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):
self._ref_topology = topology.copy()
self._N_coeff = np.array([0.48318, 0.70328, -0.18643]) # CA_i-1, CA_i, O_i-1
self._C_coeff = np.array([0.44365, 0.23520, 0.32115]) # CA_i, CA_i+1, O_i
self._H_coeff = np.array([0.84100, 0.89296, -0.73389]) # CA_i-1, CA_i, O_i-1
newTopology = Topology()
CACBO_idxs = []; N_idxs = []; C_idxs = []; H_idxs = []
res_idx = 1
atm_idx = 0
prev_ca = None
prev_o = None
for chain in topology._chains:
newChain = newTopology.add_chain()
for residue in chain._residues:
newTopology, atm_idx, res_idx = self._add_residue(newTopology,
newChain, residue, chain, res_idx, atm_idx,
N_idxs, C_idxs, H_idxs, CACBO_idxs, prev_ca, prev_o)
self._CACBO_idxs = np.array(CACBO_idxs)
self._N_idxs = np.array(N_idxs)
self._C_idxs = np.array(C_idxs)
self._H_idxs = np.array(H_idxs)
self.topology = newTopology
def __init__(self, topology):
self._ref_topology = topology.copy()
# Used for recentering sidechain atom.
self._N_CB_dist = 0.2470955 # distance from N_i to CB_i
self._CA_CB_dist = 0.1533931 # distance from CA_i to CB_i
self._C_CB_dist = 0.2510052 # distance from C_i to CB_i
self._N_coeff = np.array([0.48318, 0.70328, -0.18643]) # CA_i-1, CA_i, O_i-1
self._C_coeff = np.array([0.44365, 0.23520, 0.32115]) # CA_i, CA_i+1, O_i
self._H_coeff = np.array([0.84100, 0.89296, -0.73389]) # CA_i-1, CA_i, O_i-1
# Used for positioning HB's for the moment.
self._HB_w = 3
self._disulfides = []
# Build new topology
newTopology = Topology()
CACBO_idxs = []
HB_idxs = []
res_charges = []
res_idx = 1
atm_idx = 0
for chain in topology._chains:
newChain = newTopology.add_chain()
for residue in chain._residues:
newTopology, atm_idx, res_idx = self._add_residue(newTopology,
newChain, residue, res_charges, res_idx, atm_idx,
CACBO_idxs, HB_idxs)
self._charged_residues = res_charges
self._HB_idxs = np.array(HB_idxs)
self._CACBO_idxs = np.array(CACBO_idxs)
self.topology = newTopology
def _initialize(self, simulation):
"""Deferred initialization of the reporter, which happens before
processing the first report.
At the time that the first report is processed, we now have access
to the simulation object, which we don't have at the point when the
reporter is instantiated
Parameters
----------
simulation : simtk.openmm.app.Simulation
The Simulation to generate a report for
"""
if self._atomSubset is not None:
if not min(self._atomSubset) >= 0:
raise ValueError('atomSubset must be zero indexed. '
'the smallest allowable value is zero')
if not max(self._atomSubset) < simulation.system.getNumParticles():
raise ValueError(
'atomSubset must be zero indexed. '
'the largest value must be less than the number '
'of particles in the system')
if not all(a == int(a) for a in self._atomSubset):
raise ValueError(
'all of the indices in atomSubset must be integers')
self._atomSlice = self._atomSubset
if hasattr(self._traj_file, 'topology'):
self._traj_file.topology = _topology_from_subset(
Topology.from_openmm(simulation.topology),
self._atomSubset)
else:
self._atomSlice = slice(None)
if hasattr(self._traj_file, 'topology'):
self._traj_file.topology = Topology.from_openmm(
simulation.topology)
system = simulation.system
if self._temperature:
# Compute the number of degrees of freedom.
dof = 0
for i in range(system.getNumParticles()):
if system.getParticleMass(i) > 0 * units.dalton:
dof += 3
dof -= system.getNumConstraints()
if any(
type(system.getForce(i)) == mm.CMMotionRemover
for i in range(system.getNumForces())):
dof -= 3
self._dof = dof
if simulation.topology.getUnitCellDimensions() is None:
self._cell = False
def _initialize(self, simulation):
"""Deferred initialization of the reporter, which happens before
processing the first report.
At the time that the first report is processed, we now have access
to the simulation object, which we don't have at the point when the
reporter is instantiated
Parameters
----------
simulation : simtk.openmm.app.Simulation
The Simulation to generate a report for
"""
if self._atomSubset is not None:
if not min(self._atomSubset) >= 0:
raise ValueError('atomSubset must be zero indexed. '
'the smallest allowable value is zero')
if not max(self._atomSubset) < simulation.system.getNumParticles():
raise ValueError('atomSubset must be zero indexed. '
'the largest value must be less than the number '
'of particles in the system')
if not all(a==int(a) for a in self._atomSubset):
raise ValueError('all of the indices in atomSubset must be integers')
self._atomSlice = self._atomSubset
if hasattr(self._traj_file, 'topology'):
self._traj_file.topology = _topology_from_subset(
Topology.from_openmm(simulation.topology), self._atomSubset)
else:
self._atomSlice = slice(None)
if hasattr(self._traj_file, 'topology'):
self._traj_file.topology = Topology.from_openmm(simulation.topology)
system = simulation.system
if self._temperature:
# Compute the number of degrees of freedom.
dof = 0
for i in range(system.getNumParticles()):
if system.getParticleMass(i) > 0*units.dalton:
dof += 3
dof -= system.getNumConstraints()
if any(type(system.getForce(i)) == mm.CMMotionRemover for i in range(system.getNumForces())):
dof -= 3
self._dof = dof
if simulation.topology.getUnitCellDimensions() is None:
self._cell = False
def to_mdtraj_Topology(item, atom_indices='all', check=True):
if check:
try:
is_openmm_Topology(item)
except:
raise WrongFormError('openmm.Topology')
try:
atom_indices = digest_atom_indices(atom_indices)
except:
raise WrongAtomIndicesError()
try:
from mdtraj.core.topology import Topology as mdtraj_Topology
except:
raise LibraryNotFoundError('MDTraj')
from . import extract
tmp_item = extract(item,
atom_indices=atom_indices,
copy_if_all=False,
check=False)
tmp_item = mdtraj_Topology.from_openmm(tmp_item)
return tmp_item
def load_mol2(filename):
"""Load a TRIPOS mol2 file from disk.
Parameters
----------
filename : str
Path to the prmtop file on disk.
Returns
-------
traj : md.Trajectory
The resulting topology, as an md.Topology object.
Notes
-----
This function should work on GAFF and sybyl style MOL2 files, but has
been primarily tested on GAFF mol2 files.
This function does NOT accept multi-structure MOL2 files!!!
