def test_ff_assignment_doesnt_change_topology(pdb3aid):
m = pdb3aid
protein = mdt.Molecule(m.get_atoms('protein'))
ligand = mdt.Molecule(m.get_atoms('unknown'))
ligff = mdt.create_ff_parameters(ligand, charges='gasteiger')
mdt.guess_histidine_states(protein)
mol = protein.combine(ligand)
ff = mdt.forcefields.DefaultAmber()
ff.add_ff(ligff)
mdready = ff.create_prepped_molecule(mol)
assert mdready.num_residues == mol.num_residues
assert mdready.num_chains == mol.num_chains
for c1, c2 in zip(mdready.chains, mol.chains):
assert c1.name == c2.name
assert c1.num_residues == c2.num_residues
assert c1.index == c2.index
for newr, oldr in zip(mdready.residues, mol.residues):
assert newr.index == oldr.index
if newr.resname == 'HIS':
assert oldr.resname in 'HIS HID HIE HIP'.split()
else:
assert newr.resname == oldr.resname
assert newr.pdbindex == oldr.pdbindex
assert newr.chain.index == oldr.chain.index
for atom in oldr:
assert atom.name in newr
def test_chain_rename(pdb3aid):
res1 = mdt.Molecule(pdb3aid.residues[3])
res2 = mdt.Molecule(pdb3aid.residues[4])
newmol = mdt.Molecule([res1, res2])
assert newmol.num_chains == 2
assert newmol.num_residues == 2
assert newmol.residues[0].name == res1.residues[0].name
assert newmol.residues[1].name == res2.residues[0].name
assert newmol.chains[0].name == 'A'
assert newmol.chains[1].name == 'B'
def biopy_to_mol(struc):
"""Convert a biopython PDB structure to an MDT molecule.
Because Biopython doesn't assign bonds, assign connectivity using templates.
Args:
struc (Bio.PDB.Structure.Structure): Biopython PDB structure to convert
Returns:
moldesign.Molecule: converted molecule
"""
# TODO: assign bonds using 1) CONECT records, 2) residue templates, 3) distance
newatoms = []
for chain in struc.get_chains():
tmp, pdbidx, pdbid = chain.get_full_id()
newchain = mdt.Chain(pdbname=pdbid.strip())
for residue in chain.get_residues():
newresidue = mdt.Residue(pdbname=residue.resname.strip(),
pdbindex=residue.id[1])
newchain.add(newresidue)
for atom in residue.get_atom():
newatom = mdt.Atom(element=atom.element,
name=atom.get_name(),
pdbname=atom.get_name(),
pdbindex=atom.get_serial_number())
newatom.position = atom.coord * u.angstrom
newresidue.add(newatom)
newatoms.append(newatom)
return mdt.Molecule(newatoms, name=struc.get_full_id()[0])
def combine_molecules(mdtfile1, mdtfile2):
import moldesign as mdt
m1 = mdt.read(mdtfile1)
m2 = mdt.read(mdtfile2)
newmol = mdt.Molecule(m1.atoms + m2.atoms)
newmol.write('out.pkl')
def _build_mm_system(self):
# Set up the MM subsystem - includes all atoms,
# but we remove all FF terms except LJ
self.mmmol = mdt.Molecule(self.mol)
for atom, subatom in zip(self.mol.atoms, self.mmmol.atoms):
subatom.molecule_atom = atom
atom.props.mm_child = subatom
self.mmmol.set_energy_model(self.mm_model)
# Remove charges and bonds for QM atoms
ff = self.mm_model.params.ff
forcefield = ff.get_parameters(self.mmmol)
