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 __init__(self, topology):
r"""Calpha representation mapping
Maps an all-atom representation to just the C-alpha's of the backbone.
Holds default assignment of .
Parameters
----------
topology : mdtraj.Topology object
"""
n_calphas = len(
[atm.index for atm in topology.atoms if atm.name == "CA"])
assert (
n_calphas == topology.n_residues
), " number of C-alpha is not equal to number of residues! check for missing or non-standard amino acids."
self._ref_topology = topology.copy()
# Build new topology
newTopology = Topology()
prev_ca = None
ca_idxs = []
atm_idx = 0
for chain in topology._chains:
newChain = newTopology.add_chain()
for residue in chain._residues:
resSeq = getattr(residue, 'resSeq', None) or residue.index
newResidue = newTopology.add_residue(residue.name, newChain,
resSeq)
# map CA
new_ca = newTopology.add_atom(
'CA',
md.core.element.get_by_symbol('C'),
newResidue,
serial=atm_idx)
ca_idxs.append([[ atm.index for atm in residue.atoms if \
(atm.name == "CA") ][0], atm_idx ])
if prev_ca is None:
prev_ca = new_ca
else:
if prev_ca.residue.chain.index == new_ca.residue.chain.index:
# Only bond atoms in same chain
newTopology.add_bond(prev_ca, new_ca)
prev_ca = new_ca
atm_idx += 1
self._ca_idxs = np.array(ca_idxs)
self.topology = newTopology
def topology(self):
"""Get the topology out from the file
Returns
-------
topology : mdtraj.Topology
A topology object
"""
try:
raw = self._get_node(self._handle.root, name='topology')[0]
if not isinstance(raw, string_types):
raw = raw.decode()
topology_dict = json.loads(raw)
except self.tables.NoSuchNodeError:
return None
topology = Topology()
for chain_dict in sorted(topology_dict['chains'],
key=operator.itemgetter('index')):
chain = topology.add_chain()
for residue_dict in sorted(chain_dict['residues'],
key=operator.itemgetter('index')):
try:
resSeq = residue_dict["resSeq"]
except KeyError:
resSeq = None
warnings.warn(
'No resSeq information found in HDF file, defaulting to zero-based indices'
)
try:
segment_id = residue_dict["segmentID"]
except KeyError:
segment_id = ""
residue = topology.add_residue(residue_dict['name'],
chain,
resSeq=resSeq,
segment_id=segment_id)
for atom_dict in sorted(residue_dict['atoms'],
key=operator.itemgetter('index')):
try:
element = elem.get_by_symbol(atom_dict['element'])
except KeyError:
element = elem.virtual
topology.add_atom(atom_dict['name'], element, residue)
atoms = list(topology.atoms)
for index1, index2 in topology_dict['bonds']:
topology.add_bond(atoms[index1], atoms[index2])
return topology
def create_water_topology_on_disc(n):
topfile = tempfile.mktemp('.pdb')
top = Topology()
chain = top.add_chain()
for i in range(n):
res = top.add_residue('r%i' % i, chain)
h1 = top.add_atom('H', hydrogen, res)
o = top.add_atom('O', oxygen, res)
h2 = top.add_atom('H', hydrogen, res)
top.add_bond(h1, o)
top.add_bond(h2, o)
xyz = np.zeros((n * 3, 3))
Trajectory(xyz, top).save_pdb(topfile)
return topfile
def _read_models(self):
if not self._mode == 'r':
raise ValueError('file not opened for reading')
self._topology = Topology()
pdb = PdbStructure(self._file, load_all_models=True)
atomByNumber = {}
for chain in pdb.iter_chains():
c = self._topology.add_chain()
for residue in chain.iter_residues():
resName = residue.get_name()
if resName in PDBTrajectoryFile._residueNameReplacements:
resName = PDBTrajectoryFile._residueNameReplacements[resName]
r = self._topology.add_residue(resName, c, residue.number)
r.segment_id = residue.segment_id
if resName in PDBTrajectoryFile._atomNameReplacements:
atomReplacements = PDBTrajectoryFile._atomNameReplacements[resName]
else:
atomReplacements = {}
for atom in residue.atoms:
atomName = atom.get_name()
if atomName in atomReplacements:
atomName = atomReplacements[atomName]
atomName = atomName.strip()
element = atom.element
if element is None:
element = self._guess_element(atomName, residue)
newAtom = self._topology.add_atom(atomName, element, r, serial=atom.serial_number)
atomByNumber[atom.serial_number] = newAtom
# load all of the positions (from every model)
_positions = []
for model in pdb.iter_models(use_all_models=True):
coords = []
for chain in model.iter_chains():
for residue in chain.iter_residues():
for atom in residue.atoms:
coords.append(atom.get_position())
_positions.append(coords)
if not all(len(f) == len(_positions[0]) for f in _positions):
raise ValueError('PDB Error: All MODELs must contain the same number of ATOMs')
self._positions = np.array(_positions)
## The atom positions read from the PDB file
self._unitcell_lengths = pdb.get_unit_cell_lengths()
self._unitcell_angles = pdb.get_unit_cell_angles()
self._topology.create_standard_bonds()
self._topology.create_disulfide_bonds(self.positions[0])