The elements are guessed using GAFF atom types or via the atype string.
Examples
--------
>>> traj = md.load_mol2('mysystem.mol2')
"""
from mdtraj.core.trajectory import Trajectory
from mdtraj.core.topology import Topology
atoms, bonds = mol2_to_dataframes(filename)
atoms_mdtraj = atoms[["name", "resName"]].copy()
atoms_mdtraj["serial"] = atoms.index
#Figure out 1 letter element names
# IF this is a GAFF mol2, this line should work without issues
atoms_mdtraj["element"] = atoms.atype.map(gaff_elements)
# If this is a sybyl mol2, there should be NAN (null) values
if atoms_mdtraj.element.isnull().any():
# If this is a sybyl mol2, I think this works generally.
atoms_mdtraj["element"] = atoms.atype.apply(lambda x: x.strip(".")[0])
atoms_mdtraj["resSeq"] = np.ones(len(atoms), 'int')
atoms_mdtraj["chainID"] = np.ones(len(atoms), 'int')
if bonds is not None:
bonds_mdtraj = bonds[["id0", "id1"]].values
offset = bonds_mdtraj.min() # Should this just be 1???
bonds_mdtraj -= offset
else:
bonds_mdtraj = None
top = Topology.from_dataframe(atoms_mdtraj, bonds_mdtraj)
xyzlist = np.array([atoms[["x", "y", "z"]].values])
xyzlist /= 10.0 # Convert from angstrom to nanometer
traj = Trajectory(xyzlist, top)
return traj
def load_mol2(filename):
"""Load a TRIPOS mol2 file from disk.
Parameters
----------
filename : str
Path to the prmtop file on disk.
Returns
-------
traj : md.Trajectory
The resulting topology, as an md.Topology object.
Notes
-----
This function should work on GAFF and sybyl style MOL2 files, but has
been primarily tested on GAFF mol2 files.
This function does NOT accept multi-structure MOL2 files!!!
The elements are guessed using GAFF atom types or via the atype string.
Examples
--------
>>> traj = md.load_mol2('mysystem.mol2')
"""
from mdtraj.core.trajectory import Trajectory
from mdtraj.core.topology import Topology
atoms, bonds = mol2_to_dataframes(filename)
atoms_mdtraj = atoms[["name", "resName"]].copy()
atoms_mdtraj["serial"] = atoms.index
#Figure out 1 letter element names
# IF this is a GAFF mol2, this line should work without issues
atoms_mdtraj["element"] = atoms.atype.map(gaff_elements)
# If this is a sybyl mol2, there should be NAN (null) values
if atoms_mdtraj.element.isnull().any():
# If this is a sybyl mol2, I think this works generally.
atoms_mdtraj["element"] = atoms.atype.apply(lambda x: x.strip(".")[0])
atoms_mdtraj["resSeq"] = np.ones(len(atoms), 'int')
atoms_mdtraj["chainID"] = np.ones(len(atoms), 'int')
if bonds is not None:
bonds_mdtraj = bonds[["id0", "id1"]].values
offset = bonds_mdtraj.min() # Should this just be 1???
bonds_mdtraj -= offset
else:
bonds_mdtraj = None
top = Topology.from_dataframe(atoms_mdtraj, bonds_mdtraj)
xyzlist = np.array([atoms[["x", "y", "z"]].values])
xyzlist /= 10.0 # Convert from angstrom to nanometer
traj = Trajectory(xyzlist, top)
return traj
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 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 xrange(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 __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 to_mdtraj_Topology(item,
molecular_system=None,
atom_indices='all',
structure_indices='all'):
from mdtraj.core.topology import Topology
tmp_item, tmp_molecular_system = to_openmm_Topology(
item,
molecular_system=molecular_system,
atom_indices=atom_indices,
structure_indices=structure_indices)
tmp_item = Topology.from_openmm(tmp_item)
if tmp_molecular_system is not None:
tmp_molecular_system = tmp_molecular_system.combine_with_items(
tmp_item)
return tmp_item, tmp_molecular_system
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 topology(self):
"""Get the topology out from the file
Returns
-------
topology : mdtraj.Topology
A topology object
"""
try:
raw = self._get_node('/', 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 topology(self):
"""Get the topology out from the file
Returns
-------
topology : mdtraj.Topology
A topology object
"""
try:
raw = self._get_node("/", 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")
residue = topology.add_residue(residue_dict["name"], chain, resSeq=resSeq)
for atom_dict in sorted(residue_dict["atoms"], key=operator.itemgetter("index")):
try:
element = elem.get_by_symbol(atom_dict["element"])
except KeyError:
element = None
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 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 topology(self, topology_object):
"""Set the topology in the file
Parameters
----------
topology_object : mdtraj.Topology
A topology object
"""
_check_mode(self.mode, ('w', 'a'))
# we want to be able to handle the simtk.openmm Topology object
# here too, so if it's not an mdtraj topology we'll just guess
# that it's probably an openmm topology and convert
if not isinstance(topology_object, Topology):
topology_object = Topology.from_openmm(topology_object)
try:
topology_dict = {
'chains': [],
'bonds': []
}
for chain in topology_object.chains:
chain_dict = {
'residues': [],
'index': int(chain.index)
}
for residue in chain.residues:
residue_dict = {
'index': int(residue.index),
'name': str(residue.name),
'atoms': [],
"resSeq": int(residue.resSeq),
"segmentID": str(residue.segment_id)
}
for atom in residue.atoms:
try:
element_symbol_string = str(atom.element.symbol)
except AttributeError:
element_symbol_string = ""
residue_dict['atoms'].append({
'index': int(atom.index),
'name': str(atom.name),
'element': element_symbol_string
})
chain_dict['residues'].append(residue_dict)
topology_dict['chains'].append(chain_dict)
for atom1, atom2 in topology_object.bonds:
topology_dict['bonds'].append([
int(atom1.index),
int(atom2.index)
])
except AttributeError as e:
raise AttributeError('topology_object fails to implement the'
'chains() -> residue() -> atoms() and bond() protocol. '
'Specifically, we encountered the following %s' % e)
# actually set the tables
try:
self._remove_node(where='/', name='topology')
except self.tables.NoSuchNodeError:
pass
data = json.dumps(topology_dict)
if not isinstance(data, bytes):
data = data.encode('ascii')
if self.tables.__version__ >= '3.0.0':
self._handle.create_array(where='/', name='topology', obj=[data])
else:
self._handle.createArray(where='/', name='topology', object=[data])
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
def load_mol2(filename):
"""Load a TRIPOS mol2 file from disk.
Parameters
----------
filename : str
Path to the prmtop file on disk.
Returns
-------
traj : md.Trajectory
The resulting topology, as an md.Topology object.