def prune(termlist): # remove intra-qm region forces
# TODO: what about LJ forces???
newterms = []
for term in termlist:
subsystems = set([atom.molecule_atom.props.subsystem for atom in termlist.atom])
subsystems = list(subsystems)
if len(subsystems) == 1 and subsystems[0] == 'mm':
newterms.append(subsystems)
elif len(subsystems) > 1:
raise ValueError('Bond crosses boundary: %s' % term)
forcefield.bonds = prune(forcefield.bonds)
forcefield.angles = prune(forcefield.angles)
forcefield.dihedrals = prune(forcefield.dihedrals)
forcefield.impropers = prune(forcefield.impropers)
def test_initialization_charges():
a1 = mdt.Atom('Na', formal_charge=-1)
mol = mdt.Molecule([a1])
assert mol.charge == -1 * u.q_e
with pytest.raises(TypeError):
mdt.Atom(
'H', charge=3
) # it needs to be "formal_charge" to distinguish from partial charge
m2 = mdt.Molecule([a1], charge=-1)
assert m2.charge == -1 * u.q_e
m2 = mdt.Molecule([a1], charge=-3 * u.q_e)
# TODO: test for warning
assert m2.charge == -3 * u.q_e
def restore_topology(mol, topo):
""" Restores chain IDs and residue indices (these are stripped by some methods)
Args:
mol (mdt.Molecule): molecule to restore topology to
topo (mdt.Molecule): reference topology
Returns:
mdt.Molecule: a copy of ``mol`` with a restored topology
"""
import moldesign as mdt
assert mol.num_residues == topo.num_residues
assert mol.num_chains == 1
chain_map = {}
for chain in topo.chains:
chain_map[chain] = mdt.Chain(name=chain.name)
for res, refres in zip(mol.residues, topo.residues):
if refres.resname != res.resname:
print((
'INFO: Residue #{res.index} residue code changed from "{refres.resname}"'
' to "{res.resname}".').format(res=res, refres=refres))
res.pdbindex = refres.pdbindex
res.name = refres.name
res.chain = chain_map[refres.chain]
return mdt.Molecule(mol.atoms)
def split_chains(mol, distance_threshold=1.75*u.angstrom):
""" Split a molecule's chains into unbroken biopolymers and groups of non-polymers
This function is non-destructive - the passed molecule will not be modified.
Specifically, this function will:
- Split any chain with non-contiguous biopolymeric pieces into single, contiguous polymers
- Remove any solvent molecules from a chain into their own chain
- Isolate ligands from each chain into their own chains
Args:
mol (mdt.Molecule): Input molecule
distance_threshold (u.Scalar[length]): if not ``None``, the maximum distance between
adjacent residues for which we consider them "contiguous". For PDB data, values greater
than 1.4 Angstrom are eminently reasonable; the default threshold of 1.75 Angstrom is
purposefully set to be extremely cautious (and still much lower than the distance to
the *next* nearest neighbor, generally around 2.5 Angstrom)
Returns:
mdt.Molecule: molecule with separated chains
"""
tempmol = mol.copy()
def bonded(r1, r2):
if r2 not in r1.bonded_residues:
return False
if distance_threshold is not None and r1.distance(r2) > distance_threshold:
return False
return True
def addto(chain, res):
res.chain = None
chain.add(res)
allchains = [mdt.Chain(tempmol.chains[0].name)]
for chain in tempmol.chains:
chaintype = chain.residues[0].type
solventchain = mdt.Chain(None)
ligandchain = mdt.Chain(None)
for ires, residue in enumerate(chain.residues):
if residue.type == 'unknown':
thischain = ligandchain
elif residue.type in ('water', 'solvent', 'ion'):
thischain = solventchain
else:
assert residue.type == chaintype
if ires != 0 and not bonded(residue.prev_residue, residue):
allchains.append(mdt.Chain(None))
thischain = allchains[-1]
addto(thischain, residue)
for c in (solventchain, ligandchain):
if c.num_atoms > 0:
allchains.append(c)
return mdt.Molecule(allchains)
def h2():
atom1 = mdt.Atom('H')
atom1.x = 0.5 * u.angstrom
atom2 = mdt.Atom(atnum=1)
atom2.position = [-0.5, 0.0, 0.0] * u.angstrom
h2 = mdt.Molecule([atom1, atom2], name='h2')
atom1.bond_to(atom2, 1)
return h2
def test_charge_from_number(h2):
h2plus = mdt.Molecule(h2, charge=1)
assert h2plus.charge == 1 * u.q_e
h2plus.charge = 2 * u.q_e
assert h2plus.charge == 2 * u.q_e
h2plus.charge = 3
assert h2plus.charge == 3 * u.q_e
def build_assembly(mol, assembly_name):
""" Create biological assembly using a bioassembly specification.
