def fastrocs_query(self, qmol, numHits, host):
with self.logger("fastrocs_query") as logger:
ofs = oechem.oemolostream()
ofs.SetFormat(oechem.OEFormat_OEB)
ofs.openstring()
oechem.OEWriteMolecule(ofs, qmol)
bytes = ofs.GetString()
s = ServerProxy("http://" + host)
data = Binary(bytes)
# idx = s.SubmitQuery(data, numHits)
dargs = {
'altStarts': 'random',
'tversky': False,
'shapeOnly': False
}
assert (numHits is not None)
assert (data is not None)
assert (dargs is not None)
idx = s.SubmitQuery(data, numHits, 'oeb', 'oeb', dargs)
first = False
while True:
try:
current, total = s.QueryStatus(idx, True)
except Fault as e:
logger.error((str(e)))
return 1
if total == 0:
continue
if first:
# logger.log("%s/%s" % ("current", "total"))
first = False
# logger.log("%i/%i" % (current, total))
if total <= current:
break
results = s.QueryResults(idx)
ifs = oechem.oemolistream()
ifs.openstring(results.data)
ifs.SetFormat(oechem.OEFormat_OEB)
mols = []
for mol in ifs.GetOEMols():
good_mol = oechem.OEMol(mol)
oechem.OEAddExplicitHydrogens(good_mol)
oechem.OEClearSDData(good_mol)
oeshape.OEDeleteCompressedColorAtoms(good_mol)
oeshape.OEClearCachedSelfColor(good_mol)
oeshape.OEClearCachedSelfShape(good_mol)
oeshape.OERemoveColorAtoms(good_mol)
mols.append(good_mol)
return mols
def from_oemol(self, from_oemol):
with self.logger("from_oemol") as logger:
tautomer_options = oequacpac.OETautomerOptions()
tautomer_options.SetMaxTautomersGenerated(4096)
tautomer_options.SetMaxTautomersToReturn(16)
tautomer_options.SetCarbonHybridization(True)
tautomer_options.SetMaxZoneSize(50)
tautomer_options.SetApplyWarts(True)
pKa_norm = True
omegaOpts = oeomega.OEOmegaOptions(oeomega.OEOmegaSampling_Pose)
omegaOpts.SetStrictAtomTypes(False)
omegaOpts.SetSampleHydrogens(True)
omegaOpts.SetMaxSearchTime(30)
omegaOpts.SetFixDeleteH(True)
omega = oeomega.OEOmega(omegaOpts)
options = oeshape.OEROCSOptions()
overlayoptions = oeshape.OEOverlayOptions()
overlayoptions.SetOverlapFunc(
oeshape.OEOverlapFunc(oeshape.OEAnalyticShapeFunc()))
options.SetOverlayOptions(overlayoptions)
# options.SetNumBestHits(10)
options.SetConfsPerHit(200)
# options.SetMaxHits(10000)
rocs = oeshape.OEROCS(options)
for tautomer in oequacpac.OEGetReasonableTautomers(
from_oemol, tautomer_options, pKa_norm):
logger.log("got enantiomer")
for enantiomer in oeomega.OEFlipper(tautomer, 4, False):
logger.log("got tautomer ")
enantiomer_ = oechem.OEMol(enantiomer)
ret_code = omega.Build(enantiomer_)
if ret_code != oeomega.OEOmegaReturnCode_Success:
logger.error("got oemeg_failed",
oeomega.OEGetOmegaError(ret_code))
else:
rocs.AddMolecule(oechem.OEMol(enantiomer_))
for res in rocs.Overlay(self.refmol):
outmol = oechem.OEMol(res.GetOverlayConfs())
good_mol = oechem.OEMol(outmol)
oechem.OEAddExplicitHydrogens(good_mol)
oechem.OEClearSDData(good_mol)
oeshape.OEDeleteCompressedColorAtoms(good_mol)
oeshape.OEClearCachedSelfColor(good_mol)
oeshape.OEClearCachedSelfShape(good_mol)
oeshape.OERemoveColorAtoms(good_mol)
return good_mol
logger.error("Returning None.")