Notes
-----
This function should work on GAFF and sybyl style MOL2 files, but has
been primarily tested on GAFF mol2 files.
This function does NOT accept multi-structure MOL2 files!!!
The elements are guessed using GAFF atom types or via the atype string.
Examples
--------
>>> traj = md.load_mol2('mysystem.mol2')
"""
from mdtraj.core.trajectory import Trajectory
from mdtraj.core.topology import Topology, Single, Double, Triple, Aromatic, Amide
atoms, bonds = mol2_to_dataframes(filename)
atoms_mdtraj = atoms[["name", "resName"]].copy()
atoms_mdtraj["serial"] = atoms.index
#Figure out 1 letter element names
# IF this is a GAFF mol2, this line should work without issues
atoms_mdtraj["element"] = atoms.atype.map(gaff_elements)
# If this is a sybyl mol2, there should be NAN (null) values
if atoms_mdtraj.element.isnull().any():
# If this is a sybyl mol2, I think this works generally.
atoms_mdtraj["element"] = atoms.atype.apply(lambda x: x.strip(".")[0])
atoms_mdtraj["resSeq"] = np.ones(len(atoms), 'int')
atoms_mdtraj["chainID"] = np.ones(len(atoms), 'int')
bond_type_map = {
'1': Single,
'2': Double,
'3': Triple,
'am': Amide,
'ar': Aromatic
}
if bonds is not None:
bonds_mdtraj = bonds[["id0", "id1"]].values
offset = bonds_mdtraj.min() # Should this just be 1???
bonds_mdtraj -= offset
# Create the bond augment information
n_bonds = bonds_mdtraj.shape[0]
bond_augment = np.zeros([n_bonds, 2], dtype=float)
# Add bond type information
bond_augment[:, 0] = [float(bond_type_map[bond_value]) for bond_value in bonds["bond_type"].values]
# Add Bond "order" information, this is not known from Mol2 files
bond_augment[:, 1] = [0.0 for _ in range(n_bonds)]
# Augment array, dtype is cast to minimal representation of float
bonds_mdtraj = np.append(bonds_mdtraj, bond_augment, axis=-1)
else:
bonds_mdtraj = None
top = Topology.from_dataframe(atoms_mdtraj, bonds_mdtraj)
xyzlist = np.array([atoms[["x", "y", "z"]].values])
xyzlist /= 10.0 # Convert from angstrom to nanometer
traj = Trajectory(xyzlist, top)
return traj
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
"""
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
def topology(self, topology_object):
"""Set the topology in the file
Parameters
----------
topology_object : mdtraj.Topology
A topology object
"""
_check_mode(self.mode, ('w', 'a'))
# we want to be able to handle the simtk.openmm Topology object
# here too, so if it's not an mdtraj topology we'll just guess
# that it's probably an openmm topology and convert
if not isinstance(topology_object, Topology):
topology_object = Topology.from_openmm(topology_object)
try:
topology_dict = {
'chains': [],
'bonds': []
}
for chain in topology_object.chains:
chain_dict = {
'residues': [],
'index': int(chain.index)
}
for residue in chain.residues:
residue_dict = {
'index': int(residue.index),
'name': str(residue.name),
'atoms': [],
"resSeq": int(residue.resSeq),
"segmentID": str(residue.segment_id)
}
for atom in residue.atoms:
try:
element_symbol_string = str(atom.element.symbol)
except AttributeError:
element_symbol_string = ""
residue_dict['atoms'].append({
'index': int(atom.index),
'name': str(atom.name),
'element': element_symbol_string
})
chain_dict['residues'].append(residue_dict)
topology_dict['chains'].append(chain_dict)
for atom1, atom2 in topology_object.bonds:
topology_dict['bonds'].append([
int(atom1.index),
int(atom2.index)
])
except AttributeError as e:
raise AttributeError('topology_object fails to implement the'
'chains() -> residue() -> atoms() and bond() protocol. '
'Specifically, we encountered the following %s' % e)
# actually set the tables
try:
self._remove_node(where='/', name='topology')
except self.tables.NoSuchNodeError:
pass
data = json.dumps(topology_dict)
if not isinstance(data, bytes):
data = data.encode('ascii')
if self.tables.__version__ >= '3.0.0':
self._handle.create_array(where='/', name='topology', obj=[data])
else:
self._handle.createArray(where='/', name='topology', object=[data])
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 load_mol2(filename):
"""Load a TRIPOS mol2 file from disk.
Parameters
----------
filename : path-like
Path to the prmtop file on disk.
Returns
-------
traj : md.Trajectory
The resulting topology, as an md.Topology object.
Notes
-----
This function should work on GAFF and sybyl style MOL2 files, but has
been primarily tested on GAFF mol2 files.
This function does NOT accept multi-structure MOL2 files!!!
The elements are guessed using GAFF atom types or via the atype string.
Examples
--------
>>> traj = md.load_mol2('mysystem.mol2')
"""
from mdtraj.core.trajectory import Trajectory
from mdtraj.core.topology import Topology, Single, Double, Triple, Aromatic, Amide
atoms, bonds = mol2_to_dataframes(filename)
atoms_mdtraj = atoms[["name", "resName"]].copy()
atoms_mdtraj["serial"] = atoms.index
#Figure out 1 letter element names
# IF this is a GAFF mol2, this line should work without issues
atoms_mdtraj["element"] = atoms.atype.map(gaff_elements)
# If this is a sybyl mol2, there should be NAN (null) values
if atoms_mdtraj.element.isnull().any():
# If this is a sybyl mol2, I think this works generally.
# Argument x is being passed as a list with only one element.
def to_element(x):
if isinstance(x, (list, tuple)):
assert len(x) == 1
x = x[0]
if '.' in x: # orbital-hybridizations in SYBL
return x.split('.')[0]
try:
# check if we can convert the whole str to an Element,
# if not, we only pass the first letter.
from mdtraj.core.element import Element
Element.getBySymbol(x)
except KeyError:
return x[0]
return x
atoms_mdtraj["element"] = atoms.atype.apply(to_element)
# Check if elements inferred from atoms.atype are valid
# If not, try to infer elements from atoms.name
try:
atoms_mdtraj['element'].apply(elem.get_by_symbol)
except KeyError:
try:
atoms_mdtraj["element"] = atoms.name.apply(to_element)
atoms_mdtraj['element'].apply(elem.get_by_symbol)
except KeyError:
raise KeyError('Invalid element passed to atoms DataFrame')
atoms_mdtraj['resSeq'] = atoms['code']
atoms_mdtraj["chainID"] = np.ones(len(atoms), 'int')
bond_type_map = {
'1': Single,
'2': Double,
'3': Triple,
'am': Amide,
'ar': Aromatic
}
if bonds is not None:
bonds_mdtraj = bonds[["id0", "id1"]].values
offset = bonds_mdtraj.min() # Should this just be 1???
