def main(argv=[__name__]):
if len(argv) != 4:
oechem.OEThrow.Usage("%s <reffile> <rocs_hits_file> <output.sdf>" %
argv[0])
reffs = oechem.oemolistream(argv[1])
fitfs = oechem.oemolistream(argv[2])
outfs = oechem.oemolostream(argv[3])
refmol = oechem.OEGraphMol()
oechem.OEReadMolecule(reffs, refmol)
# Prepare reference molecule for calculation
# With default options this will add required color atoms
prep = oeshape.OEOverlapPrep()
prep.Prep(refmol)
# Get appropriate function to calculate analytic color
colorFunc = oeshape.OEAnalyticColorFunc()
colorFunc.SetupRef(refmol)
res = oeshape.OEOverlapResults()
for fitmol in fitfs.GetOEGraphMols():
prep.Prep(fitmol)
colorFunc.Overlap(fitmol, res)
oechem.OESetSDData(fitmol, "AnalyticColorTanimoto",
"%.2f" % res.GetColorTanimoto())
oechem.OEWriteMolecule(outfs, fitmol)
print("Fit Title: %s Color Tanimoto: %.2f" %
(fitmol.GetTitle(), res.GetColorTanimoto()))
def main(argv=[__name__]):
if len(argv) != 3:
oechem.OEThrow.Usage("%s <reffile> <fitfile>" % argv[0])
reffs = oechem.oemolistream(argv[1])
fitfs = oechem.oemolistream(argv[2])
refmol = oechem.OEGraphMol()
oechem.OEReadMolecule(reffs, refmol)
# Prepare reference molecule for calculation
# With default options this will remove any explicit
# hydrogens present and add color atoms
prep = oeshape.OEOverlapPrep()
prep.Prep(refmol)
# Get appropriate function to calculate both shape and color
# By default the OEOverlapFunc contains OEGridShapeFunc for shape
# and OEExactColorFunc for color
func = oeshape.OEOverlapFunc()
func.SetupRef(refmol)
res = oeshape.OEOverlapResults()
for fitmol in fitfs.GetOEGraphMols():
prep.Prep(fitmol)
func.Overlap(fitmol, res)
print("title: %s tanimoto combo = %.2f shape tanimoto = %.2f color tanimoto = %.2f" %
(fitmol.GetTitle(), res.GetTanimotoCombo(),
res.GetTanimoto(), res.GetColorTanimoto()))
def tanimotocombo(ref_mol, query_mol):
"""
This is the TanimotoCombo function. It takes in two OEMols.
It does not matter (for our purposes) which mol is the reference mol and
which one is the query mol. The TanimotoCombo distance between the two
mols is the same no matter which is the reference and which is the query.
The GetTanimotoCombo() function comes from openeye.
Args:
ref_mol (oemol) An oemol that has already been read in.
query_mol (oemol) Another oemol that has already been read in.
Returns:
res.GetTanimotoCombo() (float) The TanimotoCombo value.
