def delete_shell(core_mol, del_mol, cut_off, in_out='in'):
"""
This function deletes molecules present in the passed argument
del_mol that are far (in_out=out) or close (in_out=in) than the
selected cutoff distance (in A) from the passed molecules core_mol
Parameters:
-----------
core_mol: OEMol molecule
The core molecules
del_mol: OEMol molecule
The molecules to be deleted if their distances from the core_mol
molecules are greater or closer that the selected cutoff distance
cut_off: python float number
The threshold distance in A used to mark atom for deletion
in_out: python string
A flag used to select if delete molecules far or close than
the cutoff distance from the core_mol
Return:
-------
reset_del: copy of del_mol where atoms have been deleted with
reset atom indexes
"""
if in_out not in ['in', 'out']:
raise ValueError(
"The passed in_out parameter is not recognized: {}".format(in_out))
# Copy the passed molecule to delete in
to_del = oechem.OEMol(del_mol)
# Create a OE bit vector mask for each atoms of the
# molecule to delete
bv = oechem.OEBitVector(to_del.GetMaxAtomIdx())
bv.NegateBits()
# Create the Nearest neighbours
nn = oechem.OENearestNbrs(to_del, cut_off)
for nbrs in nn.GetNbrs(core_mol):
# bv.SetBitOff(nbrs.GetBgn().GetIdx())
for atom in oechem.OEGetResidueAtoms(nbrs.GetBgn()):
bv.SetBitOff(atom.GetIdx())
# Invert selection mask
if in_out == 'in':
bv.NegateBits()
pred = oechem.OEAtomIdxSelected(bv)
for atom in to_del.GetAtoms(pred):
to_del.DeleteAtom(atom)
# It is necessary to reset the atom indexes of the molecule with
# delete atoms to avoid possible mismatching
reset_del = oechem.OEMol(to_del)
return reset_del
def strip_water_ions(in_system):
"""
This function remove waters and ions molecules
from the input system
Parameters:
----------
in_system : oechem.OEMol
The bio-molecular system to clean
opt: python dictionary
The system option
Output:
-------
clean_system : oechem.OEMol
The cleaned system
"""
# Copy the input system
system = in_system.CreateCopy()
# Create a bit vector mask
bv = oechem.OEBitVector(system.GetMaxAtomIdx())
bv.NegateBits()
# Create a Hierarchical View of the protein system
hv = oechem.OEHierView(
system,
oechem.OEAssumption_BondedResidue + oechem.OEAssumption_ResPerceived)
# Looping over the system residues
for chain in hv.GetChains():
for frag in chain.GetFragments():
for hres in frag.GetResidues():
res = hres.GetOEResidue()
# Check if a residue is a mono atomic ion
natoms = 0
for at in hres.GetAtoms():
natoms += 1
# Set the atom bit mask off
if oechem.OEGetResidueIndex(
res) == oechem.OEResidueIndex_HOH or natoms == 1:
# Set Bit mask
atms = hres.GetAtoms()
for at in atms:
bv.SetBitOff(at.GetIdx())
# Extract the system without waters or ions
pred = oechem.OEAtomIdxSelected(bv)
clean_system = oechem.OEMol()
oechem.OESubsetMol(clean_system, system, pred)
return clean_system
def check_shell(core_mol, check_mol, cutoff):
"""
This function checks if at least one atomic distance from the passed
check_mol molecule to the core_mol molecule is less than the selected
cutoff distance in A.
Parameters:
-----------
core_mol: OEMol molecule
The core molecule
check_mol: OEMol molecule
The molecule to be checked if inside or outside a shell
surrounding the core_mole with radius equal to the cutoff
threshold
cut_off: python float number
The threshold distance in A used to mark atom inside or outside
the shell
Return:
-------
in_out: python boolean
True if at least one of check_mol atom distance from core_mole
is less than the selected cutoff threshold
"""
# Create a OE bit vector mask for each atoms of the
# molecule to be checked
bv = oechem.OEBitVector(check_mol.GetMaxAtomIdx())
# Create the Nearest neighbours
nn = oechem.OENearestNbrs(check_mol, cutoff)
# Check neighbours setting the atom bit mask
for nbrs in nn.GetNbrs(core_mol):
bv.SetBitOn(nbrs.GetBgn().GetIdx())
# Create predicate based on the atom bit mask
pred = oechem.OEAtomIdxSelected(bv)
# Checking flag
in_out = False
# If just one chem_mol atom is inside the cutoff distance return True
for atom in check_mol.GetAtoms(pred):
in_out = True
break
return in_out
def around(dist, ls):
"""
This function select atom not far than the threshold distance from
the current selection. The threshold distance is in Angstrom
selection can be:
mask = '5.0 around ligand'
"""
# at = system.GetAtom(oechem.OEHasAtomIdx(idx))
# Atom set selection
atom_set_around = set()
# Create a OE bit vector mask for each atoms
bv_around = oechem.OEBitVector(system.GetMaxAtomIdx())
# Set the mask atom
for at in system.GetAtoms():
if at.GetIdx() in ls:
bv_around.SetBitOn(at.GetIdx())
# Predicate
pred = oechem.OEAtomIdxSelected(bv_around)
# Create the system molecule based on the atom mask
molecules = oechem.OEMol()
oechem.OESubsetMol(molecules, system, pred)
# Create the Nearest neighbours
nn = oechem.OENearestNbrs(system, float(dist))
for nbrs in nn.GetNbrs(molecules):
for atom in oechem.OEGetResidueAtoms(nbrs.GetBgn()):
if atom.GetIdx() in ls:
continue
atom_set_around.add(atom.GetIdx())
return atom_set_around
def DropLigandFromProtein(prot, lig):
"""delete atoms from the protein w/same coords as the ligand
as well as any waters"""
approximatelyTheSame = 0.05
nn = oechem.OENearestNbrs(prot, approximatelyTheSame)
# mark ligand atoms for deletion
bv = oechem.OEBitVector(prot.GetMaxAtomIdx())
for nbrs in nn.GetNbrs(lig):
r1 = oechem.OEAtomGetResidue(nbrs.GetBgn())
r2 = oechem.OEAtomGetResidue(nbrs.GetEnd())
if r1.GetModelNumber() == r2.GetModelNumber():
bv.SetBitOn(nbrs.GetBgn().GetIdx())
# mark waters for deletion too
for atom in prot.GetAtoms():
res = oechem.OEAtomGetResidue(atom)
if oechem.OEGetResidueIndex(res) == oechem.OEResidueIndex_HOH:
bv.SetBitOn(atom.GetIdx())
pred = oechem.OEAtomIdxSelected(bv)
for atom in prot.GetAtoms(pred):
prot.DeleteAtom(atom)
def oesolvate(solute,
density=1.0,
padding_distance=10.0,
distance_between_atoms=2.5,
solvents='tip3p',
molar_fractions='1.0',
geometry='box',
close_solvent=True,
salt='[Na+], [Cl-]',
salt_concentration=0.0,
neutralize_solute=True,
verbose=False,
return_components=False,
**kargs):
"""
This function solvates the passed solute in a cubic box or a sphere by using Packmol. Packmol
creates an initial point for molecular dynamics simulations by packing molecule in defined regions
of space. For additional info:
http://www.ime.unicamp.br/~martinez/packmol/home.shtml
The geometry volume is estimated by the using the padding parameter and the solute size.