return None
def reorder_sd_props(mol: oechem.OEGraphMol):
strain1 = oechem.OEGetSDData(mol, TOTAL_STRAIN_TAG)
data_pairs = [(TOTAL_STRAIN_TAG, strain1)]
for dp in oechem.OEGetSDDataPairs(mol):
if dp.GetTag() == TOTAL_STRAIN_TAG:
pass
else:
data_pairs.append((dp.GetTag(), dp.GetValue()))
oechem.OEClearSDData(mol)
for k, v in data_pairs:
oechem.OESetSDData(mol, k, v)
data_pairs = []
for dp in oechem.OEGetSDDataPairs(mol):
data_pairs.append((dp.GetTag(), dp.GetValue()))
oechem.OEClearSDData(mol)
for k, v in data_pairs:
oechem.OESetSDData(mol, k, v)
def FilterMolData(self, mol):
if not oechem.OEHasSDData(mol):
return 0
if self.fields is None:
return -1
if len(self.fields) == 0:
oechem.OEClearSDData(mol)
return 0
validdata = 0
deletefields = []
for dp in oechem.OEGetSDDataPairs(mol):
tag = dp.GetTag()
if tag not in self.fields:
deletefields.append(tag)
continue
value = oechem.OEGetSDData(mol, tag)
if self.asFloating:
try:
float(value)
except ValueError:
oechem.OEThrow.Warning("Failed to convert %s to numeric value (%s) in %s" %
(tag, value, mol.GetTitle()))
deletefields.append(tag)
continue
validdata += 1
if not validdata:
oechem.OEClearSDData(mol)
else:
for nuke in deletefields:
oechem.OEDeleteSDData(mol, nuke)
return validdata
# Read all submitted designs: Compounds with the key substructure will be retained
print('Reading submitted designs...')
# Drop columns that cause trouble for OpenEye
import pandas as pd
drop_columns = ['Submission Rationale', 'Submission Notes']
df = pd.read_csv(submissions_csv_filename, dtype=str)
df.drop(columns=drop_columns, inplace=True)
import tempfile
with tempfile.NamedTemporaryFile(suffix='.csv') as csv_file:
df.to_csv(csv_file.name, header=True, index=False)
# Read file
with oechem.oemolistream(csv_file.name) as ifs:
mol = oechem.OEGraphMol()
while oechem.OEReadMolecule(ifs, mol):
# Clear SD tags
oechem.OEClearSDData(mol)
# Store the molecule
mols.append(mol.CreateCopy())
print(f'{len(mols)} molecules read')
# Aggregate all compound designs
source_filenames = [
# Filtered synthetic designs
'filtered/transformations-final-ligands.csv',
]
for source_filename in source_filenames:
with oechem.oemolistream(source_filename) as ifs:
mol = oechem.OEGraphMol()
while oechem.OEReadMolecule(ifs, mol):
# Clear SD tags
oechem.OEClearSDData(mol)
def construct_dihedral_energy_profile(torsion_conformers, num_points=24):
angle_list = np.array([360 * i / num_points for i in range(num_points)])
num_confs = 0
profile = np.full(num_points, np.nan)
for mol in torsion_conformers:
if not mol:
continue
num_confs += 1
conf = mol.GetActive()
conf_title = get_sd_data(conf, "CONFORMER_LABEL")
tor_atoms = get_sd_data(mol, "TORSION_ATOMS_ParentMol").split()
parent_name = conf_title[:-3]
dih_label = "_".join(str(x) for x in tor_atoms)
fragment_label = parent_name + "_" + dih_label
angle_idx = int(conf_title[-2:])
profile[angle_idx] = np.float(get_sd_data(conf, "PSI4_ENERGY"))
logging.debug("angle_idx: %d", angle_idx)
logging.debug("Psi4 Energy: %f",
float(get_sd_data(conf, "PSI4_ENERGY")))
# check for angles where no energies are available
for angle in angle_list[np.all(np.isnan(profile))]:
logging.warning(
"Warning: No energies found for angle {:.1f} for fragment: {}".
format(angle, fragment_label))
# calculate relative energies
min_energy = np.nanmin(profile)
profile -= min_energy
profile[np.isnan(profile)] = -1 # set nans to -1
torsional_strain = np.column_stack((angle_list, profile))
# combine conformers
output_conformers = oechem.OEMol(torsion_conformers[0])
output_conformers.DeleteConfs()
title = fragment_label
output_conformers.SetTitle(title)
# setup normalization
torsion_tag = "TORSION_ATOMS_FRAGMENT"
torsion_atoms_in_fragment = get_sd_data(mol, torsion_tag).split()
print(torsion_atoms_in_fragment)
dihedral_atom_indices = [int(x) - 1 for x in torsion_atoms_in_fragment]
dih, _ = get_dihedral(output_conformers, dihedral_atom_indices)
for old_conf in torsion_conformers:
if old_conf:
new_conf = output_conformers.NewConf(old_conf)
normalize_coordinates(new_conf, dih)
oechem.OEClearSDData(new_conf)
for dp in oechem.OEGetSDDataPairs(old_conf.GetActive()):
if dp.GetTag() not in ["OEConfTitle", "CONFORMER_LABEL"]:
oechem.OESetSDData(new_conf, dp.GetTag(), dp.GetValue())
torsion_angle = get_sd_data(old_conf, "TORSION_ANGLE")
title = fragment_label + ": Angle " + torsion_angle
new_conf.SetTitle(title)
write_energy_profile_to_sddata(output_conformers, torsional_strain.copy())
# Calculate all possible torsion inchi keys for this fragment
torsion_inchi_list = []
inchi_key = oechem.OECreateInChIKey(output_conformers)
_, b, c, _ = get_torsion_oeatom_list(output_conformers)
for a in b.GetAtoms(oechem.OEIsHeavy()):
for d in c.GetAtoms(oechem.OEIsHeavy()):
if a.GetIdx() == c.GetIdx() or d.GetIdx() == b.GetIdx():
continue
torsion_inchi = inchi_key + get_modified_inchi_key(
output_conformers, [a, b, c, d])
torsion_inchi_list.append(torsion_inchi)
return output_conformers, torsional_strain, torsion_inchi_list
def gen_starting_confs(
mol,
torsion_library,
max_one_bond_away=True,
num_conformers=MAX_CONFS,
rms_cutoff=0.0,
energy_window=25,
):
# Identify the atoms in the dihedral
TAGNAME = "TORSION_ATOMS_FRAGMENT"
if not has_sd_data(mol, TAGNAME):
raise ValueError(
"Molecule does not have the SD Data Tag '{}'.".format(TAGNAME))
dihedralAtomIndices = [
int(x) - 1 for x in get_sd_data(mol, TAGNAME).split()
]
inDih = oechem.OEOrAtom(
oechem.OEOrAtom(
oechem.OEHasAtomIdx(dihedralAtomIndices[0]),
oechem.OEHasAtomIdx(dihedralAtomIndices[1]),
),
oechem.OEOrAtom(
oechem.OEHasAtomIdx(dihedralAtomIndices[2]),
oechem.OEHasAtomIdx(dihedralAtomIndices[3]),
),
)
mol1 = mol.CreateCopy()
mc_mol = oechem.OEMol(mol1)
# Tag torsion atoms with their dihedral index
for atom in mc_mol.GetAtoms():
if atom.GetIdx() == dihedralAtomIndices[0]:
atom.SetData("dihidx", 0)
if atom.GetIdx() == dihedralAtomIndices[1]:
atom.SetData("dihidx", 1)
if atom.GetIdx() == dihedralAtomIndices[2]:
atom.SetData("dihidx", 2)
if atom.GetIdx() == dihedralAtomIndices[3]:
atom.SetData("dihidx", 3)
if num_conformers > 1:
# Set criterion for rotatable bond
if False and max_one_bond_away:
# this max function makes this seem potentially broken
only_one_bond_away = distance_predicate(dihedralAtomIndices[1],
dihedralAtomIndices[2])
rotor_predicate = oechem.OEAndBond(
only_one_bond_away, oechem.PyBondPredicate(isRotatableBond))
elif False:
# this ONLY samples special bonds & neglects "regualr" torsions
rotor_predicate = oechem.PyBondPredicate(isRotatableBond)
else:
# try this more general sampling, but leave prior versions untouched
rotor_predicate = oechem.OEOrBond(
oechem.OEIsRotor(), oechem.PyBondPredicate(isRotatableBond))
# Initialize conformer generator and multi-conformer library
conf_generator = configure_omega(torsion_library, rotor_predicate,
rms_cutoff, energy_window,
num_conformers)
# Generator conformers
if not conf_generator(mc_mol, inDih):
raise ValueError("Conformers cannot be generated.")
logging.debug(
"Generated a total of %d conformers for %s.",
mc_mol.NumConfs(),
mol.GetTitle(),
)
# Reassign
new_didx = [-1, -1, -1, -1]
for atom in mc_mol.GetAtoms():
if atom.HasData("dihidx"):
new_didx[atom.GetData("dihidx")] = atom.GetIdx()
oechem.OEClearSDData(mc_mol)
oechem.OESetSDData(mc_mol, TAGNAME,
" ".join(str(x + 1) for x in new_didx))
oechem.OESetSDData(
mc_mol,
"TORSION_ATOMS_ParentMol",
get_sd_data(mol, "TORSION_ATOMS_ParentMol"),
)
oechem.OESetSDData(
mc_mol,
"TORSION_ATOMPROP",
f"cs1:0:1;1%{new_didx[0]+1}:1%{new_didx[1]+1}:1%{new_didx[2]+1}:1%{new_didx[3]+1}",
)
for conf_no, conf in enumerate(mc_mol.GetConfs()):
conformer_label = (
mol.GetTitle() + "_" +
"_".join(get_sd_data(mol, "TORSION_ATOMS_ParentMol").split()) +
"_{:02d}".format(conf_no))
oechem.OESetSDData(conf, "CONFORMER_LABEL", conformer_label)
conf.SetTitle(conformer_label)
return mc_mol
def ClearProps(ifs, ofs):
for mol in ifs.GetOEGraphMols():
oechem.OEClearSDData(mol)
oechem.OEWriteMolecule(ofs, mol)