"""
# Prepare reference molecule for calculation
# With default options this will remove any explicit
# hydrogens present and add color atoms
prep = oeshape.OEOverlapPrep()
prep.Prep(ref_mol)
# Get appropriate function to calculate both shape and color
# By default the OEOverlapFunc contains OEGridShapeFunc for shape
# and OEExactColorFunc for color
func = oeshape.OEOverlapFunc()
func.SetupRef(ref_mol)
res = oeshape.OEOverlapResults()
prep.Prep(query_mol)
func.Overlap(query_mol, res)
return (res.GetTanimotoCombo())
def main(argv=[__name__]):
if len(argv) != 3:
oechem.OEThrow.Usage("%s <reffile> <fitfile>" % argv[0])
reffs = oechem.oemolistream(argv[1])
fitfs = oechem.oemolistream(argv[2])
refmol = oechem.OEGraphMol()
oechem.OEReadMolecule(reffs, refmol)
# Prepare reference molecule for calculation
# With default options this will remove any explicit hydrogens present
prep = oeshape.OEOverlapPrep()
prep.Prep(refmol)
# Get appropriate function to calculate exact shape
shapeFunc = oeshape.OEExactShapeFunc()
shapeFunc.SetupRef(refmol)
res = oeshape.OEOverlapResults()
for fitmol in fitfs.GetOEGraphMols():
prep.Prep(fitmol)
shapeFunc.Overlap(fitmol, res)
print("title: %s exact tanimoto = %.2f" %
(fitmol.GetTitle(), res.GetTanimoto()))
def main(argv=[__name__]):
if len(argv) != 3:
oechem.OEThrow.Usage("%s <reffile> <fitfile>" % argv[0])
reffs = oechem.oemolistream(argv[1])
fitfs = oechem.oemolistream(argv[2])
refmol = oechem.OEGraphMol()
oechem.OEReadMolecule(reffs, refmol)
# Prepare reference molecule for calculation
# With default options this will remove any explicit
# hydrogens present, and add required color atoms
prep = oeshape.OEOverlapPrep()
prep.Prep(refmol)
# Get appropriate function to calculate exact color
colorFunc = oeshape.OEExactColorFunc()
colorFunc.SetupRef(refmol)
res = oeshape.OEOverlapResults()
fitmol = oechem.OEGraphMol()
while oechem.OEReadMolecule(fitfs, fitmol):
prep.Prep(fitmol)
colorFunc.Overlap(fitmol, res)
print("title: %s color score = %.2f" %
(fitmol.GetTitle(), res.GetColorScore()))
def main(argv=[__name__]):
itf = oechem.OEInterface(InterfaceData, argv)
# Set up best overlay to the query molecule
qfs = oechem.oemolistream()
if not qfs.open(itf.GetString("-q")):
oechem.OEThrow.Fatal("Unable to open %s" % itf.GetString("-q"))
qmol = oechem.OEMol()
oechem.OEReadMolecule(qfs, qmol)
# Set up overlap to protein exclusion volume
efs = oechem.oemolistream()
if not efs.open(itf.GetString("-e")):
oechem.OEThrow.Fatal("Unable to open %s" % itf.GetString("-e"))
emol = oechem.OEMol()
oechem.OEReadMolecule(efs, emol)
evol = oeshape.OEExactShapeFunc()
evol.SetupRef(emol)
# open database and output streams
ifs = oechem.oemolistream()
if not ifs.open(itf.GetString("-d")):
oechem.OEThrow.Fatal("Unable to open %s" % itf.GetString("-d"))
ofs = oechem.oemolostream()
if not ofs.open(itf.GetString("-o")):
oechem.OEThrow.Fatal("Unable to open %s" % itf.GetString("-o"))
print("Title Combo Rescore")
for mol in ifs.GetOEMols():
res = oeshape.OEROCSResult()
oeshape.OEROCSOverlay(res, qmol, mol)
outmol = res.GetOverlayConf()
# calculate overlap with protein
eres = oeshape.OEOverlapResults()