The number of solvent molecules is calculated by using the specified density and volume.
Solvent molecules are specified as comma separated smiles strings. The molar fractions
of each solvent molecule are specified in a similar fashion. By default if the solute is
charged counter ions are added to neutralize it
Parameters:
-----------
solute: OEMol molecule
The solute to solvate
density: float
The solution density in g/ml
padding_distance: float
The largest dimension of the solute (along the x, y, or z axis) is determined (in A),
and a cubic box of size (largest dimension)+2*padding is used
distance_between_atoms: float
The minimum distance between atoms in A
solvents: python string
A comma separated smiles string or keywords for the solvent molecules.
Special water models can be selected by using the keywords:
tip3p for TIP3P water model geometry
molar_fractions: python string
A comma separated molar fraction string of the solvent molecules
close_solvent: boolean
If True solvent molecules will be placed very close to the solute
salt: python string
A comma separated string of the dissociated salt in solution
salt_concentration: float
Salt concentration in millimolar
neutralize_solute: boolean
If True counter-ions will be added to the solution to neutralize the solute
verbose: Bool
If True verbose mode is enabled
return_components: Bool
If True the added solvent molecules are also returned as OEMol
Return:
-------
oe_mol: OEMol
The solvated system. If the selected geometry is a box a SD tag with
name 'box_vector' is attached the output molecule containing
the system box vectors.
oe_mol_components: OEMol
If the return_components flag is True the added solvent molecules are
returned as an additional OEMol
"""
def BoundingBox(molecule):
"""
This function calculates the Bounding Box of the passed
molecule
molecule: OEMol
return: bb (numpy array)
the calculated bounding box is returned as numpy array:
[(xmin,ymin,zmin), (xmax,ymax,zmax)]
"""
coords = [v for k, v in molecule.GetCoords().items()]
np_coords = np.array(coords)
min_coord = np_coords.min(axis=0)
max_coord = np_coords.max(axis=0)
bb = np.array([min_coord, max_coord])
return bb
if shutil.which("packmol") is None:
raise (IOError("Packmol executable not found"))
# Extract solvent smiles strings and mole fractions
solvents = [sm.strip() for sm in solvents.split(',')]
fractions = [float(mf) for mf in molar_fractions.split(',')]
# If the smiles string and mole fractions lists have different lengths raise an error
if len(solvents) != len(fractions):
raise ValueError(
"Selected solvent number and selected molar fraction number mismatch: {} vs {}"
.format(len(solvents), len(fractions)))
# Remove smiles string with 0.0 mole fraction
solvent_smiles = [
solvents[i] for i, v in enumerate(fractions) if fractions[i]
]
mol_fractions = [mf for mf in fractions if mf]
# Mole fractions are non-negative numbers
if any([v < 0.0 for v in mol_fractions]):
raise ValueError("Error: Mole fractions are non-negative real numbers")
# Mole fractions must sum up to 1.0
if abs(sum(mol_fractions) - 1.0) > 0.001:
oechem.OEThrow.Error("Error: Mole fractions do not sum up to 1.0")
if geometry not in ['box', 'sphere']:
raise ValueError(
"Error geometry: the supported geometries are box and sphere not {}"
.format(geometry))
# Set Units
density = density * unit.grams / unit.milliliter
padding_distance = padding_distance * unit.angstrom
salt_concentration = salt_concentration * unit.millimolar
# Calculate the Solute Bounding Box
BB_solute = BoundingBox(solute)
# Estimate of the box cube length
box_edge = 2.0 * padding_distance + np.max(BB_solute[1] -
BB_solute[0]) * unit.angstrom
if geometry == 'box':
# Box Volume
Volume = box_edge**3
if geometry == 'sphere':
Volume = (4.0 / 3.0) * 3.14159265 * (0.5 * box_edge)**3
# Omega engine is used to generate conformations
omegaOpts = oeomega.OEOmegaOptions()
omegaOpts.SetMaxConfs(1)
omegaOpts.SetStrictStereo(False)
omega = oeomega.OEOmega(omegaOpts)
# Create a string code to identify the solute residues. The code ID used is based
# on the residue number id, the residue name and the chain id:
# id+resname+chainID
hv_solute = oechem.OEHierView(
solute,
oechem.OEAssumption_BondedResidue + oechem.OEAssumption_ResPerceived)
solute_resid_list = []