bonds_mdtraj -= offset
# Create the bond augment information
n_bonds = bonds_mdtraj.shape[0]
bond_augment = np.zeros([n_bonds, 2], dtype=float)
# Add bond type information
bond_augment[:, 0] = [
float(bond_type_map[str(bond_value)])
for bond_value in bonds["bond_type"].values
]
# Add Bond "order" information, this is not known from Mol2 files
bond_augment[:, 1] = [0.0 for _ in range(n_bonds)]
# Augment array, dtype is cast to minimal representation of float
bonds_mdtraj = np.append(bonds_mdtraj, bond_augment, axis=-1)
else:
bonds_mdtraj = None
top = Topology.from_dataframe(atoms_mdtraj, bonds_mdtraj)
xyzlist = np.array([atoms[["x", "y", "z"]].values])
xyzlist /= 10.0 # Convert from angstrom to nanometer
traj = Trajectory(xyzlist, top)
return traj
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)
class PDBTrajectoryFile(object):
"""Interface for reading and writing Protein Data Bank (PDB) files
Parameters
----------
filename : str
The filename to open. A path to a file on disk.
mode : {'r', 'w'}
The mode in which to open the file, either 'r' for read or 'w' for write.
force_overwrite : bool
If opened in write mode, and a file by the name of `filename` already
exists on disk, should we overwrite it?
Attributes
----------
positions : np.ndarray, shape=(n_frames, n_atoms, 3)
topology : mdtraj.Topology
closed : bool
Notes
-----
When writing pdb files, mdtraj follows the PDB3.0 standard as closely as
possible. During *reading* however, we try to be more lenient. For instance,
we will parse common nonstandard atom names during reading, and convert them
into the standard names. The replacement table used by mdtraj is at
{mdtraj_source}/formats/pdb/data/pdbNames.xml.
See Also
--------
mdtraj.load_pdb : High-level wrapper that returns a ``md.Trajectory``
"""
distance_unit = 'angstroms'
_residueNameReplacements = {}
_atomNameReplacements = {}
_chain_names = [chr(ord('A') + i) for i in range(26)]
def __init__(self, filename, mode='r', force_overwrite=True):
self._open = False
self._file = None
self._topology = None
self._positions = None
self._mode = mode
self._last_topology = None
if mode == 'r':
PDBTrajectoryFile._loadNameReplacementTables()
if _is_url(filename):
self._file = urlopen(filename)
if filename.lower().endswith('.gz'):
if six.PY3:
self._file = gzip.GzipFile(fileobj=self._file)
else:
self._file = gzip.GzipFile(fileobj=six.StringIO(
self._file.read()))
if six.PY3:
self._file = six.StringIO(self._file.read().decode('utf-8'))
else:
if filename.lower().endswith('.gz'):
self._file = gzip.open(filename, 'r')
self._file = six.StringIO(self._file.read().decode('utf-8'))
else:
self._file = open(filename, 'r')
self._read_models()
elif mode == 'w':
self._header_written = False
self._footer_written = False
if os.path.exists(filename) and not force_overwrite:
raise IOError('"%s" already exists' % filename)
self._file = open(filename, 'w')
else:
raise ValueError("invalid mode: %s" % mode)
self._open = True
def write(self, positions, topology, modelIndex=None, unitcell_lengths=None,
unitcell_angles=None, bfactors=None):
"""Write a PDB file to disk
Parameters
----------
positions : array_like
The list of atomic positions to write.
topology : mdtraj.Topology
The Topology defining the model to write.
modelIndex : {int, None}
If not None, the model will be surrounded by MODEL/ENDMDL records
with this index
unitcell_lengths : {tuple, None}
Lengths of the three unit cell vectors, or None for a non-periodic system
unitcell_angles : {tuple, None}
Angles between the three unit cell vectors, or None for a non-periodic system
bfactors : array_like, default=None, shape=(n_atoms,)
Save bfactors with pdb file. Should contain a single number for
each atom in the topology
"""
if not self._mode == 'w':
raise ValueError('file not opened for writing')
if not self._header_written:
self._write_header(unitcell_lengths, unitcell_angles)
self._header_written = True
if ilen(topology.atoms) != len(positions):
raise ValueError('The number of positions must match the number of atoms')
if np.any(np.isnan(positions)):
raise ValueError('Particle position is NaN')
if np.any(np.isinf(positions)):
raise ValueError('Particle position is infinite')
self._last_topology = topology # Hack to save the topology of the last frame written, allows us to output CONECT entries in write_footer()
if bfactors is None:
bfactors = ['{0:5.2f}'.format(0.0)] * len(positions)
else:
if (np.max(bfactors) >= 100) or (np.min(bfactors) <= -10):
raise ValueError("bfactors must be in (-10, 100)")
bfactors = ['{0:5.2f}'.format(b) for b in bfactors]
atomIndex = 1
posIndex = 0
if modelIndex is not None:
print("MODEL %4d" % modelIndex, file=self._file)
for (chainIndex, chain) in enumerate(topology.chains):
chainName = self._chain_names[chainIndex % len(self._chain_names)]
residues = list(chain.residues)
for (resIndex, res) in enumerate(residues):
if len(res.name) > 3:
resName = res.name[:3]
else:
resName = res.name
for atom in res.atoms:
if len(atom.name) < 4 and atom.name[:1].isalpha() and (atom.element is None or len(atom.element.symbol) < 2):
atomName = ' '+atom.name
elif len(atom.name) > 4:
atomName = atom.name[:4]
else:
atomName = atom.name
coords = positions[posIndex]
if atom.element is not None:
symbol = atom.element.symbol
else:
symbol = ' '
line = "ATOM %5d %-4s %3s %s%4d %s%s%s 1.00 %s %2s " % (
atomIndex % 100000, atomName, resName, chainName,
(res.resSeq) % 10000, _format_83(coords[0]),
_format_83(coords[1]), _format_83(coords[2]),
bfactors[posIndex], symbol)
assert len(line) == 80, 'Fixed width overflow detected'
print(line, file=self._file)
posIndex += 1
atomIndex += 1
if resIndex == len(residues)-1:
print("TER %5d %3s %s%4d" % (atomIndex, resName, chainName, res.resSeq), file=self._file)
atomIndex += 1
if modelIndex is not None:
print("ENDMDL", file=self._file)
def _write_header(self, unitcell_lengths, unitcell_angles, write_metadata=True):
"""Write out the header for a PDB file.