evol.Overlap(outmol, eres)
frac = eres.GetOverlap() / eres.GetFitSelfOverlap()
rescore = res.GetTanimotoCombo() - frac
# attach data to molecule and write it
oechem.OESetSDData(outmol, "TanimotoCombo", "%-.3f" % res.GetTanimotoCombo())
oechem.OESetSDData(outmol, "Exclusion Volume", "%-.3f" % eres.overlap)
oechem.OESetSDData(outmol, "Fraction Overlap", "%-.3f" % frac)
oechem.OESetSDData(outmol, "Rescore", "%-.3f" % rescore)
oechem.OEWriteMolecule(ofs, outmol)
print("%-20s %.3f %.3f" %
(outmol.GetTitle(), res.GetTanimotoCombo(), rescore))
def main(argv=[__name__]):
if len(argv) != 4:
oechem.OEThrow.Usage("%s <reffile> <rocs_hits_file> <output.sdf>" % argv[0])
reffs = oechem.oemolistream(argv[1])
fitfs = oechem.oemolistream(argv[2])
outfs = oechem.oemolostream(argv[3])
refmol = oechem.OEGraphMol()
oechem.OEReadMolecule(reffs, refmol)
# Get appropriate function to calculate analytic shape
shapeFunc = oeshape.OEAnalyticShapeFunc()
shapeFunc.SetupRef(refmol)
res = oeshape.OEOverlapResults()
for fitmol in fitfs.GetOEGraphMols():
shapeFunc.Overlap(fitmol, res)
oechem.OESetSDData(fitmol, "AnalyticTanimoto", "%.2f" % res.GetTanimoto())
oechem.OEWriteMolecule(outfs, fitmol)
def main(argv=[__name__]):
if len(argv) != 3:
oechem.OEThrow.Usage("%s <reffile> <overlayfile>" % argv[0])
reffs = oechem.oemolistream(argv[1])
fitfs = oechem.oemolistream(argv[2])
refmol = oechem.OEGraphMol()
oechem.OEReadMolecule(reffs, refmol)
# Modify ImplicitMillsDean color force field by
# adding user defined color interactions
cff = oeshape.OEColorForceField()
cff.Init(oeshape.OEColorFFType_ImplicitMillsDean)
cff.ClearInteractions()
donorType = cff.GetType("donor")
accepType = cff.GetType("acceptor")
cff.AddInteraction(donorType, donorType, "gaussian", -1.0, 1.0)
cff.AddInteraction(accepType, accepType, "gaussian", -1.0, 1.0)
# Prepare reference molecule for calculation
# With default options this will add required color atoms
# Set the modified color force field for addignment
prep = oeshape.OEOverlapPrep()
prep.SetColorForceField(cff)
prep.Prep(refmol)
# Get appropriate function to calculate exact color
# Set appropriate options to use the user defined color
options = oeshape.OEColorOptions()
options.SetColorForceField(cff)
colorFunc = oeshape.OEExactColorFunc(options)
colorFunc.SetupRef(refmol)
res = oeshape.OEOverlapResults()
for fitmol in fitfs.GetOEGraphMols():
prep.Prep(fitmol)
colorFunc.Overlap(fitmol, res)
print("Fit Title: %s Color Tanimoto: %.2f" %
(fitmol.GetTitle(), res.GetColorTanimoto()))
def generate_restricted_conformers(receptor, refmol, mol, core_smarts=None):
"""
Generate and select a conformer of the specified molecule using the reference molecule
Parameters
----------
receptor : openeye.oechem.OEGraphMol
Receptor (already prepped for docking) for identifying optimal pose
refmol : openeye.oechem.OEGraphMol
Reference molecule which shares some part in common with the proposed molecule
mol : openeye.oechem.OEGraphMol
Molecule whose conformers are to be enumerated
core_smarts : str, optional, default=None
If core_smarts is specified, substructure will be extracted using SMARTS.