for chain in hv_solute.GetChains():
for frag in chain.GetFragments():
for hres in frag.GetResidues():
oe_res = hres.GetOEResidue()
solute_resid_list.append(
str(oe_res.GetResidueNumber()) + oe_res.GetName() +
chain.GetChainID())
# Solvent component list_names
solvent_resid_dic_names = dict()
# Neutralize solute
ion_sum_wgt_n_ions = 0.0 * unit.grams / unit.mole
if neutralize_solute:
# Container for the counter-ions
oe_ions = []
# Container for the ion smiles strings
ions_smiles = []
solute_formal_charge = 0
for at in solute.GetAtoms():
solute_formal_charge += at.GetFormalCharge()
if solute_formal_charge > 0:
ions_smiles.append("[Cl-]")
elif solute_formal_charge < 0:
ions_smiles.append("[Na+]")
else:
pass
# Total number of counter-ions to neutralize the solute
n_ions = abs(solute_formal_charge)
# print("Counter ions to add = {} of {}".format(n_ions, ions_smiles[0]))
# Ions
if n_ions >= 1:
for sm in ions_smiles:
mol = oechem.OEMol()
if not oechem.OESmilesToMol(mol, sm):
raise ValueError(
"Error counter ions: SMILES string parsing fails for the string: {}"
.format(sm))
# Generate conformer
if not omega(mol):
raise ValueError(
"Error counter ions: Conformer generation fails for the molecule with "
"smiles string: {}".format(sm))
oe_ions.append(mol)
if sm == '[Na+]':
solvent_resid_dic_names[' NA'] = mol
else:
solvent_resid_dic_names[' CL'] = mol
ion_sum_wgt = 0.0 * unit.grams / unit.mole
for ion in oe_ions:
# Molecular weight
ion_sum_wgt += oechem.OECalculateMolecularWeight(
ion) * unit.grams / unit.mole
ion_sum_wgt_n_ions = ion_sum_wgt * n_ions
# Create ions .pdb files
ions_smiles_pdbs = []
for i in range(0, len(ions_smiles)):
pdb_name = os.path.basename(tempfile.mktemp(suffix='.pdb'))
pdb_name = ions_smiles[i] + '_' + pdb_name
ions_smiles_pdbs.append(pdb_name)
for i in range(0, len(ions_smiles)):
ofs = oechem.oemolostream(ions_smiles_pdbs[i])
oechem.OEWriteConstMolecule(ofs, oe_ions[i])
# Add salts to the solution
# Solvent smiles string parsing
char_set = string.ascii_uppercase
salt_sum_wgt_n_salt = 0.0 * unit.grams / unit.mole
if salt_concentration > 0.0 * unit.millimolar:
salt_smiles = [sm.strip() for sm in salt.split(',')]
# Container list of oemol salt molecules generated by using smiles strings
oe_salt = []
for sm in salt_smiles:
mol_salt = oechem.OEMol()
if not oechem.OESmilesToMol(mol_salt, sm):
raise ValueError(
"Error salt: SMILES string parsing fails for the string: {}"
.format(sm))
# Generate conformer
if not omega(mol_salt):
raise ValueError(
"Error salt: Conformer generation fails for the "
"molecule with smiles string: {}".format(sm))
# Unique 3 code letter are set as solvent residue names
solv_id = ''.join(random.sample(char_set * 3, 3))
# Try to recognize the residue name
oechem.OEPerceiveResidues(mol_salt)
for atmol in mol_salt.GetAtoms():
res = oechem.OEAtomGetResidue(atmol)
if res.GetName() == 'UNL':
res.SetName(solv_id)
oechem.OEAtomSetResidue(atmol, res)
if solv_id not in solvent_resid_dic_names:
solvent_resid_dic_names[solv_id] = mol_salt
else:
if res.GetName() not in solvent_resid_dic_names:
solvent_resid_dic_names[res.GetName()] = mol_salt
break
oe_salt.append(mol_salt)
n_salt = int(
round(unit.AVOGADRO_CONSTANT_NA * salt_concentration *
Volume.in_units_of(unit.liter)))
# for i in range(0, len(salt_smiles)):
# print("Number of molecules for the salt component {} = {}".format(salt_smiles[i], n_salt))
salt_sum_wgt = 0.0 * unit.grams / unit.mole
for salt in oe_salt:
# Molecular weight
salt_sum_wgt += oechem.OECalculateMolecularWeight(
salt) * unit.grams / unit.mole
salt_sum_wgt_n_salt = salt_sum_wgt * n_salt
# Create salt .pdb files
if n_salt >= 1:
salt_pdbs = []
for i in range(0, len(salt_smiles)):
pdb_name = os.path.basename(tempfile.mktemp(suffix='.pdb'))
# pdb_name = salt_smiles[i] + '_' + pdb_name
salt_pdbs.append(pdb_name)
for i in range(0, len(salt_smiles)):
ofs = oechem.oemolostream(salt_pdbs[i])
oechem.OEWriteConstMolecule(ofs, oe_salt[i])
# Container list of oemol solvent molecules generated by using smiles strings
oe_solvents = []
for sm in solvent_smiles:
if sm == 'tip3p':
tip3p_fn = os.path.join(PACKAGE_DIR, 'oeommtools', 'data',
'tip3p.pdb')
ifs = oechem.oemolistream(tip3p_fn)
mol_sol = oechem.OEMol()
if not oechem.OEReadMolecule(ifs, mol_sol):