This routine builds a biomolecular assembly using the specification from a PDB header (if
present, this data can be found in the "REMARK 350" lines in the PDB file). Assemblies are
author-assigned structures created by copying, translating, and rotating a subset of the
chains in the PDB file.
See Also:
http://pdb101.rcsb.org/learn/guide-to-understanding-pdb-data/biological-assemblies
Args:
mol (moldesign.Molecule): Molecule with assembly data (assembly data will be created by the
PDB parser at ``molecule.properties.bioassembly``)
assembly_name (str OR int): id of the biomolecular assembly to build.
Returns:
mol (moldesign.Molecule): molecule containing the complete assembly
Raises:
AttributeError: If the molecule does not contain any biomolecular assembly data
KeyError: If the specified assembly is not present
"""
if isinstance(assembly_name, int): assembly_name = str(assembly_name)
if 'bioassemblies' not in mol.properties:
raise AttributeError(
'This molecule does not contain any biomolecular assembly data')
try:
asm = mol.properties.bioassemblies[assembly_name]
except KeyError:
raise KeyError(
('The specified assembly name ("%s") was not found. The following '
'assemblies are present: %s') %
(assembly_name, ', '.join(mol.properties.bioassemblies.keys())))
# Make sure each chain gets a unique name - up to all the letters in the alphabet, anyway
used_chain_names = set()
alpha = iter(string.ascii_uppercase)
# Create the new molecule by copying, transforming, and renaming the original chains
all_atoms = moldesign.molecules.atomcollections.AtomList()
for i, t in enumerate(asm.transforms):
for chain_name in asm.chains:
chain = mol.chains[chain_name].copy()
chain.transform(t)
while chain.name in used_chain_names:
chain.name = alpha.next()
used_chain_names.add(chain.name)
chain.pdbname = chain.pdbindex = chain.name
all_atoms.extend(chain.atoms)
newmol = mdt.Molecule(all_atoms,
name="%s (bioassembly %s)" %
(mol.name, assembly_name))
return newmol
def test_solvate_protein_padding(pdb1yu8):
newmol = mdt.add_water(pdb1yu8, padding=5.0 * u.angstrom)
assert newmol.num_atoms > pdb1yu8.num_atoms
oldmol = mdt.Molecule(newmol.residues[:pdb1yu8.num_residues])
assert oldmol.same_topology(pdb1yu8, verbose=True)
np.testing.assert_allclose(pdb1yu8.positions.value_in(u.angstrom),
oldmol.positions.value_in(u.angstrom),
atol=1e-3)
def linear():
expected = {'C1': 1,
'Cinf_v': 1}
a1 = mdt.Atom(1)
a1.position = [-1, 0, 0] * u.angstrom
a2 = mdt.Atom(2)
a2.position = [1, 0, 0] * u.angstrom
return mdt.Molecule([a1, a2], name='linear_h2'), expected, True
def get_input_file(self):
if len(self.mol.atoms) <= 250:
fmt = 'sdf'
else:
fmt = 'pdb'
if not hasattr(self.mol, 'write'):
writemol = mdt.Molecule(self.mol)
else:
writemol = self.mol
instring = writemol.write(format=fmt)
return instring, fmt
def __init__(self, mol, unit_system=None, first_frame=False):
self._init = True
self.info = "Trajectory"
self.frames = []
self.mol = mol
self.unit_system = utils.if_not_none(unit_system, mdt.units.default)
self._property_keys = None
self._tempmol = mdt.Molecule(self.mol.atoms, copy_atoms=True)
self._tempmol.dynamic_dof = self.mol.dynamic_dof
self._viz = None