"""
from openeye import oechem, oeomega
# DEBUG: For benzotriazoles, truncate refmol
core_smarts = 'c1ccc(NC(=O)[C,N]n2nnc3ccccc32)cc1' # prospective
core_smarts = 'NC(=O)[C,N]n2nnc3ccccc32' # retrospective
# Get core fragment
if core_smarts:
# Truncate refmol to SMARTS if specified
#print(f'Trunctating using SMARTS {refmol_smarts}')
ss = oechem.OESubSearch(core_smarts)
oechem.OEPrepareSearch(refmol, ss)
for match in ss.Match(refmol):
core_fragment = oechem.OEGraphMol()
oechem.OESubsetMol(core_fragment, match)
break
#print(f'refmol has {refmol.NumAtoms()} atoms')
else:
core_fragment = GetCoreFragment(refmol, [mol])
oechem.OESuppressHydrogens(core_fragment)
#print(f' Core fragment has {core_fragment.NumAtoms()} heavy atoms')
MIN_CORE_ATOMS = 6
if core_fragment.NumAtoms() < MIN_CORE_ATOMS:
return None
# Create an Omega instance
#omegaOpts = oeomega.OEOmegaOptions()
omegaOpts = oeomega.OEOmegaOptions(oeomega.OEOmegaSampling_Dense)
# Set the fixed reference molecule
omegaFixOpts = oeomega.OEConfFixOptions()
omegaFixOpts.SetFixMaxMatch(10) # allow multiple MCSS matches
omegaFixOpts.SetFixDeleteH(True) # only use heavy atoms
omegaFixOpts.SetFixMol(core_fragment)
#omegaFixOpts.SetFixSmarts(smarts)
omegaFixOpts.SetFixRMS(0.5)
atomexpr = oechem.OEExprOpts_Aromaticity | oechem.OEExprOpts_Hybridization
bondexpr = oechem.OEExprOpts_BondOrder | oechem.OEExprOpts_Aromaticity
omegaFixOpts.SetAtomExpr(atomexpr)
omegaFixOpts.SetBondExpr(bondexpr)
omegaOpts.SetConfFixOptions(omegaFixOpts)
molBuilderOpts = oeomega.OEMolBuilderOptions()
molBuilderOpts.SetStrictAtomTypes(False) # don't give up if MMFF types are not found
omegaOpts.SetMolBuilderOptions(molBuilderOpts)
omegaOpts.SetWarts(False) # expand molecule title
omegaOpts.SetStrictStereo(False) # set strict stereochemistry
omegaOpts.SetIncludeInput(False) # don't include input
omegaOpts.SetMaxConfs(1000) # generate lots of conformers
#omegaOpts.SetEnergyWindow(10.0) # allow high energies
omega = oeomega.OEOmega(omegaOpts)
from openeye import oequacpac
if not oequacpac.OEGetReasonableProtomer(mol):
print('No reasonable protomer found')
return None
mol = oechem.OEMol(mol) # multi-conformer molecule
ret_code = omega.Build(mol)
if (mol.GetDimension() != 3) or (ret_code != oeomega.OEOmegaReturnCode_Success):
print(f'Omega failure: {mol.GetDimension()} and {oeomega.OEGetOmegaError(ret_code)}')
return None
# Extract poses
class Pose(object):
def __init__(self, conformer):
self.conformer = conformer
self.clash_score = None
self.docking_score = None
self.overlap_score = None
poses = [ Pose(conf) for conf in mol.GetConfs() ]
# Score clashes
bump_check = BumpCheck(receptor)
for pose in poses:
pose.clash_score = bump_check.count(pose.conformer)
# Score docking poses
from openeye import oedocking
score = oedocking.OEScore(oedocking.OEScoreType_Chemgauss4)
score.Initialize(receptor)
for pose in poses:
pose.docking_score = score.ScoreLigand(pose.conformer)
# Compute overlap scores
from openeye import oeshape
overlap_prep = oeshape.OEOverlapPrep()
overlap_prep.Prep(refmol)
shapeFunc = oeshape.OEExactShapeFunc()
shapeFunc.SetupRef(refmol)
oeshape_result = oeshape.OEOverlapResults()
for pose in poses:
tmpmol = oechem.OEGraphMol(pose.conformer)
overlap_prep.Prep(tmpmol)
shapeFunc.Overlap(tmpmol, oeshape_result)
pose.overlap_score = oeshape_result.GetRefTversky()
# Filter poses based on top 10% of overlap
poses = sorted(poses, key= lambda pose : pose.overlap_score)
poses = poses[int(0.9*len(poses)):]
# Select the best docking score
import numpy as np
poses = sorted(poses, key=lambda pose : pose.docking_score)
pose = poses[0]
mol.SetActive(pose.conformer)
oechem.OESetSDData(mol, 'clash_score', str(pose.clash_score))
oechem.OESetSDData(mol, 'docking_score', str(pose.docking_score))
oechem.OESetSDData(mol, 'overlap_score', str(pose.overlap_score))
# Convert to single-conformer molecule
mol = oechem.OEGraphMol(mol)
return mol
shapeFunc = oeshape.OEOverlapFunc()
# Compute distinct fragments
compute_fragment_basis = True
n_clusters = 40
if compute_fragment_basis:
# TODO: Use PCCA+ when we have time to implement this
print('Extracting a fragment basis...')