raise IOError(
"It was not possible to read the tip3p molecule file")
else:
mol_sol = oechem.OEMol()
if not oechem.OESmilesToMol(mol_sol, sm):
raise ValueError(
"Error solvent: SMILES string parsing fails for the string: {}"
.format(sm))
# Generate conformer
if not omega(mol_sol):
raise ValueError(
"Error solvent: Conformer generation fails for "
"the molecule with smiles string: {}".format(sm))
# Unique 3 code letter are set as solvent residue names
solv_id = ''.join(random.sample(char_set * 3, 3))
# Try to recognize the residue name
oechem.OEPerceiveResidues(mol_sol)
for atmol in mol_sol.GetAtoms():
res = oechem.OEAtomGetResidue(atmol)
if res.GetName() == 'UNL':
res.SetName(solv_id)
oechem.OEAtomSetResidue(atmol, res)
if solv_id not in solvent_resid_dic_names:
solvent_resid_dic_names[solv_id] = mol_sol
else:
if res.GetName() not in solvent_resid_dic_names:
solvent_resid_dic_names[res.GetName()] = mol_sol
break
oe_solvents.append(mol_sol)
# Sum of the solvent molecular weights
solvent_sum_wgt_frac = 0.0 * unit.grams / unit.mole
for idx in range(0, len(oe_solvents)):
# Molecular weight
wgt = oechem.OECalculateMolecularWeight(
oe_solvents[idx]) * unit.grams / unit.mole
solvent_sum_wgt_frac += wgt * mol_fractions[idx]
# Solute molecular weight
solute_wgt = oechem.OECalculateMolecularWeight(
solute) * unit.gram / unit.mole
# Estimate of the number of each molecular species present in the solution accordingly
# to their molar fraction fi:
#
# ni = fi*(density*volume*NA - wgt_solute - sum_k(wgt_salt_k*nk) - wgt_ion*n_ion)/sum_j(wgt_nj * fj)
#
# where ni is the number of molecule of specie i, density the mixture density, volume the
# mixture volume, wgt_solute the molecular weight of the solute, wgt_salt_k the molecular
# weight of the salt component k, nk the number of molecule of salt component k, wgt_ion
# the counter ion molecular weight, n_ions the number of counter ions and wgt_nj the molecular
# weight of the molecule specie j with molar fraction fj
div = (unit.AVOGADRO_CONSTANT_NA * density * Volume -
(solute_wgt + salt_sum_wgt_n_salt +
ion_sum_wgt_n_ions)) / solvent_sum_wgt_frac
# Solvent number of monomers
n_monomers = [int(round(mf * div)) for mf in mol_fractions]
if not all([nm > 0 for nm in n_monomers]):
raise ValueError(
"Error negative number of solvent components: the density could be too low"
)
# for i in range(0, len(solvent_smiles)):
# print("Number of molecules for the component {} = {}".format(solvent_smiles[i], n_monomers[i]))
# Packmol Configuration file setting
if close_solvent:
header_template = """\n# Mixture\ntolerance {}\nfiletype pdb\noutput {}\nadd_amber_ter\navoid_overlap no"""
else:
header_template = """\n# Mixture\ntolerance {}\nfiletype pdb\noutput {}\nadd_amber_ter\navoid_overlap yes"""
# Templates strings
solute_template = """\n\n# Solute\nstructure {}\nnumber 1\nfixed 0. 0. 0. 0. 0. 0.\nresnumbers 1\nend structure"""
if geometry == 'box':
solvent_template = """\nstructure {}\nnumber {}\ninside box {:0.3f} {:0.3f} {:0.3f} {:0.3f} {:0.3f} {:0.3f}\
\nchain !\nresnumbers 3\nend structure"""
if geometry == 'sphere':
solvent_template = """\nstructure {}\nnumber {}\ninside sphere {:0.3f} {:0.3f} {:0.3f} {:0.3f}\
\nchain !\nresnumbers 3\nend structure"""
# Create solvents .pdb files
solvent_pdbs = []
for i in range(0, len(solvent_smiles)):
pdb_name = os.path.basename(tempfile.mktemp(suffix='.pdb'))
solvent_pdbs.append(pdb_name)
for i in range(0, len(solvent_smiles)):
ofs = oechem.oemolostream(solvent_pdbs[i])
oechem.OEWriteConstMolecule(ofs, oe_solvents[i])
solute_pdb = 'solute' + '_' + os.path.basename(
tempfile.mktemp(suffix='.pdb'))
ofs = oechem.oemolostream(solute_pdb)
if solute.GetMaxConfIdx() > 1:
raise ValueError("Solutes with multiple conformers are not supported")
else:
oechem.OEWriteConstMolecule(ofs, solute)
# Write Packmol header section
mixture_pdb = 'mixture' + '_' + os.path.basename(
tempfile.mktemp(suffix='.pdb'))
body = header_template.format(distance_between_atoms, mixture_pdb)
# Write Packmol configuration file solute section
body += solute_template.format(solute_pdb)
# The solute is centered inside the box
xc = (BB_solute[0][0] + BB_solute[1][0]) / 2.
yc = (BB_solute[0][1] + BB_solute[1][1]) / 2.
zc = (BB_solute[0][2] + BB_solute[1][2]) / 2.
