopology.add_chain()
for residue in chain._residues:
res_idx = residue.index
if res_idx == mut_res_idx:
# create mutated residue
self._add_mutated_residue(mut_new_resname, newTopology, newChain, res_idx, residue)
else:
# copy old residue atoms directly
newResidue = newTopology.add_residue(residue.name, newChain, res_idx)
for atom in residue.atoms:
newTopology.add_atom(atom.name,
md.core.element.get_by_symbol(atom.element.symbol),
newResidue, serial=atom.index)
# The bond connectivity should stay identical
for atm1, atm2 in self.topology._bonds:
new_atm1 = newTopology.atom(atm1.index)
new_atm2 = newTopology.atom(atm2.index)
newTopology.add_bond(new_atm1, new_atm2)
self._prev_topology = self.topology.copy()
self.topology = newTopology
def load_hoomdxml(filename, top=None):
"""Load a single conformation from an HOOMD-Blue XML file.
For more information on this file format, see:
http://codeblue.umich.edu/hoomd-blue/doc/page_xml_file_format.html
Notably, all node names and attributes are in all lower case.
HOOMD-Blue does not contain residue and chain information explicitly.
For this reason, chains will be found by looping over all the bonds and
finding what is bonded to what.
Each chain consisists of exactly one residue.
Parameters
----------
filename : string
The path on disk to the XML file
top : None
This argumet is ignored
Returns
-------
trajectory : md.Trajectory
The resulting trajectory, as an md.Trajectory object, with corresponding
Topology.
Notes
-----
This function requires the NetworkX python package.
"""
from mdtraj.core.trajectory import Trajectory
from mdtraj.core.topology import Topology
topology = Topology()
tree = cElementTree.parse(filename)
config = tree.getroot().find('configuration')
position = config.find('position')
bond = config.find('bond')
atom_type = config.find('type') # MDTraj calls this "name"
box = config.find('box')
box.attrib = dict((key.lower(), val) for key, val in box.attrib.items())
# be generous for case of box attributes
lx = float(box.attrib['lx'])
ly = float(box.attrib['ly'])
lz = float(box.attrib['lz'])
try:
xy = float(box.attrib['xy'])
xz = float(box.attrib['xz'])
yz = float(box.attrib['yz'])
except (ValueError, KeyError):
xy = 0.0
xz = 0.0
yz = 0.0
unitcell_vectors = np.array([[[lx, xy * ly, xz * lz], [0.0, ly, yz * lz],
[0.0, 0.0, lz]]])
positions, types = [], {}
for pos in position.text.splitlines()[1:]:
positions.append((float(pos.split()[0]), float(pos.split()[1]),
float(pos.split()[2])))
for idx, atom_name in enumerate(atom_type.text.splitlines()[1:]):
types[idx] = str(atom_name.split()[0])
if len(types) != len(positions):
raise ValueError('Different number of types and positions in xml file')
# ignore the bond type
if hasattr(bond, 'text'):
bonds = [(int(b.split()[1]), int(b.split()[2]))
for b in bond.text.splitlines()[1:]]
chains = _find_chains(bonds)
else:
chains = []
bonds = []
# Relate the first index in the bonded-group to mdtraj.Residue
bonded_to_residue = {}
for i, _ in enumerate(types):
bonded_group = _in_chain(chains, i)
if bonded_group is not None:
if bonded_group[0] not in bonded_to_residue:
t_chain = topology.add_chain()
t_residue = topology.add_residue('A', t_chain)
bonded_to_residue[bonded_group[0]] = t_residue
topology.add_atom(types[i], virtual_site,
bonded_to_residue[bonded_group[0]])
if bonded_group is None:
t_chain = topology.add_chain()
t_residue = topology.add_residue('A', t_chain)
topology.add_atom(types[i], virtual_site, t_residue)
for bond in bonds:
atom1, atom2 = bond[0], bond[1]
topology.add_bond(topology.atom(atom1), topology.atom(atom2))
traj = Trajectory(xyz=np.array(positions), topology=topology)
traj.unitcell_vectors = unitcell_vectors
return traj
def load_hoomdxml(filename, top=None):
"""Load a single conformation from an HOOMD-Blue XML file.
For more information on this file format, see:
http://codeblue.umich.edu/hoomd-blue/doc/page_xml_file_format.html
Notably, all node names and attributes are in all lower case.
HOOMD-Blue does not contain residue and chain information explicitly.
For this reason, chains will be found by looping over all the bonds and
finding what is bonded to what.
Each chain consisists of exactly one residue.
Parameters
----------
filename : string
The path on disk to the XML file
Returns
-------
trajectory : md.Trajectory
The resulting trajectory, as an md.Trajectory object, with corresponding
Topology.
Notes
-----
This function requires the NetworkX python package.
"""
from mdtraj.core.trajectory import Trajectory
from mdtraj.core.topology import Topology
topology = Topology()
tree = cElementTree.parse(filename)
config = tree.getroot().find('configuration')
position = config.find('position')
bond = config.find('bond')
atom_type = config.find('type') # MDTraj calls this "name"
box = config.find('box')
box.attrib = dict((key.lower(), val) for key, val in box.attrib.items())
# be generous for case of box attributes
lx = float(box.attrib['lx'])
ly = float(box.attrib['ly'])
lz = float(box.attrib['lz'])
try:
xy = float(box.attrib['xy'])
xz = float(box.attrib['xz'])
yz = float(box.attrib['yz'])
except:
xy = 0.0
xz = 0.0
yz = 0.0
unitcell_vectors = np.array([[[lx, xy*ly, xz*lz],
[0.0, ly, yz*lz],
[0.0, 0.0, lz ]]])
positions, types = [], {}
for pos in position.text.splitlines()[1:]:
positions.append((float(pos.split()[0]),
float(pos.split()[1]),
float(pos.split()[2])))
for idx, atom_name in enumerate(atom_type.text.splitlines()[1:]):
types[idx] = str(atom_name.split()[0])
if len(types) != len(positions):
raise ValueError('Different number of types and positions in xml file')
# ignore the bond type
bonds = [(int(b.split()[1]), int(b.split()[2])) for b in bond.text.splitlines()[1:]]
chains = _find_chains(bonds)
ions = [i for i in range(len(types)) if not _in_chain(chains, i)]
# add chains, bonds and ions (each chain = 1 residue)
for chain in chains:
t_chain = topology.add_chain()
t_residue = topology.add_residue('A', t_chain)
for atom in chain:
topology.add_atom(types[atom], 'U', t_residue)
for ion in ions:
t_chain = topology.add_chain()
t_residue = topology.add_residue('A', t_chain)
topology.add_atom(types[atom], 'U', t_residue)
for bond in bonds:
atom1, atom2 = bond[0], bond[1]
topology.add_bond(topology.atom(atom1), topology.atom(atom2))
traj = Trajectory(xyz=np.array(positions), topology=topology)
traj.unitcell_vectors = unitcell_vectors
return traj