self.atoms = [_TrajAtom(self, i) for i in xrange(self.mol.num_atoms)]
if first_frame: self.new_frame()
def planar():
expected = {'C1': 1,
'Cs': 1}
a1 = mdt.Atom(1)
a1.position = [-1, 0, 0] * u.angstrom
a2 = mdt.Atom(2)
a2.position = [0.9, 0.2, 0] * u.angstrom
a3 = mdt.Atom(3)
a3.position = [-0.9, 3.2, 0] * u.angstrom
return mdt.Molecule([a1, a2, a3], name='planar_h3'), expected, True
def test_set_hybridization_and_saturate():
# Creates just the carbons of ethylene, expects the routine to figure out the rest
atom1 = mdt.Atom(6)
atom2 = mdt.Atom(6)
atom2.x = 1.35 * u.angstrom
atom1.bond_to(atom2, 1)
mol = mdt.Molecule([atom1, atom2])
newmol = mdt.set_hybridization_and_saturate(mol)
pytest.xfail('This is apparently broken')
assert newmol.num_atoms == 6
assert newmol.atoms[0].bond_graph[atom1] == 2
assert len(newmol.get_atoms(atnum=1)) == 4
def test_copy_breaks_link(h2):
h2copy = mdt.Molecule(h2)
h2.atoms[0].y = 4.0 * u.bohr
assert h2copy.atoms[0].y == 0.0 * u.angstrom
np.testing.assert_almost_equal(h2.positions[0, 1].value_in(u.bohr), 4.0, 7)
assert h2copy.positions[0, 1] == 0.0 * u.bohr
h2copy.momenta[1, 0] = 2.0 * u.default.momentum
np.testing.assert_almost_equal(
h2copy.atoms[1].px.value_in(u.default.momentum), 2.0, 7)
assert h2.momenta[1, 0] == 0.0 * u.default.momentum
assert h2.atoms[1].px == 0.0 * u.default.momentum
def combine(self, *others):
""" Create a new molecule from a group of other AtomContainers
Notes:
- Chain IDs and sequence numbers are automatically assigned if they are missing
- Chains will be renamed to prevent chain ID clashes
- Residue resnames are not changed.
Args:
*others (AtomContainer or AtomList or List[moldesign.Atom]):
Returns:
mdt.Molecule: a new Molecule that's the union of this structure with all
others. Chains will be renamed as necessary to avoid clashes.
"""
new_atoms = []
charge = 0
names = []
chain_names = collections.OrderedDict(
(x, None) for x in string.ascii_uppercase)
taken_names = set()
seen_chains = set()
for obj in itertools.chain([self], others):
objatoms = mdt.helpers.get_all_atoms(obj).copy()
for atom in objatoms:
chain = atom.chain
if chain not in seen_chains:
seen_chains.add(chain)
if chain.pdbindex is None or chain.pdbindex in taken_names:
chain.pdbindex = chain.name = next(
iter(chain_names.keys()))
chain_names.pop(chain.pdbindex, None)
taken_names.add(chain.pdbindex)
new_atoms.extend(objatoms)
charge += getattr(obj, 'charge', 0 * u.q_e)
if hasattr(obj, 'name'):
names.append(obj.name)
elif objatoms[0].molecule is not None:
names.append('%d atoms from %s' %
(len(objatoms), objatoms[0].molecule.name))
else:
names.append('list of %d unowned atoms' % len(objatoms))
return mdt.Molecule(new_atoms,
copy_atoms=True,
charge=charge,
name='%s extended with %d atoms' %
(self.name, len(new_atoms) - self.num_atoms),
metadata=utils.DotDict(description='Union of %s' %
', '.join(names)))
def __init__(self, mol, unit_system=None, first_frame=False, name=None):
self._init = True
self.info = "Trajectory"
self.frames = []
self.mol = mol
self.unit_system = utils.if_not_none(unit_system, mdt.units.default)
self.properties = utils.DotDict()
self._tempmol = mdt.Molecule(self.mol.atoms, copy_atoms=True)