# Correct for periodic box conditions to avoid
# steric clashes at the box edges
pbc_correction = 1.0 * unit.angstrom
xmin = xc - ((box_edge - pbc_correction) / 2.) / unit.angstrom
xmax = xc + ((box_edge - pbc_correction) / 2.) / unit.angstrom
ymin = yc - ((box_edge - pbc_correction) / 2.) / unit.angstrom
ymax = yc + ((box_edge - pbc_correction) / 2.) / unit.angstrom
zmin = zc - ((box_edge - pbc_correction) / 2.) / unit.angstrom
zmax = zc + ((box_edge - pbc_correction) / 2.) / unit.angstrom
# Packmol setting for the solvent section
body += '\n\n# Solvent'
for i in range(0, len(solvent_smiles)):
if geometry == 'box':
body += solvent_template.format(solvent_pdbs[i], n_monomers[i],
xmin, ymin, zmin, xmax, ymax, zmax)
if geometry == 'sphere':
body += solvent_template.format(solvent_pdbs[i], n_monomers[i], xc,
yc, zc,
0.5 * box_edge / unit.angstrom)
# Packmol setting for the salt section
if salt_concentration > 0.0 * unit.millimolar and n_salt >= 1:
body += '\n\n# Salt'
for i in range(0, len(salt_smiles)):
if geometry == 'box':
body += solvent_template.format(salt_pdbs[i],
int(round(n_salt)), xmin, ymin,
zmin, xmax, ymax, zmax)
if geometry == 'sphere':
body += solvent_template.format(salt_pdbs[i],
int(round(n_salt)), xc, yc, zc,
0.5 * box_edge / unit.angstrom)
# Packmol setting for the ions section
if neutralize_solute and n_ions >= 1:
body += '\n\n# Counter Ions'
for i in range(0, len(ions_smiles)):
if geometry == 'box':
body += solvent_template.format(ions_smiles_pdbs[i], n_ions,
xmin, ymin, zmin, xmax, ymax,
zmax)
if geometry == 'sphere':
body += solvent_template.format(ions_smiles_pdbs[i], n_ions,
xc, yc, zc,
0.5 * box_edge / unit.angstrom)
# Packmol configuration file
packmol_filename = os.path.basename(tempfile.mktemp(suffix='.inp'))
with open(packmol_filename, 'w') as file_handle:
file_handle.write(body)
# Call Packmol
if not verbose:
mute_output = open(os.devnull, 'w')
with open(packmol_filename, 'r') as file_handle:
subprocess.check_call(['packmol'],
stdin=file_handle,
stdout=mute_output,
stderr=mute_output)
else:
with open(packmol_filename, 'r') as file_handle:
subprocess.check_call(['packmol'], stdin=file_handle)
# Read in the Packmol solvated system
solvated = oechem.OEMol()
if os.path.exists(mixture_pdb + '_FORCED'):
os.rename(mixture_pdb + '_FORCED', mixture_pdb)
print("Warning: Packing solution is not optimal")
ifs = oechem.oemolistream(mixture_pdb)
oechem.OEReadMolecule(ifs, solvated)
# To avoid to change the user oemol starting solute by reading in
# the generated mixture pdb file and loosing molecule info, the
# solvent molecules are extracted from the mixture system and
# added back to the starting solute
# Extract from the solution system the solvent molecules
# by checking the previous solute generated ID: id+resname+chainID
hv_solvated = oechem.OEHierView(
solvated,
oechem.OEAssumption_BondedResidue + oechem.OEAssumption_ResPerceived)
# This molecule will hold the solvent molecules generated directly from
# the omega conformers. This is useful to avoid problems related to read in
# the solvent molecules from pdb files and triggering unwanted perceiving actions
new_components = oechem.OEMol()
bv = oechem.OEBitVector(solvated.GetMaxAtomIdx())
for chain in hv_solvated.GetChains():
for frag in chain.GetFragments():
for hres in frag.GetResidues():
oe_res = hres.GetOEResidue()
if str(oe_res.GetResidueNumber()) + oe_res.GetName(
) + chain.GetChainID() not in solute_resid_list:
oechem.OEAddMols(new_components,
solvent_resid_dic_names[oe_res.GetName()])
atms = hres.GetAtoms()
for at in atms:
bv.SetBitOn(at.GetIdx())
pred = oechem.OEAtomIdxSelected(bv)
components = oechem.OEMol()
oechem.OESubsetMol(components, solvated, pred)
new_components.SetCoords(components.GetCoords())
# This is necessary otherwise just one big residue is created
oechem.OEPerceiveResidues(new_components)
# Add the solvent molecules to the solute copy
solvated_system = solute.CreateCopy()
oechem.OEAddMols(solvated_system, new_components)
# Set Title
solvated_system.SetTitle(solute.GetTitle())
# Set ions resname to Na+ and Cl-
for at in solvated_system.GetAtoms():
res = oechem.OEAtomGetResidue(at)
if res.GetName() == ' NA':
res.SetName("Na+")
oechem.OEAtomSetResidue(atmol, res)
elif res.GetName() == ' CL':
res.SetName("Cl-")
oechem.OEAtomSetResidue(atmol, res)
else:
pass
# Cleaning
to_delete = solvent_pdbs + [packmol_filename, solute_pdb, mixture_pdb]
if salt_concentration > 0.0 * unit.millimolar and n_salt >= 1:
to_delete += salt_pdbs
if neutralize_solute and n_ions >= 1:
to_delete += ions_smiles_pdbs
for fn in to_delete:
try:
os.remove(fn)
except:
pass
# Calculate the solution total density
total_wgt = oechem.OECalculateMolecularWeight(
solvated_system) * unit.gram / unit.mole
density_mix = (1 / unit.AVOGADRO_CONSTANT_NA) * total_wgt / Volume
print("Computed Solution Density = {}".format(
density_mix.in_units_of(unit.gram / unit.milliliter)))
# Threshold checking
ths = 0.1 * unit.gram / unit.milliliter
if not abs(density -
density_mix.in_units_of(unit.gram / unit.milliliter)) < ths:
raise ValueError(
"Error: the computed density for the solute {} does not match the selected density {} vs {}"
.format(solute.GetTitle(), density_mix, density))
if geometry == 'box':
# Define the box vector and attached it as SD tag to the solvated system
# with ID tag: 'box_vectors'
box_vectors = (Vec3(box_edge / unit.angstrom, 0.0,
0.0), Vec3(0.0, box_edge / unit.angstrom, 0.0),
Vec3(0.0, 0.0,
box_edge / unit.angstrom)) * unit.angstrom
box_vectors = data_utils.encodePyObj(box_vectors)
solvated_system.SetData(oechem.OEGetTag('box_vectors'), box_vectors)
if return_components:
new_components.SetTitle(solute.GetTitle() + '_solvent_comp')
return solvated_system, new_components
else:
return solvated_system
def applyffExcipients(excipients, opt):
"""
This function applies the selected force field to the