Parameters
----------
unitcell_lengths : {tuple, None}
The lengths of the three unitcell vectors, ``a``, ``b``, ``c``
unitcell_angles : {tuple, None}
The angles between the three unitcell vectors, ``alpha``,
``beta``, ``gamma``
"""
if not self._mode == 'w':
raise ValueError('file not opened for writing')
if unitcell_lengths is None and unitcell_angles is None:
return
if unitcell_lengths is not None and unitcell_angles is not None:
if not len(unitcell_lengths) == 3:
raise ValueError('unitcell_lengths must be length 3')
if not len(unitcell_angles) == 3:
raise ValueError('unitcell_angles must be length 3')
else:
raise ValueError('either unitcell_lengths and unitcell_angles'
'should both be spefied, or neither')
box = list(unitcell_lengths) + list(unitcell_angles)
assert len(box) == 6
if write_metadata:
print("REMARK 1 CREATED WITH MDTraj %s, %s" % (version.version, str(date.today())), file=self._file)
print("CRYST1%9.3f%9.3f%9.3f%7.2f%7.2f%7.2f P 1 1 " % tuple(box), file=self._file)
def _write_footer(self):
if not self._mode == 'w':
raise ValueError('file not opened for writing')
# Identify bonds that should be listed as CONECT records.
standardResidues = ['ALA', 'ASN', 'CYS', 'GLU', 'HIS', 'LEU', 'MET', 'PRO', 'THR', 'TYR',
'ARG', 'ASP', 'GLN', 'GLY', 'ILE', 'LYS', 'PHE', 'SER', 'TRP', 'VAL',
'A', 'G', 'C', 'U', 'I', 'DA', 'DG', 'DC', 'DT', 'DI', 'HOH']
conectBonds = []
if self._last_topology is not None:
for atom1, atom2 in self._last_topology.bonds:
if atom1.residue.name not in standardResidues or atom2.residue.name not in standardResidues:
conectBonds.append((atom1, atom2))
elif atom1.name == 'SG' and atom2.name == 'SG' and atom1.residue.name == 'CYS' and atom2.residue.name == 'CYS':
conectBonds.append((atom1, atom2))
if len(conectBonds) > 0:
# Work out the index used in the PDB file for each atom.
atomIndex = {}
nextAtomIndex = 0
prevChain = None
for chain in self._last_topology.chains:
for atom in chain.atoms:
if atom.residue.chain != prevChain:
nextAtomIndex += 1
prevChain = atom.residue.chain
atomIndex[atom] = nextAtomIndex
nextAtomIndex += 1
# Record which other atoms each atom is bonded to.
atomBonds = {}
for atom1, atom2 in conectBonds:
index1 = atomIndex[atom1]
index2 = atomIndex[atom2]
if index1 not in atomBonds:
atomBonds[index1] = []
if index2 not in atomBonds:
atomBonds[index2] = []
atomBonds[index1].append(index2)
atomBonds[index2].append(index1)
# Write the CONECT records.
for index1 in sorted(atomBonds):
bonded = atomBonds[index1]
while len(bonded) > 4:
print("CONECT%5d%5d%5d%5d" % (index1, bonded[0], bonded[1], bonded[2]), file=self._file)
del bonded[:4]
line = "CONECT%5d" % index1
for index2 in bonded:
line = "%s%5d" % (line, index2)
print(line, file=self._file)
print("END", file=self._file)
self._footer_written = True
@classmethod
def set_chain_names(cls, values):
"""Set the cycle of chain names used when writing PDB files
When writing PDB files, PDBTrajectoryFile translates each chain's
index into a name -- the name is what's written in the file. By
default, chains are named with the letters A-Z.
Parameters
----------
values : list
A list of chacters (strings of length 1) that the PDB writer will
cycle through to choose chain names.
"""
for item in values:
if not isinstance(item, six.string_types) and len(item) == 1:
raise TypeError('Names must be a single character string')
cls._chain_names = values
@property
def positions(self):
"""The cartesian coordinates of all of the atoms in each frame. Available when a file is opened in mode='r'
"""
return self._positions
@property
def topology(self):
"""The topology from this PDB file. Available when a file is opened in mode='r'
"""
return self._topology
@property
def unitcell_lengths(self):
"The unitcell lengths (3-tuple) in this PDB file. May be None"
return self._unitcell_lengths
@property
def unitcell_angles(self):
"The unitcell angles (3-tuple) in this PDB file. May be None"
return self._unitcell_angles
@property
def closed(self):
"Whether the file is closed"
return not self._open
def close(self):
"Close the PDB file"
if self._mode == 'w' and not self._footer_written:
self._write_footer()
if self._open:
self._file.close()
self._open = False
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)
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[0].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)
@staticmethod
def _loadNameReplacementTables():
"""Load the list of atom and residue name replacements."""
if len(PDBTrajectoryFile._residueNameReplacements) == 0:
tree = etree.parse(os.path.join(os.path.dirname(__file__), 'data', 'pdbNames.xml'))
allResidues = {}
proteinResidues = {}
nucleicAcidResidues = {}
for residue in tree.getroot().findall('Residue'):
name = residue.attrib['name']
if name == 'All':
PDBTrajectoryFile._parseResidueAtoms(residue, allResidues)
elif name == 'Protein':
PDBTrajectoryFile._parseResidueAtoms(residue, proteinResidues)
elif name == 'Nucleic':
PDBTrajectoryFile._parseResidueAtoms(residue, nucleicAcidResidues)
for atom in allResidues:
proteinResidues[atom] = allResidues[atom]
nucleicAcidResidues[atom] = allResidues[atom]
for residue in tree.getroot().findall('Residue'):
name = residue.attrib['name']
for id in residue.attrib:
if id == 'name' or id.startswith('alt'):
PDBTrajectoryFile._residueNameReplacements[residue.attrib[id]] = name
if 'type' not in residue.attrib:
atoms = copy(allResidues)
elif residue.attrib['type'] == 'Protein':
atoms = copy(proteinResidues)
elif residue.attrib['type'] == 'Nucleic':
atoms = copy(nucleicAcidResidues)
else:
atoms = copy(allResidues)
PDBTrajectoryFile._parseResidueAtoms(residue, atoms)
PDBTrajectoryFile._atomNameReplacements[name] = atoms
def _guess_element(self, atom_name, residue):
"Try to guess the element name"
upper = atom_name.upper()
if upper.startswith('CL'):
element = elem.chlorine
elif upper.startswith('NA'):
element = elem.sodium
elif upper.startswith('MG'):
element = elem.magnesium
elif upper.startswith('BE'):
element = elem.beryllium
elif upper.startswith('LI'):
element = elem.lithium
elif upper.startswith('K'):
element = elem.potassium
elif upper.startswith('ZN'):
element = elem.zinc
elif len(residue) == 1 and upper.startswith('CA'):
element = elem.calcium
# TJL has edited this. There are a few issues here. First,
# parsing for the element is non-trivial, so I do my best
# below. Second, there is additional parsing code in
# pdbstructure.py, and I am unsure why it doesn't get used
# here...
elif len(residue) > 1 and upper.startswith('CE'):
element = elem.carbon # (probably) not Celenium...