self._tempmol.dynamic_dof = self.mol.dynamic_dof
self._viz = None
self._atoms = None
self.name = utils.if_not_none(name, 'untitled')
if first_frame: self.new_frame()
def test_copying_doesnt_corrupt_original_h2_harmonic(h2_harmonic):
mol = h2_harmonic
integ = mol.integrator
model = mol.energy_model
residue = mol.residues[0]
chain = list(mol.chains)[0]
m2 = mdt.Molecule(mol)
assert integ is mol.integrator
assert model is mol.energy_model
assert len(mol.chains) == 1
assert len(mol.residues) == 1
assert residue == mol.residues[0]
assert chain == list(mol.chains)[0]
def _build_qm_system(self):
# Set up the QM system
self.qmmol = mdt.Molecule(self.mol)
for atom, subatom in zip(self.mol.atoms, self.qmmol.atoms):
subatom.molecule_atom = atom
atom.props.qm_child = subatom
# Make the MM atoms point charges
point_charges = {atom: atom.props.mm_child.ff.charge for atom in self.qmmol.atoms
if atom.molecule_atom.params.subsystem == 'mm'}
self.qm_model.params.point_charges = point_charges
self.qmmol.set_energy_model(self.qm_model)
self._subatoms = self.qmmol.atoms + self.mmol.atoms
def _setup_qm_subsystem(self):
qmmol = mdt.Molecule(self.mol)
self.mol.ff.copy_to(qmmol)
self.qm_link_atoms = mdt.helpers.qmmm.create_link_atoms(
self.mol, self.qm_atoms)
if self.qm_link_atoms:
raise ValueError('The %s model does not support link atoms' %
self.__class__.__name__)
qmmol.set_energy_model(self.params.qm_model)
qmmol.energy_model.params.qm_atom_indices = self.params.qm_atom_indices
return qmmol
def test_numeric_residue_name_1PYN(request, mol):
""" The ligand in this residue is named "941", which causes a little trickiness
"""
import parmed
mol = request.getfixturevalue(mol)
ligand = mdt.Molecule(mol.residues[283])
params = mdt.create_ff_parameters(ligand, charges='gasteiger')
params._file_list['mol.lib'].put('/tmp/tmp.lib')
contents = parmed.load_file('/tmp/tmp.lib')
assert len(contents) == 1
assert list(contents.keys())[0] == '941'
def test_numeric_residue_name_1PYN():
""" The ligand in this residue is named "941", which causes a little trickiness
"""
import parmed
mol = mdt.read(get_data_path('1pyn.pdb'))
ligand = mdt.Molecule(mol.residues[283])
params = mdt.parameterize(ligand, charges='gasteiger')
params.lib.put('/tmp/tmp.lib')
contents = parmed.load_file('/tmp/tmp.lib')
assert len(contents) == 1
assert contents.keys()[0] == '941'
def finish_job(job):
mol = mdt.fileio.read_pdb(job.get_output('helix.pdb').open(), assign_ccd_bonds=False)
if mol.num_chains == 1:
assert mol.num_residues % 2 == 0
oldchain = mol.chains[0]
oldchain.name = oldchain.pdbindex = oldchain.pdbname = 'A'
newchain = mdt.Chain('B')
for residue in mol.residues[mol.num_residues//2:]:
residue.chain = newchain
mol = mdt.Molecule(mol)
mdt.helpers.assign_biopolymer_bonds(mol)
mol.name = '%s-DNA Helix: %s' % (helix_type.upper(), sequence)
return mol
def _setup_qm_subsystem(self):
""" QM subsystem for mechanical embedding is the QM atoms + any link atoms
"""
qm_atoms = [
self.mol.atoms[iatom] for iatom in self.params.qm_atom_indices
]
self.qm_link_atoms = mdt.helpers.qmmm.create_link_atoms(
self.mol, qm_atoms)
qmmol = mdt.Molecule(qm_atoms + self.qm_link_atoms,