excipients
Parameters:
-----------
excipients: OEMol molecule
The excipients molecules to parametrize
opt: python dictionary
The options used to parametrize the excipients
Return:
-------
excipient_structure: Parmed structure instance
The parametrized excipient parmed structure
"""
# OpenMM topology and positions from OEMol
topology, positions = oeommutils.oemol_to_openmmTop(excipients)
# Try to apply the selected FF on the excipients
forcefield = app.ForceField(opt['protein_forcefield'])
# List of the unrecognized excipients
unmatched_res_list = forcefield.getUnmatchedResidues(topology)
# Unique unrecognized excipient names
templates = set()
for res in unmatched_res_list:
templates.add(res.name)
if templates: # Some excipients are not recognized
oechem.OEThrow.Info("The following excipients are not recognized "
"by the protein FF: {}"
"\nThey will be parametrized by using the FF: {}".format(templates, opt['other_forcefield']))
# Create a bit vector mask used to split recognized from un-recognize excipients
bv = oechem.OEBitVector(excipients.GetMaxAtomIdx())
bv.NegateBits()
# Dictionary containing the name and the parmed structures of the unrecognized excipients
unrc_excipient_structures = {}
# Dictionary used to skip already selected unrecognized excipients and count them
unmatched_excp = {}
# Ordered list of the unrecognized excipients
unmatched_res_order = []
for r_name in templates:
unmatched_excp[r_name] = 0
hv = oechem.OEHierView(excipients)
for chain in hv.GetChains():
for frag in chain.GetFragments():
for hres in frag.GetResidues():
r_name = hres.GetOEResidue().GetName()
if r_name not in unmatched_excp:
continue
else:
unmatched_res_order.append(r_name)
if unmatched_excp[r_name]: # Test if we have selected the unknown excipient
# Set Bit mask
atms = hres.GetAtoms()
for at in atms:
bv.SetBitOff(at.GetIdx())
unmatched_excp[r_name] += 1
else:
unmatched_excp[r_name] = 1
# Create AtomBondSet to extract from the whole excipient system
# the current selected FF unknown excipient
atms = hres.GetAtoms()
bond_set = set()
for at in atms:
bv.SetBitOff(at.GetIdx())
bonds = at.GetBonds()
for bond in bonds:
bond_set.add(bond)
atom_bond_set = oechem.OEAtomBondSet(atms)
for bond in bond_set:
atom_bond_set.AddBond(bond)
# Create the unrecognized excipient OEMol
unrc_excp = oechem.OEMol()
if not oechem.OESubsetMol(unrc_excp, excipients, atom_bond_set):
oechem.OEThrow.Fatal("Is was not possible extract the residue: {}".format(r_name))
# Charge the unrecognized excipient
if not oequacpac.OEAssignCharges(unrc_excp,
oequacpac.OEAM1BCCCharges(symmetrize=True)):
oechem.OEThrow.Fatal("Is was not possible to "
"charge the extract residue: {}".format(r_name))
# If GAFF or GAFF2 is selected as FF check for tleap command
if opt['other_forcefield'] in ['GAFF', 'GAFF2']:
ff_utils.ParamLigStructure(oechem.OEMol(), opt['other_forcefield']).checkTleap
if opt['other_forcefield'] == 'SMIRNOFF':
unrc_excp = oeommutils.sanitizeOEMolecule(unrc_excp)
# Parametrize the unrecognized excipient by using the selected FF
pmd = ff_utils.ParamLigStructure(unrc_excp, opt['other_forcefield'],
prefix_name=opt['prefix_name']+'_'+r_name)
unrc_excp_struc = pmd.parameterize()
unrc_excp_struc.residues[0].name = r_name
unrc_excipient_structures[r_name] = unrc_excp_struc
# Recognized FF excipients
pred_rec = oechem.OEAtomIdxSelected(bv)
rec_excp = oechem.OEMol()
oechem.OESubsetMol(rec_excp, excipients, pred_rec)
if rec_excp.NumAtoms() > 0:
top_known, pos_known = oeommutils.oemol_to_openmmTop(rec_excp)
ff_rec = app.ForceField(opt['protein_forcefield'])
try:
omm_system = ff_rec.createSystem(top_known, rigidWater=False)
rec_struc = parmed.openmm.load_topology(top_known, omm_system, xyz=pos_known)
except:
oechem.OEThrow.Fatal("Error in the recognised excipient parametrization")
# Unrecognized FF excipients
bv.NegateBits()
pred_unrc = oechem.OEAtomIdxSelected(bv)
unrc_excp = oechem.OEMol()
oechem.OESubsetMol(unrc_excp, excipients, pred_unrc)
# Unrecognized FF excipients coordinates
oe_coord_dic = unrc_excp.GetCoords()
unrc_coords = np.ndarray(shape=(unrc_excp.NumAtoms(), 3))
for at_idx in oe_coord_dic:
unrc_coords[at_idx] = oe_coord_dic[at_idx]
# It is important the order used to assemble the structures. In order to
# avoid mismatch between the coordinates and the structures, it is convenient
# to use the unrecognized residue order
unmatched_res_order_count = []
i = 0
while i < len(unmatched_res_order):
res_name = unmatched_res_order[i]
for j in range(i+1, len(unmatched_res_order)):
if unmatched_res_order[j] == res_name:
continue
else:
break
if i == (len(unmatched_res_order) - 1):
num = 1
unmatched_res_order_count.append((res_name, num))
break
else:
num = j - i
unmatched_res_order_count.append((res_name, num))
i = j
# Merge all the unrecognized Parmed structure
unrc_struc = parmed.Structure()
for pair in unmatched_res_order_count:
res_name = pair[0]
nums = pair[1]
unrc_struc = unrc_struc + nums*unrc_excipient_structures[res_name]
# Set the unrecognized coordinates
unrc_struc.coordinates = unrc_coords
# Set the parmed excipient structure merging
# the unrecognized and recognized parmed
# structures together
if rec_excp.NumAtoms() > 0:
excipients_structure = unrc_struc + rec_struc
else:
excipients_structure = unrc_struc
return excipients_structure
else: # All the excipients are recognized by the selected FF
omm_system = forcefield.createSystem(topology, rigidWater=False)
excipients_structure = parmed.openmm.load_topology(topology, omm_system, xyz=positions)
return excipients_structure
def extract_aligned_prot_lig_wat_traj(md_components,
flask,
trj_fn,
opt,
nmax=30,
water_cutoff=15.0):
"""
Extracts the aligned protein trajectory and aligned ligand trajectory and aligned
Water trajectory from a MD trajectory of a larger system that includes other
components (eg water).