# Add bonds based on CONECT records.
connectBonds = []
for connect in pdb.models[-1].connects:
i = connect[0]
for j in connect[1:]:
if i in atomByNumber and j in atomByNumber:
connectBonds.append((atomByNumber[i], atomByNumber[j]))
if len(connectBonds) > 0:
# Only add bonds that don't already exist.
existingBonds = set(self._topology.bonds)
for bond in connectBonds:
if bond not in existingBonds and (bond[1], bond[0]) not in existingBonds:
self._topology.add_bond(bond[0], bond[1])
existingBonds.add(bond)
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 _read(self):
"Read a single frame"
from mdtraj.core.topology import Topology
from mdtraj.core.element import Element, virtual
# Read in the number of atoms.
line = self._fh.readline()
if line == '':
raise _EOF()
self._n_atoms = int(line.split()[0])
self._line_counter += 1
coords = np.empty((self._n_atoms, 3), dtype=np.float32)
bond_partners = [[] for i in xrange(self._n_atoms)]
atom_names = ['' for i in xrange(self._n_atoms)]
line = self._fh.readline()
s = line.split()
self._line_counter += 1
# See if we have box info on this line or not
cell_lengths = cell_angles = None
if len(s) == 6:
try:
cell_lengths = np.asarray(
[float(s[0]), float(s[1]),
float(s[2])])
cell_angles = np.asarray(
[float(s[3]), float(s[4]),
float(s[5])])
line = self._fh.readline()
self._line_counter += 1
except ValueError:
pass
i = 0
while i < self._n_atoms - 1:
atom_names[i] = s[1]
bond_partners[i] = [int(x) for x in s[6:]]
coords[i, :] = [float(s[pos]) for pos in [2, 3, 4]]
i += 1
line = self._fh.readline()
s = line.split()
self._line_counter += 1
# Now do the last atom
atom_names[i] = s[1]
bond_partners[i] = [int(x) for x in s[6:]]
coords[i, :] = [float(s[pos]) for pos in [2, 3, 4]]
# Now see if we have to build a topology
if self.topology is None:
self.topology = top = Topology()
chain = top.add_chain() # only 1 chain
res = top.add_residue('RES', chain, 1) # only 1 residue
for at in atom_names:
# First get the element. Try for common 2-letter elements, then
# use the first letter only (default to None if I can't find it)
if at[:2].upper() in ('NA', 'CL', 'MG'):
elem = Element.getBySymbol(at[:2])
else:
try:
elem = Element.getBySymbol(at[0])
except KeyError:
elem = virtual
top.add_atom(at, elem, res)
# Now add the bonds
atoms = list(top.atoms)
for i, bonds in enumerate(bond_partners):
me = atoms[i]
for b in bonds:
b -= 1
if b < i: continue
it = atoms[b]
top.add_bond(me, it)
self._frame_index += 1
return coords, cell_lengths, cell_angles
def __init__(self, topology, use_chains=None):
if use_chains is None:
use_chains = range(len(topology._chains))
self._ref_topology = topology.copy()
# Build new topology
newTopology = Topology()
new_atm_idx = 0
res_idx = 1
prev_ca = None
ca_idxs = []
self._sidechain_idxs = []
self._sidechain_mass = []
self._chain_indices = []
for chain_count, chain in enumerate(topology._chains):
if chain_count in use_chains:
newChain = newTopology.add_chain()
for residue in chain._residues:
#resSeq = getattr(residue, 'resSeq', None) or residue.index
newResidue = newTopology.add_residue(
residue.name, newChain, res_idx)
# map CA
new_ca = newTopology.add_atom(
'CA',
md.core.element.get_by_symbol('C'),
newResidue,
serial=new_atm_idx)
self._chain_indices.append(chain_count)
if prev_ca is None:
prev_ca = new_ca
else:
# only bond atoms in the same chain.
if new_ca.residue.chain.index == prev_ca.residue.chain.index:
newTopology.add_bond(prev_ca, new_ca)
prev_ca = new_ca
try:
ca_idxs.append([[ atm.index for atm in residue.atoms if \
(atm.name == "CA") ][0], new_atm_idx ])
except:
print(residue)
print(chain)
for atm in residue.atoms:
atm.name
raise
new_atm_idx += 1
if residue.name == 'GLY':
self._sidechain_idxs.append([])
self._sidechain_mass.append([])
else:
# map CB
cb_name = "CB%s" % atom_types.residue_code[
residue.name]
new_cb = newTopology.add_atom(
cb_name,
md.core.element.get_by_symbol('C'),
newResidue,
serial=new_atm_idx)
self._chain_indices.append(chain_count)
newTopology.add_bond(new_cb, new_ca)
self._sidechain_idxs.append([[ atm.index for atm in residue.atoms if \
(atm.is_sidechain) and (atm.element.symbol != "H") ], new_atm_idx ])
self._sidechain_mass.append(np.array([ atm.element.mass for atm in residue.atoms if \
(atm.is_sidechain) and (atm.element.symbol != "H") ]))
new_atm_idx += 1
res_idx += 1
self._ca_idxs = np.array(ca_idxs)
self.topology = newTopology
assert self.topology.n_atoms == len(self._chain_indices)
def extract(item, atom_indices='all', copy_if_all=True, check=True):
if check:
digest_item(item, 'mdtraj.Topology')
atom_indices = digest_atom_indices(atom_indices)
if atom_indices is 'all':
if copy_if_all:
from copy import deepcopy
tmp_item = deepcopy(item)
else:
tmp_item = item
else:
from mdtraj.core.topology import Topology
from mdtraj.utils import ilen
atom_indices_to_be_kept = set(atom_indices)
newTopology = Topology()
old_atom_to_new_atom = {}
for chain in item._chains:
newChain = newTopology.add_chain()
for group in chain._groups:
resSeq = getattr(group, 'resSeq', None) or group.index
newResidue = newTopology.add_group(group.name, newChain,
resSeq, group.segment_id)
for atom in group._atoms:
if atom.index in atom_indices_to_be_kept:
try: # OpenMM Topology objects don't have serial attributes, so we have to check first.
serial = atom.serial
except AttributeError:
serial = None
newAtom = newTopology.add_atom(atom.name,
atom.element,
newResidue,
serial=serial)
old_atom_to_new_atom[atom] = newAtom
bondsiter = item.bonds
if not hasattr(bondsiter, '__iter__'):
bondsiter = bondsiter()
for bond in bondsiter:
try:
atom1, atom2 = bond
newTopology.add_bond(old_atom_to_new_atom[atom1],
old_atom_to_new_atom[atom2],
type=bond.type,
order=bond.order)
except KeyError:
pass
# we only put bonds into the new topology if both of their partners
# were indexed and thus HAVE a new atom
# Delete empty groups
newTopology._groups = [
r for r in newTopology._groups if len(r._atoms) > 0
]
for chain in newTopology._chains:
chain._groups = [r for r in chain._groups if len(r._atoms) > 0]
# Delete empty chains
newTopology._chains = [
c for c in newTopology._chains if len(c._groups) > 0
]
# Re-set the numAtoms and numResidues
newTopology._numAtoms = ilen(newTopology.atoms)
newTopology._numResidues = ilen(newTopology.groups)
tmp_item = newTopology
return tmp_item
def _to_topology(self, atom_list, chain_types=None, residue_types=None):
"""Create a mdtraj.Topology from a Compound.
Parameters
----------
atom_list :
chain_types :
residue_types :
Returns
-------
top : mtraj.Topology
"""
from mdtraj.core.element import get_by_symbol
from mdtraj.core.topology import Topology
if isinstance(chain_types, Compound):
chain_types = [Compound]
if isinstance(chain_types, (list, set)):
chain_types = tuple(chain_types)
if isinstance(residue_types, Compound):
residue_types = [Compound]
if isinstance(residue_types, (list, set)):
residue_types = tuple(residue_types)
top = Topology()
atom_mapping = {}
default_chain = top.add_chain()
default_residue = top.add_residue('RES', default_chain)
last_residue_compound = None
last_chain_compound = None
last_residue = None
last_chain = None
for atom in atom_list:
# Chains
for parent in atom.ancestors():
if chain_types and isinstance(parent, chain_types):
if parent != last_chain_compound:
last_chain_compound = parent
last_chain = top.add_chain()
last_chain_default_residue = top.add_residue(
'RES', last_chain)
last_chain.compound = last_chain_compound
break
else:
last_chain = default_chain
last_chain.compound = last_chain_compound
# Residues
for parent in atom.ancestors():
if residue_types and isinstance(parent, residue_types):
if parent != last_residue_compound:
last_residue_compound = parent
last_residue = top.add_residue(
parent.__class__.__name__, last_chain)
last_residue.compound = last_residue_compound
break
else:
if last_chain != default_chain:
last_residue = last_chain_default_residue
else:
last_residue = default_residue
last_residue.compound = last_residue_compound
# Add the actual atoms
try:
elem = get_by_symbol(atom.name)
except KeyError:
elem = get_by_symbol("VS")
at = top.add_atom(atom.name, elem, last_residue)
at.charge = atom.charge
atom_mapping[atom] = at
# Remove empty default residues.
chains_to_remove = [
chain for chain in top.chains if chain.n_atoms == 0
]
residues_to_remove = [res for res in top.residues if res.n_atoms == 0]
for chain in chains_to_remove:
top._chains.remove(chain)
for res in residues_to_remove:
for chain in top.chains:
try:
chain._residues.remove(res)
except ValueError: # Already gone.
pass
for atom1, atom2 in self.bonds():
# Ensure that both atoms are part of the compound. This becomes an
# issue if you try to convert a sub-compound to a topology which is
# bonded to a different subcompound.
if all(a in atom_mapping.keys() for a in [atom1, atom2]):
top.add_bond(atom_mapping[atom1], atom_mapping[atom2])
return top