import numpy as np
from sklearn.cluster import SpectralClustering
fragment_names = list(fragments.keys())
nfragments = len(fragment_names)
affinity_matrix = np.ones([nfragments, nfragments], np.float32)
for i in range(nfragments):
shapeFunc.SetupRef(fragments[fragment_names[i]])
result = oeshape.OEOverlapResults()
for j in range(i + 1, nfragments):
shapeFunc.Overlap(fragments[fragment_names[j]], result)
overlap = result.GetTanimoto()
affinity_matrix[i, j] = overlap
affinity_matrix[j, i] = overlap
clustering = SpectralClustering(
n_clusters=n_clusters, affinity='precomputed').fit(affinity_matrix)
unique_fragment_names = [index for index in range(n_clusters)]
for fragment_index, cluster_index in enumerate(clustering.labels_):
unique_fragment_names[cluster_index] = fragment_names[
fragment_index]
fragments = {
fragment_name: fragments[fragment_name]
for fragment_name in unique_fragment_names
}
def generate_restricted_conformers(receptor, refmol, mol, core_smarts=None):
"""
Generate and select a conformer of the specified molecule using the reference molecule
Parameters
----------
receptor : openeye.oechem.OEGraphMol
Receptor (already prepped for docking) for identifying optimal pose
refmol : openeye.oechem.OEGraphMol
Reference molecule which shares some part in common with the proposed molecule
mol : openeye.oechem.OEGraphMol
Molecule whose conformers are to be enumerated
core_smarts : str, optional, default=None
If core_smarts is specified, substructure will be extracted using SMARTS.
"""
from openeye import oechem, oeomega
logging.debug(
f'mol: {oechem.OEMolToSmiles(mol)} | core_smarts: {core_smarts}')
# Be quiet
from openeye import oechem
oechem.OEThrow.SetLevel(oechem.OEErrorLevel_Quiet)
#oechem.OEThrow.SetLevel(oechem.OEErrorLevel_Error)
# Get core fragment
if core_smarts:
# Truncate refmol to SMARTS if specified
#print(f'Trunctating using SMARTS {refmol_smarts}')
ss = oechem.OESubSearch(core_smarts)
oechem.OEPrepareSearch(refmol, ss)
for match in ss.Match(refmol):
core_fragment = oechem.OEGraphMol()
oechem.OESubsetMol(core_fragment, match)
logging.debug(
f'Truncated refmol to generate core_fragment: {oechem.OEMolToSmiles(core_fragment)}'
)
break
#print(f'refmol has {refmol.NumAtoms()} atoms')
else:
core_fragment = GetCoreFragment(refmol, [mol])
oechem.OESuppressHydrogens(core_fragment)
#print(f' Core fragment has {core_fragment.NumAtoms()} heavy atoms')
MIN_CORE_ATOMS = 6
if core_fragment.NumAtoms() < MIN_CORE_ATOMS:
return None
# Create an Omega instance
#omegaOpts = oeomega.OEOmegaOptions()
omegaOpts = oeomega.OEOmegaOptions(oeomega.OEOmegaSampling_Dense)
# Set the fixed reference molecule
omegaFixOpts = oeomega.OEConfFixOptions()
omegaFixOpts.SetFixMaxMatch(10) # allow multiple MCSS matches
omegaFixOpts.SetFixDeleteH(True) # only use heavy atoms
omegaFixOpts.SetFixMol(core_fragment)
#omegaFixOpts.SetFixSmarts(core_smarts) # DEBUG
omegaFixOpts.SetFixRMS(0.5)
# This causes a warning:
#Warning: OESubSearch::Match() is unable to match unset hybridization in the target (EN300-221518_3_1) for patterns with set hybridization, call OEPrepareSearch on the target first
#atomexpr = oechem.OEExprOpts_Aromaticity | oechem.OEExprOpts_Hybridization