The passed in setup mol must have the topology that matches the trajectory, and its xyz
coordinates are the reference for the alignment. The alignment is done on the
alpha carbons (atom name CA) of the active site residues within cutoff
from the ligand. Once the alignment is done, the protein and ligand trajectories
are each placed into a separate OEMol, one conformer per trajectory frame.
Water trajectory is selecting the nmax waters from the ligand and protein CA
within the cutoff distance for each trajectory snapshot
Inputs:
md_components: MDComponents object
The md components carrying the setup starting flask.
flask: OEMol
The system flask
trj_fn: String
The filename of the hdf5-format MD trajectory or Gromacs .trr file format
water_cutoff: Float
The cutoff distance between the PL binding site and the waters in angstroms
nmax: Integer
max number of waters to select
Outputs:
multi_conf_protein: A multi conformer OEMol for the protein, one conformer per frame.
multi_conf_ligand: A multi conformer OEMol for the ligand, one conformer per frame.
multi_conf_water: A multi conformer OEMol for the waters, one conformer per frame.
"""
# Extract protein, ligand, water and excipients from the flask
# protein, ligand, water, excipients = oeommutils.split(flask, ligand_res_name="LIG")
set_up_flask, map_dic = md_components.create_flask
protein = md_components.get_protein
ligand = md_components.get_ligand
check_nmax = nmax_waters(protein, ligand, water_cutoff)
if check_nmax < nmax:
opt['Logger'].warn(
"The selected number of max waters cannot fit around the protein binding site: {} vs {}"
.format(nmax, check_nmax))
void, traj_ext = os.path.splitext(trj_fn)
traj_dir = os.path.dirname(trj_fn)
if traj_ext == '.h5':
trj = md.load_hdf5(trj_fn)
elif traj_ext == '.trr':
pdb_fn = glob.glob(os.path.join(traj_dir, '*.pdb'))[0]
trj = md.load_trr(trj_fn, top=pdb_fn)
trj = trj[1:]
else:
raise ValueError(
"Trajectory file format {} not recognized in the trajectory {}".
format(traj_ext, trj_fn))
# System topology
top_trj = trj.topology
# Ligand indexes
# lig_idx = top_trj.select("resname LIG")
lig_idx = map_dic['ligand']
# Protein indexes
# prot_idx = top_trj.select("protein")
# It is safer to use OE toolkits than mdtraj which is missing the protein caps
prot_idx = map_dic['protein']
# for at in protein.GetAtoms():
# prot_idx.append(at.GetIdx())
# Water oxygen indexes
water_O_idx = top_trj.select("water and element O")
# Protein carbon alpha indexes
prot_ca_idx = top_trj.select("backbone and element C")
# Cutoff for the selection of the binding site atoms in A
cutoff_bs = 5.0
# Carbon alpha binding site indexes
ca_bs_idx = md.compute_neighbors(trj[0],
cutoff_bs / 10.0,
lig_idx,
haystack_indices=prot_ca_idx,
periodic=True)[0]
# Carbon alpha binding site and ligand indexes
ca_bs_lig_idx = np.concatenate((ca_bs_idx, lig_idx))
# Image the protein-ligand trajectory so the complex does not jump across box boundaries
protlig = trj[0].atom_slice(np.concatenate((prot_idx, lig_idx)))
protligAtoms = [atom for atom in protlig.topology.atoms]
with open(os.devnull, 'w') as devnull:
with contextlib.redirect_stderr(devnull):
trjImaged = trj.image_molecules(inplace=False,
anchor_molecules=[protligAtoms],
make_whole=True)
# trjImaged = trj.image_molecules(inplace=False, anchor_molecules=[protligAtoms], make_whole=True)
count = 0
water_max_frames = []
# TODO DEBUG
# trjImaged = trjImaged[:10]
for frame in trjImaged:
# print(count, flush=True)
# Water oxygen binding site indexes
water_O_bs_idx = md.compute_neighbors(frame,
water_cutoff / 10.0,
ca_bs_lig_idx,
haystack_indices=water_O_idx,
periodic=True)
# Pair combination water indexes times ligand indexes
wat_lig_pairs = np.array(np.meshgrid(water_O_bs_idx,
lig_idx)).T.reshape(-1, 2)
# Distances between the waters and the ligand in nm
wat_lig_distances = md.compute_distances(frame,
wat_lig_pairs,
periodic=True,
opt=True)
# Reshape the wat_lig_distances
ns = np.reshape(wat_lig_distances,
(len(water_O_bs_idx[0]), len(lig_idx)))
# Min distances in nm between the oxygen waters and the ligand
min_wat_O_lig_distances = np.min(ns, axis=1)
# Pair combination water indexes times protein binding site carbon alpha indexes
wat_ca_bs_pairs = np.array(np.meshgrid(water_O_bs_idx,
ca_bs_idx)).T.reshape(-1, 2)
# Distances between the waters and the protein binding site carbon alpha in nm
wat_ca_bs_distances = md.compute_distances(frame,
wat_ca_bs_pairs,
periodic=True,
opt=True)
# Reshape the wat_ca_bs_distances
ns = np.reshape(wat_ca_bs_distances,