name='%s QM subsystem' % self.mol.name)
for real_atom, qm_atom in zip(self.qm_atoms, qmmol.atoms):
qm_atom.metadata.real_atom = real_atom
qmmol.set_energy_model(self.params.qm_model)
return qmmol
def test_mechanical_embedding_wfn(h2_h2_mechanical_embedding_rhf):
mol = h2_h2_mechanical_embedding_rhf
mol.calculate()
qmprops = mol.properties.qmprops
mmprops = mol.properties.mmprops
h2_qm = mdt.Molecule(mol.residues[0])
h2_qm.set_energy_model(mdt.models.RHF, basis='sto-3g')
h2_qm.calculate()
assert abs(h2_qm.potential_energy -
qmprops.potential_energy) < 1e-8 * u.hartree
helpers.assert_almost_equal(h2_qm.wfn.fock_ao, qmprops.wfn.fock_ao)
assert qmprops.potential_energy + mmprops.potential_energy == mol.potential_energy
def _prep_for_tleap(mol):
""" Returns a modified *copy* that's been modified for input to tleap
Makes the following modifications:
1. Reassigns all residue IDs
2. Assigns tleap-appropriate cysteine resnames
"""
change = False
clean = mdt.Molecule(mol.atoms)
for residue in clean.residues:
residue.pdbindex = residue.index + 1
if residue.resname == 'CYS': # deal with cysteine states
if 'SG' not in residue.atoms or 'HG' in residue.atoms:
continue # sulfur's missing, we'll let tleap create it
else:
sulfur = residue.atoms['SG']
if sulfur.formal_charge == -1 * u.q_e:
residue.resname = 'CYM'
change = True
continue
# check for a reasonable hybridization state
if sulfur.formal_charge != 0 or sulfur.num_bonds not in (1, 2):
raise ValueError("Unknown sulfur hybridization state for %s" %
sulfur)
# check for a disulfide bond
for otheratom in sulfur.bonded_atoms:
if otheratom.residue is not residue:
if otheratom.name != 'SG' or otheratom.residue.resname not in (
'CYS', 'CYX'):
raise ValueError(
'Unknown bond from cysteine sulfur (%s)' % sulfur)
# if we're here, this is a cystine with a disulfide bond
print(
'INFO: disulfide bond detected. Renaming %s from CYS to CYX'
% residue)
sulfur.residue.resname = 'CYX'
clean._rebuild_from_atoms()
return clean
def parmed_to_mdt(pmdmol):
""" Convert parmed Structure to MDT Structure
Args:
pmdmol (parmed.Structure): parmed structure to convert
Returns:
mdt.Molecule: converted molecule
"""
atoms = collections.OrderedDict()
residues = {}
chains = {}
masses = [pa.mass for pa in pmdmol.atoms] * u.dalton
positions = [[pa.xx, pa.xy, pa.xz] for pa in pmdmol.atoms] * u.angstrom
for iatom, patm in enumerate(pmdmol.atoms):
if patm.residue.chain not in chains:
chains[patm.residue.chain] = mdt.Chain(pdbname=patm.residue.chain)
chain = chains[patm.residue.chain]
if patm.residue not in residues:
residues[patm.residue] = mdt.Residue(resname=patm.residue.name,
pdbindex=patm.residue.number)
residues[patm.residue].chain = chain
chain.add(residues[patm.residue])
residue = residues[patm.residue]
atom = mdt.Atom(name=patm.name,
atnum=patm.atomic_number,
pdbindex=patm.number,
mass=masses[iatom])
atom.position = positions[iatom]
atom.residue = residue
residue.add(atom)
assert patm not in atoms
atoms[patm] = atom
for pbnd in pmdmol.bonds:
atoms[pbnd.atom1].bond_to(atoms[pbnd.atom2], int(pbnd.order))
mol = mdt.Molecule(list(atoms.values()),
metadata=_get_pdb_metadata(pmdmol))
return mol