elif len(residue) > 1 and upper.startswith('CD'):
element = elem.carbon # (probably) not Cadmium...
elif residue.name in ['TRP', 'ARG', 'GLN', 'HIS'] and upper.startswith('NE'):
element = elem.nitrogen # (probably) not Neon...
elif residue.name in ['ASN'] and upper.startswith('ND'):
element = elem.nitrogen # (probably) not ND...
elif residue.name == 'CYS' and upper.startswith('SG'):
element = elem.sulfur # (probably) not SG...
else:
try:
element = elem.get_by_symbol(atom_name[0])
except KeyError:
try:
symbol = atom_name[0:2].strip().rstrip("AB0123456789").lstrip("0123456789")
element = elem.get_by_symbol(symbol)
except KeyError:
element = None
return element
@staticmethod
def _parseResidueAtoms(residue, map):
for atom in residue.findall('Atom'):
name = atom.attrib['name']
for id in atom.attrib:
map[atom.attrib[id]] = name
def __del__(self):
self.close()
def __enter__(self):
return self
def __exit__(self, *exc_info):
self.close()
def __len__(self):
"Number of frames in the file"
if str(self._mode) != 'r':
raise NotImplementedError('len() only available in mode="r" currently')
if not self._open:
raise ValueError('I/O operation on closed file')
return len(self._positions)
class PDBTrajectoryFile(object):
"""Interface for reading and writing Protein Data Bank (PDB) files
Parameters
----------
filename : str
The filename to open. A path to a file on disk.
mode : {'r', 'w'}
The mode in which to open the file, either 'r' for read or 'w' for write.
force_overwrite : bool
If opened in write mode, and a file by the name of `filename` already
exists on disk, should we overwrite it?
Attributes
----------
positions : np.ndarray, shape=(n_frames, n_atoms, 3)
topology : mdtraj.Topology
closed : bool
Notes
-----
When writing pdb files, mdtraj follows the PDB3.0 standard as closely as
possible. During *reading* however, we try to be more lenient. For instance,
we will parse common nonstandard atom names during reading, and convert them
into the standard names. The replacement table used by mdtraj is at
{mdtraj_source}/formats/pdb/data/pdbNames.xml.
See Also
--------
mdtraj.load_pdb : High-level wrapper that returns a ``md.Trajectory``
"""
distance_unit = 'angstroms'
_residueNameReplacements = {}
_atomNameReplacements = {}
_chain_names = [chr(ord('A') + i) for i in range(26)]
def __init__(self, filename, mode='r', force_overwrite=True):
self._open = False
self._file = None
self._topology = None
self._positions = None
self._mode = mode
self._last_topology = None
if mode == 'r':
PDBTrajectoryFile._loadNameReplacementTables()
if _is_url(filename):
self._file = urlopen(filename)
if filename.lower().endswith('.gz'):
if six.PY3:
self._file = gzip.GzipFile(fileobj=self._file)
else:
self._file = gzip.GzipFile(fileobj=six.StringIO(
self._file.read()))
if six.PY3:
self._file = six.StringIO(self._file.read().decode('utf-8'))
else:
self._file = open_maybe_zipped(filename, 'r')
self._read_models()
elif mode == 'w':
self._header_written = False
self._footer_written = False
self._file = open_maybe_zipped(filename, 'w', force_overwrite)
else:
raise ValueError("invalid mode: %s" % mode)
self._open = True
def write(self, positions, topology, modelIndex=None, unitcell_lengths=None,
unitcell_angles=None, bfactors=None):
"""Write a PDB file to disk
Parameters
----------
positions : array_like
The list of atomic positions to write.
topology : mdtraj.Topology
The Topology defining the model to write.
modelIndex : {int, None}
If not None, the model will be surrounded by MODEL/ENDMDL records
with this index
unitcell_lengths : {tuple, None}
Lengths of the three unit cell vectors, or None for a non-periodic system
unitcell_angles : {tuple, None}
Angles between the three unit cell vectors, or None for a non-periodic system
bfactors : array_like, default=None, shape=(n_atoms,)
Save bfactors with pdb file. Should contain a single number for
each atom in the topology
"""
if not self._mode == 'w':
raise ValueError('file not opened for writing')
if not self._header_written:
self._write_header(unitcell_lengths, unitcell_angles)
self._header_written = True
if ilen(topology.atoms) != len(positions):
raise ValueError('The number of positions must match the number of atoms')
if np.any(np.isnan(positions)):
raise ValueError('Particle position is NaN')
if np.any(np.isinf(positions)):
raise ValueError('Particle position is infinite')
self._last_topology = topology # Hack to save the topology of the last frame written, allows us to output CONECT entries in write_footer()
if bfactors is None:
bfactors = ['{0:5.2f}'.format(0.0)] * len(positions)
else:
if (np.max(bfactors) >= 100) or (np.min(bfactors) <= -10):
raise ValueError("bfactors must be in (-10, 100)")
bfactors = ['{0:5.2f}'.format(b) for b in bfactors]
atomIndex = 1
posIndex = 0
if modelIndex is not None:
print("MODEL %4d" % modelIndex, file=self._file)
for (chainIndex, chain) in enumerate(topology.chains):
chainName = self._chain_names[chainIndex % len(self._chain_names)]
residues = list(chain.residues)
for (resIndex, res) in enumerate(residues):
if len(res.name) > 3:
resName = res.name[:3]
else:
resName = res.name
for atom in res.atoms:
if len(atom.name) < 4 and atom.name[:1].isalpha() and (atom.element is None or len(atom.element.symbol) < 2):
atomName = ' '+atom.name
elif len(atom.name) > 4:
atomName = atom.name[:4]
else:
atomName = atom.name
coords = positions[posIndex]
if atom.element is not None:
symbol = atom.element.symbol
else:
symbol = ' '
line = "ATOM %5d %-4s %3s %s%4d %s%s%s 1.00 %s %2s " % (
atomIndex % 100000, atomName, resName, chainName,
(res.resSeq) % 10000, _format_83(coords[0]),
_format_83(coords[1]), _format_83(coords[2]),
bfactors[posIndex], symbol)
assert len(line) == 80, 'Fixed width overflow detected'
print(line, file=self._file)
posIndex += 1
atomIndex += 1
if resIndex == len(residues)-1:
print("TER %5d %3s %s%4d" % (atomIndex, resName, chainName, res.resSeq), file=self._file)
atomIndex += 1
if modelIndex is not None:
print("ENDMDL", file=self._file)
def _write_header(self, unitcell_lengths, unitcell_angles, write_metadata=True):
"""Write out the header for a PDB file.