(len(water_O_bs_idx[0]), len(ca_bs_idx)))
# Min distances in nm between the oxygen waters and the protein binding site carbon alpha
min_wat_O_ca_bs_distances = np.min(ns, axis=1)
metrics = min_wat_O_lig_distances + min_wat_O_ca_bs_distances
metric_distances = dict()
for wat_idx, m in zip(water_O_bs_idx[0], metrics):
metric_distances[int(wat_idx)] = m
water_list_sorted_max = sorted(metric_distances.items(),
key=lambda x: x[1])[:nmax]
if len(water_list_sorted_max) != nmax:
raise ValueError(
"The ordered water list has the wrong size {} vs expected {} for the frame {}"
.format(len(water_list_sorted_max), nmax, count))
water_max_frames.append(water_list_sorted_max)
# print(min_wat_O_ca_bs_distances)
# print(pairs[:len(lig_idx), :])
# for p,d in zip(wat_ca_bs_pairs, wat_ca_bs_distances[0]):
# print(p,d)
count += 1
# Put the reference mol xyz into the 1-frame topologyTraj to use as a reference in the fit
setup_mol_array_coords = oechem.OEDoubleArray(3 *
set_up_flask.GetMaxAtomIdx())
set_up_flask.GetCoords(setup_mol_array_coords)
setup_mol_xyzArr = np.array(setup_mol_array_coords)
setup_mol_xyzArr.shape = (-1, 3)
trj_reference = trjImaged[0]
# convert from angstroms to nanometers
trj_reference.xyz[0] = setup_mol_xyzArr / 10.0
# Fitting
trjImaged.superpose(trj_reference, 0, ca_bs_idx)
# Delete Original Trajectory to save memory
del trj
# Molecule copies
ligand_reference = oechem.OEMol(ligand)
protein_reference = oechem.OEMol(protein)
count = 0
# Create the multi conformer protein, ligand and water molecules
for frame in trjImaged.xyz:
# print("Trj Image loop", count, flush=True)
# Extract coordinates in A
xyz = frame * 10
# Set flask Coordinates as the current frame for the water extraction
flask.SetCoords(xyz.flatten())
water_list_sorted_max = water_max_frames[count]
# print(water_list_sorted_max)
# TODO The following solution to extract the waters do not
# keep the water order
# Mark the close water atoms and extract them
bv = oechem.OEBitVector(nmax * 3)
water_idx = []
for pair in water_list_sorted_max:
ow = flask.GetAtom(oechem.OEHasAtomIdx(pair[0]))
# Select the whole water molecule
for atw in oechem.OEGetResidueAtoms(ow):
bv.SetBitOn(atw.GetIdx())
water_idx.append(atw.GetIdx())
pred_vec = oechem.OEAtomIdxSelected(bv)
water_nmax_reference = oechem.OEMol()
oechem.OESubsetMol(water_nmax_reference, flask, pred_vec)
# TODO The following solution to extract the waters
# keep the water order but is it seems extremely inefficient
# water_list = []
# for pair in water_list_sorted_max:
# bv = oechem.OEBitVector(3)
# water_idx = []
# ow = flask.GetAtom(oechem.OEHasAtomIdx(pair[0]))
#
# # Select the whole water molecule
# for atw in oechem.OEGetResidueAtoms(ow):
# bv.SetBitOn(atw.GetIdx())
# water_idx.append(atw.GetIdx())
#
# pred_vec = oechem.OEAtomIdxSelected(bv)
# water = oechem.OEMol()
# oechem.OESubsetMol(water, flask, pred_vec)
#
# water_list.append(water)
#
#
# # print(len(water_list))
#
# water_nmax_reference = oechem.OEMol()
# for w in water_list:
# oechem.OEAddMols(water_nmax_reference, w)
# ligand and protein conf coordinates
lig_xyz_list = [10 * frame[idx] for idx in lig_idx]
lig_confxyz = oechem.OEFloatArray(np.array(lig_xyz_list).ravel())
prot_xyz_list = [10 * frame[idx] for idx in prot_idx]
prot_confxyz = oechem.OEFloatArray(np.array(prot_xyz_list).ravel())
# Initialize the protein, ligand and water molecule topologies
if count == 0:
multi_conf_water = oechem.OEMol(water_nmax_reference)
if multi_conf_water.NumAtoms() % 3 != 0:
raise ValueError("Number of Water atoms is not multiple of 3")
# Clean ResNumber and Chain on the multi conf water molecule
# oechem.OEPerceiveResidues(multi_conf_water, oechem.OEPreserveResInfo_All)
multi_conf_water.SetTitle("Water_" + str(nmax))
res_num = 0
i = 0
for at in multi_conf_water.GetAtoms():
res = oechem.OEAtomGetResidue(at)
res.SetSerialNumber(i)
res.SetName("HOH")
res.SetChainID("Z")
if i % 3 == 0:
res_num += 1
res.SetResidueNumber(res_num)
i += 1
ligand_reference.SetCoords(lig_confxyz)
protein_reference.SetCoords(prot_confxyz)
multi_conf_ligand = oechem.OEMol(ligand_reference)
multi_conf_protein = oechem.OEMol(protein_reference)
# Attach the conformers on the multi conformer protein, ligand and water molecules
else:
water_confxyz = oechem.OEFloatArray(
water_nmax_reference.NumAtoms() * 3)
water_nmax_reference.GetCoords(water_confxyz)
multi_conf_water.NewConf(water_confxyz)
multi_conf_ligand.NewConf(lig_confxyz)
multi_conf_protein.NewConf(prot_confxyz)
count += 1
return multi_conf_protein, multi_conf_ligand, multi_conf_water