atomexpr = oechem.OEExprOpts_Aromaticity | oechem.OEExprOpts_AtomicNumber
bondexpr = oechem.OEExprOpts_BondOrder | oechem.OEExprOpts_Aromaticity
omegaFixOpts.SetAtomExpr(atomexpr)
omegaFixOpts.SetBondExpr(bondexpr)
omegaOpts.SetConfFixOptions(omegaFixOpts)
molBuilderOpts = oeomega.OEMolBuilderOptions()
molBuilderOpts.SetStrictAtomTypes(
False) # don't give up if MMFF types are not found
omegaOpts.SetMolBuilderOptions(molBuilderOpts)
omegaOpts.SetWarts(False) # expand molecule title
omegaOpts.SetStrictStereo(True) # set strict stereochemistry
omegaOpts.SetIncludeInput(False) # don't include input
omegaOpts.SetMaxConfs(1000) # generate lots of conformers
omegaOpts.SetEnergyWindow(20.0) # allow high energies
omega = oeomega.OEOmega(omegaOpts)
# TODO: Expand protonation states and tautomers
from openeye import oequacpac
if not oequacpac.OEGetReasonableProtomer(mol):
logging.warning('No reasonable protomer found')
return None
mol = oechem.OEMol(mol) # multi-conformer molecule
ret_code = omega.Build(mol)
if (mol.GetDimension() != 3) or (ret_code !=
oeomega.OEOmegaReturnCode_Success):
msg = f'\nOmega failure for {mol.GetTitle()} : SMILES {oechem.OEMolToSmiles(mol)} : core_smarts {core_smarts} : {oeomega.OEGetOmegaError(ret_code)}\n'
logging.warning(msg)
return None
# Return the molecule with an error code
#oechem.OESetSDData(mol, 'error', '{oeomega.OEGetOmegaError(ret_code)}')
#return mol
# Extract poses
class Pose(object):
def __init__(self, conformer):
self.conformer = conformer
self.clash_score = None
self.docking_score = None
self.overlap_score = None
poses = [Pose(conf) for conf in mol.GetConfs()]
# Score clashes
bump_check = BumpCheck(receptor)
for pose in poses:
pose.clash_score = bump_check.count(pose.conformer)
# Score docking poses
from openeye import oedocking
score = oedocking.OEScore(oedocking.OEScoreType_Chemgauss4)
score.Initialize(receptor)
for pose in poses:
pose.docking_score = score.ScoreLigand(pose.conformer)
# Compute overlap scores
from openeye import oeshape
overlap_prep = oeshape.OEOverlapPrep()
overlap_prep.Prep(refmol)
shapeFunc = oeshape.OEExactShapeFunc()
shapeFunc.SetupRef(refmol)
oeshape_result = oeshape.OEOverlapResults()
for pose in poses:
tmpmol = oechem.OEGraphMol(pose.conformer)
overlap_prep.Prep(tmpmol)
shapeFunc.Overlap(tmpmol, oeshape_result)
pose.overlap_score = oeshape_result.GetRefTversky()
# Filter poses based on top 10% of overlap
poses = sorted(poses, key=lambda pose: pose.overlap_score)
poses = poses[int(0.9 * len(poses)):]
# Select the best docking score
import numpy as np
poses = sorted(poses, key=lambda pose: pose.docking_score)
pose = poses[0]
mol.SetActive(pose.conformer)
oechem.OESetSDData(mol, 'clash_score', str(pose.clash_score))
oechem.OESetSDData(mol, 'docking_score', str(pose.docking_score))
oechem.OESetSDData(mol, 'overlap_score', str(pose.overlap_score))
# Convert to single-conformer molecule
mol = oechem.OEGraphMol(mol)
# Compute MMFF energy
energy = mmff_energy(mol)
oechem.OESetSDData(mol, 'MMFF_internal_energy', str(energy))
# Store SMILES
docked_smiles = oechem.OEMolToSmiles(mol)
oechem.OESetSDData(mol, 'docked_smiles', docked_smiles)
return mol