Parameters
----------
unitcell_lengths : {tuple, None}
The lengths of the three unitcell vectors, ``a``, ``b``, ``c``
unitcell_angles : {tuple, None}
The angles between the three unitcell vectors, ``alpha``,
``beta``, ``gamma``
"""
if not self._mode == 'w':
raise ValueError('file not opened for writing')
if unitcell_lengths is None and unitcell_angles is None:
return
if unitcell_lengths is not None and unitcell_angles is not None:
if not len(unitcell_lengths) == 3:
raise ValueError('unitcell_lengths must be length 3')
if not len(unitcell_angles) == 3:
raise ValueError('unitcell_angles must be length 3')
else:
raise ValueError('either unitcell_lengths and unitcell_angles'
'should both be spefied, or neither')
box = list(unitcell_lengths) + list(unitcell_angles)
assert len(box) == 6
if write_metadata:
print("REMARK 1 CREATED WITH MDTraj %s, %s" % (version.version, str(date.today())), file=self._file)
print("CRYST1%9.3f%9.3f%9.3f%7.2f%7.2f%7.2f P 1 1 " % tuple(box), file=self._file)
def _write_footer(self):
if not self._mode == 'w':
raise ValueError('file not opened for writing')
# Identify bonds that should be listed as CONECT records.
standardResidues = ['ALA', 'ASN', 'CYS', 'GLU', 'HIS', 'LEU', 'MET', 'PRO', 'THR', 'TYR',
'ARG', 'ASP', 'GLN', 'GLY', 'ILE', 'LYS', 'PHE', 'SER', 'TRP', 'VAL',
'A', 'G', 'C', 'U', 'I', 'DA', 'DG', 'DC', 'DT', 'DI', 'HOH']
conectBonds = []
if self._last_topology is not None:
for atom1, atom2 in self._last_topology.bonds:
if atom1.residue.name not in standardResidues or atom2.residue.name not in standardResidues:
conectBonds.append((atom1, atom2))
elif atom1.name == 'SG' and atom2.name == 'SG' and atom1.residue.name == 'CYS' and atom2.residue.name == 'CYS':
conectBonds.append((atom1, atom2))
if len(conectBonds) > 0:
# Work out the index used in the PDB file for each atom.
atomIndex = {}
nextAtomIndex = 0
prevChain = None
for chain in self._last_topology.chains:
for atom in chain.atoms:
if atom.residue.chain != prevChain:
nextAtomIndex += 1
prevChain = atom.residue.chain
atomIndex[atom] = nextAtomIndex
nextAtomIndex += 1
# Record which other atoms each atom is bonded to.
atomBonds = {}
for atom1, atom2 in conectBonds:
index1 = atomIndex[atom1]
index2 = atomIndex[atom2]
if index1 not in atomBonds:
atomBonds[index1] = []
if index2 not in atomBonds:
atomBonds[index2] = []
atomBonds[index1].append(index2)
atomBonds[index2].append(index1)
# Write the CONECT records.
for index1 in sorted(atomBonds):
bonded = atomBonds[index1]
while len(bonded) > 4:
print("CONECT%5d%5d%5d%5d" % (index1, bonded[0], bonded[1], bonded[2]), file=self._file)
del bonded[:4]
line = "CONECT%5d" % index1
for index2 in bonded:
line = "%s%5d" % (line, index2)
print(line, file=self._file)
print("END", file=self._file)
self._footer_written = True
@classmethod
def set_chain_names(cls, values):
"""Set the cycle of chain names used when writing PDB files
When writing PDB files, PDBTrajectoryFile translates each chain's
index into a name -- the name is what's written in the file. By
default, chains are named with the letters A-Z.
Parameters
----------
values : list
A list of chacters (strings of length 1) that the PDB writer will
cycle through to choose chain names.
"""
for item in values:
if not isinstance(item, six.string_types) and len(item) == 1:
raise TypeError('Names must be a single character string')
cls._chain_names = values
@property
def positions(self):
"""The cartesian coordinates of all of the atoms in each frame. Available when a file is opened in mode='r'
"""
return self._positions
@property
def topology(self):
"""The topology from this PDB file. Available when a file is opened in mode='r'
"""
return self._topology
@property
def unitcell_lengths(self):
"The unitcell lengths (3-tuple) in this PDB file. May be None"
return self._unitcell_lengths
@property
def unitcell_angles(self):
"The unitcell angles (3-tuple) in this PDB file. May be None"
return self._unitcell_angles
@property
def closed(self):
"Whether the file is closed"
return not self._open
def close(self):
"Close the PDB file"
if self._mode == 'w' and not self._footer_written:
self._write_footer()
if self._open:
self._file.close()
self._open = False
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)
@staticmethod
def _loadNameReplacementTables():
"""Load the list of atom and residue name replacements."""
if len(PDBTrajectoryFile._residueNameReplacements) == 0:
tree = etree.parse(os.path.join(os.path.dirname(__file__), 'data', 'pdbNames.xml'))
allResidues = {}
proteinResidues = {}
nucleicAcidResidues = {}
for residue in tree.getroot().findall('Residue'):
name = residue.attrib['name']
if name == 'All':
PDBTrajectoryFile._parseResidueAtoms(residue, allResidues)
elif name == 'Protein':
PDBTrajectoryFile._parseResidueAtoms(residue, proteinResidues)
elif name == 'Nucleic':
PDBTrajectoryFile._parseResidueAtoms(residue, nucleicAcidResidues)
for atom in allResidues:
proteinResidues[atom] = allResidues[atom]
nucleicAcidResidues[atom] = allResidues[atom]
for residue in tree.getroot().findall('Residue'):
name = residue.attrib['name']
for id in residue.attrib:
if id == 'name' or id.startswith('alt'):
PDBTrajectoryFile._residueNameReplacements[residue.attrib[id]] = name
if 'type' not in residue.attrib:
atoms = copy(allResidues)
elif residue.attrib['type'] == 'Protein':
atoms = copy(proteinResidues)
elif residue.attrib['type'] == 'Nucleic':
atoms = copy(nucleicAcidResidues)
else:
atoms = copy(allResidues)
PDBTrajectoryFile._parseResidueAtoms(residue, atoms)
PDBTrajectoryFile._atomNameReplacements[name] = atoms
def _guess_element(self, atom_name, residue):
"Try to guess the element name"
upper = atom_name.upper()
if upper.startswith('CL'):
element = elem.chlorine
elif upper.startswith('NA'):
element = elem.sodium
elif upper.startswith('MG'):
element = elem.magnesium
elif upper.startswith('BE'):
element = elem.beryllium
elif upper.startswith('LI'):
element = elem.lithium
elif upper.startswith('K'):
element = elem.potassium
elif upper.startswith('ZN'):
element = elem.zinc
elif len(residue) == 1 and upper.startswith('CA'):
element = elem.calcium
# TJL has edited this. There are a few issues here. First,
# parsing for the element is non-trivial, so I do my best
# below. Second, there is additional parsing code in
# pdbstructure.py, and I am unsure why it doesn't get used
# here...
elif len(residue) > 1 and upper.startswith('CE'):
element = elem.carbon # (probably) not Celenium...
elif len(residue) > 1 and upper.startswith('CD'):
element = elem.carbon # (probably) not Cadmium...
elif residue.name in ['TRP', 'ARG', 'GLN', 'HIS'] and upper.startswith('NE'):
element = elem.nitrogen # (probably) not Neon...
elif residue.name in ['ASN'] and upper.startswith('ND'):
element = elem.nitrogen # (probably) not ND...
elif residue.name == 'CYS' and upper.startswith('SG'):
element = elem.sulfur # (probably) not SG...
else:
try:
element = elem.get_by_symbol(atom_name[0])
except KeyError:
try:
symbol = atom_name[0:2].strip().rstrip("AB0123456789").lstrip("0123456789")
element = elem.get_by_symbol(symbol)
except KeyError:
element = None
return element
@staticmethod
def _parseResidueAtoms(residue, map):
for atom in residue.findall('Atom'):
name = atom.attrib['name']
for id in atom.attrib:
map[atom.attrib[id]] = name
def __del__(self):
self.close()
def __enter__(self):
return self
def __exit__(self, *exc_info):
self.close()
def __len__(self):
"Number of frames in the file"
if str(self._mode) != 'r':
raise NotImplementedError('len() only available in mode="r" currently')
if not self._open:
raise ValueError('I/O operation on closed file')
return len(self._positions)
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_mol2(filename):
"""Load a TRIPOS mol2 file from disk.
Parameters
----------
filename : str
Path to the prmtop file on disk.
Returns
-------
traj : md.Trajectory
The resulting topology, as an md.Topology object.
Notes
-----
This function should work on GAFF and sybyl style MOL2 files, but has
been primarily tested on GAFF mol2 files.
This function does NOT accept multi-structure MOL2 files!!!
The elements are guessed using GAFF atom types or via the atype string.
Examples
--------
>>> traj = md.load_mol2('mysystem.mol2')
"""
from mdtraj.core.trajectory import Trajectory
from mdtraj.core.topology import Topology, Single, Double, Triple, Aromatic, Amide
atoms, bonds = mol2_to_dataframes(filename)
atoms_mdtraj = atoms[["name", "resName"]].copy()
atoms_mdtraj["serial"] = atoms.index
#Figure out 1 letter element names
# IF this is a GAFF mol2, this line should work without issues
atoms_mdtraj["element"] = atoms.atype.map(gaff_elements)
# If this is a sybyl mol2, there should be NAN (null) values
if atoms_mdtraj.element.isnull().any():
# If this is a sybyl mol2, I think this works generally.
# Argument x is being passed as a list with only one element.
def to_element(x):
if isinstance(x, (list, tuple)):
assert len(x) == 1
x = x[0]
if '.' in x: # orbital-hybridizations in SYBL
return x.split('.')[0]
try:
# check if we can convert the whole str to an Element,
# if not, we only pass the first letter.
from mdtraj.core.element import Element
Element.getBySymbol(x)
except KeyError:
return to_element(x[0])
return x
atoms_mdtraj["element"] = atoms.atype.apply(to_element)
atoms_mdtraj["resSeq"] = np.ones(len(atoms), 'int')
atoms_mdtraj["chainID"] = np.ones(len(atoms), 'int')
bond_type_map = {
'1': Single,
'2': Double,
'3': Triple,
'am': Amide,
'ar': Aromatic
}
if bonds is not None:
bonds_mdtraj = bonds[["id0", "id1"]].values
offset = bonds_mdtraj.min() # Should this just be 1???
bonds_mdtraj -= offset
# Create the bond augment information
n_bonds = bonds_mdtraj.shape[0]
bond_augment = np.zeros([n_bonds, 2], dtype=float)
# Add bond type information
bond_augment[:, 0] = [float(bond_type_map[str(bond_value)]) for bond_value in bonds["bond_type"].values]
# Add Bond "order" information, this is not known from Mol2 files
bond_augment[:, 1] = [0.0 for _ in range(n_bonds)]
# Augment array, dtype is cast to minimal representation of float
bonds_mdtraj = np.append(bonds_mdtraj, bond_augment, axis=-1)
else:
bonds_mdtraj = None
top = Topology.from_dataframe(atoms_mdtraj, bonds_mdtraj)
xyzlist = np.array([atoms[["x", "y", "z"]].values])
xyzlist /= 10.0 # Convert from angstrom to nanometer
traj = Trajectory(xyzlist, top)
return traj
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)
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[0].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
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
def topology(self, topology_object):
"""Set the topology in the file
Parameters
----------
topology_object : mdtraj.Topology
A topology object
"""
# we want to be able to handle the simtk.openmm Topology object
# here too, so if it's not an mdtraj topology we'll just guess
# that it's probably an openmm topology and convert
if not isinstance(topology_object, Topology):
topology_object = Topology.from_openmm(topology_object)
try:
topology_dict = {"chains": [], "bonds": []}
for chain in topology_object.chains:
chain_dict = {"residues": [], "index": int(chain.index)}
for residue in chain.residues:
residue_dict = {
"index": int(residue.index),
"name": str(residue.name),
"atoms": [],
"resSeq": int(residue.resSeq),
}
for atom in residue.atoms:
try:
element_symbol_string = str(atom.element.symbol)
except AttributeError:
element_symbol_string = ""
residue_dict["atoms"].append(
{"index": int(atom.index), "name": str(atom.name), "element": element_symbol_string}
)
chain_dict["residues"].append(residue_dict)
topology_dict["chains"].append(chain_dict)
for atom1, atom2 in topology_object.bonds:
topology_dict["bonds"].append([int(atom1.index), int(atom2.index)])
except AttributeError as e:
raise AttributeError(
"topology_object fails to implement the"
"chains() -> residue() -> atoms() and bond() protocol. "
"Specifically, we encountered the following %s" % e
)
# actually set the tables
try:
self._remove_node(where="/", name="topology")
except self.tables.NoSuchNodeError:
pass
data = json.dumps(topology_dict)
if not isinstance(data, bytes):
data = data.encode("ascii")
if self.tables.__version__ >= "3.0.0":
self._handle.create_array(where="/", name="topology", obj=[data])
else:
self._handle.createArray(where="/", name="topology", object=[data])
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)
