def mol_with_smiles_is_in_contnr(self, smiles):
"""Checks whether or not a given smiles string is already in this
container.
:param smiles: The smiles string to check.
:type smiles: str
:return: True if it is present, otherwise a new MyMol.MyMol object
corresponding to that smiles.
:rtype: bool or MyMol.MyMol
"""
# Checks all the mols in this container to see if a given smiles is
# already present. Returns a new MyMol object if it isn't, True
# otherwise.
# First, get the set of all cannonical smiles.
# TODO: Probably shouldn't be generating this on the fly every time
# you use it!
can_smi_in_this_container = set([m.smiles() for m in self.mols])
# Determine whether it is already in the container, and act
# accordingly.
amol = MyMol.MyMol(smiles)
if amol.smiles() in can_smi_in_this_container:
return True
else:
return amol
def update_orig_smi(self, orig_smi):
"""Updates the orig_smi string. Used by desalter (to replace with
largest fragment).
:param orig_smi: The replacement smiles string.
:type orig_smi: str
"""
# Update the MolContainer object
self.orig_smi = orig_smi
self.orig_smi_deslt = orig_smi
self.mol_orig_frm_inp_smi = MyMol.MyMol(self.orig_smi, self.name)
self.frgs = ""
self.orig_smi_canonical = self.mol_orig_frm_inp_smi.smiles()
self.num_nonaro_rngs = len(
self.mol_orig_frm_inp_smi.get_idxs_of_nonaro_rng_atms()
)
self.num_specif_chiral_cntrs = len(
self.mol_orig_frm_inp_smi.chiral_cntrs_only_asignd()
)
self.num_unspecif_chiral_cntrs = len(
self.mol_orig_frm_inp_smi.chiral_cntrs_w_unasignd()
)
# None of the mols derived to date, if present, are accurate.
self.mols = []
def parallel_add_H(contnr, protonation_settings):
"""Creates alternate ionization variants for a given molecule container.
This is the function that gets fed into the parallelizer.
:param contnr: The molecule container.
:type contnr: MolContainer.MolContainer
:param protonation_settings: Protonation settings to pass to Dimorphite-DL.
:type protonation_settings: dict
:return: [description]
:rtype: [type]
"""
# Make sure the canonical SMILES is actually a string.
if type(contnr.orig_smi_canonical) != str:
Utils.log("container.orig_smi_canonical: " + contnr.orig_smi_canonical)
Utils.log("type container.orig_smi_canonical: " +
str(type(contnr.orig_smi_canonical)))
Utils.exception("container.orig_smi_canonical: " +
contnr.orig_smi_canonical)
# Add the SMILES string to the protonation parameters.
protonation_settings["smiles"] = contnr.orig_smi_canonical
# Protonate the SMILESstring. This is Dimorphite-DL.
smis = Protonate(protonation_settings)
# Convert the protonated SMILES strings into a list of rdkit molecule
# objects.
rdkit_mols = [Chem.MolFromSmiles(smi.strip()) for smi in smis]
# Convert from rdkit mols to MyMol.MyMol.
addH_mols = [MyMol.MyMol(mol) for mol in rdkit_mols if mol is not None]
# Remove MyMols with odd substructures.
addH_mols = [
mol for mol in addH_mols if mol.remove_bizarre_substruc() is False
]
# I once saw it add a "C+"" here. So do a secondary check at this point to
# make sure it's valid. Recreate the list, moving new MyMol.MyMol objects
# into the return_values list.
return_values = []
orig_mol = contnr.mol_orig_frm_inp_smi
for Hm in addH_mols:
Hm.inherit_contnr_props(contnr)
Hm.genealogy = orig_mol.genealogy[:]
Hm.name = orig_mol.name
if Hm.smiles() != orig_mol.smiles():
Hm.genealogy.append(Hm.smiles(True) + " (protonated)")
return_values.append(Hm)
return return_values
def __init__(self, smiles, name, index, properties):
"""The constructor.
:param smiles: A list of SMILES strings.
:type smiles: str
:param name: The name of the molecule.
:type name: str
:param index: The index of this MolContainer in the main MolContainer
list.
:type index: int
:param properties: A dictionary of properties from the sdf.
:type properties: dict
"""
# Set some variables are set on the container level (not the MyMol
# level)
self.contnr_idx = index
self.contnr_idx_orig = index # Because if some circumstances (mpi),
# might be reset. But good to have
# original for filename output.
self.orig_smi = smiles
self.orig_smi_deslt = smiles # initial assumption
self.mols = []
self.name = name
self.properties = properties
self.mol_orig_frm_inp_smi = MyMol.MyMol(smiles, name)
self.mol_orig_frm_inp_smi.contnr_idx = self.contnr_idx
self.frgs = "" # For caching.
# Save the original canonical smiles
self.orig_smi_canonical = self.mol_orig_frm_inp_smi.smiles()
# Get the number of nonaromatic rings
self.num_nonaro_rngs = len(
self.mol_orig_frm_inp_smi.get_idxs_of_nonaro_rng_atms()
)
# Get the number of chiral centers, assigned
self.num_specif_chiral_cntrs = len(
self.mol_orig_frm_inp_smi.chiral_cntrs_only_asignd()
)
# Also get the number of chiral centers, unassigned
self.num_unspecif_chiral_cntrs = len(
self.mol_orig_frm_inp_smi.chiral_cntrs_w_unasignd()
)
# Get the non-acidic carbon-hydrogen footprint.
self.carbon_hydrogen_count = self.mol_orig_frm_inp_smi.count_hyd_bnd_to_carb()
def desalter(contnr):
"""Desalts molecules in a molecule container.
:param contnr: The molecule container.
:type contnr: MolContainer.MolContainer
:return: A molecule object.
:rtype: MyMol.MyMol
"""
# Split it into fragments
frags = contnr.get_frags_of_orig_smi()
if len(frags) == 1:
# It's only got one fragment, so default assumption that
# orig_smi = orig_smi_deslt is correct.
return contnr.mol_orig_frm_inp_smi
else:
Utils.log(
"\tMultiple fragments found in "
+ contnr.orig_smi
+ " ("
+ contnr.name
+ ")"
)
# Find the biggest fragment
num_heavy_atoms = []
num_heavy_atoms_to_frag = {}
for i, f in enumerate(frags):
num = f.GetNumHeavyAtoms()
num_heavy_atoms.append(num)
num_heavy_atoms_to_frag[num] = f
max_num = max(num_heavy_atoms)
biggest_frag = num_heavy_atoms_to_frag[max_num]
# Return info about that biggest fragment.
new_mol = MyMol.MyMol(biggest_frag)
new_mol.contnr_idx = contnr.contnr_idx
new_mol.name = contnr.name
new_mol.genealogy = contnr.mol_orig_frm_inp_smi.genealogy
new_mol.make_mol_frm_smiles_sanitze() # Need to update the mol.
return new_mol
def parallel_get_chiral(mol, max_variants_per_compound, thoroughness):
"""A parallelizable function for enumerating chiralities.
:param mol: The input molecule.
:type mol: MyMol.MyMol
:param max_variants_per_compound: To control the combinatorial explosion,
only this number of variants (molecules) will be advanced to the next
step.
:type max_variants_per_compound: int
:param thoroughness: How many molecules to generate per variant (molecule)
retained, for evaluation. For example, perhaps you want to advance five
molecules (max_variants_per_compound = 5). You could just generate five
and advance them all. Or you could generate ten and advance the best
five (so thoroughness = 2). Using thoroughness > 1 increases the
computational expense, but it also increases the chances of finding good
molecules.
:type thoroughness: int
:return: A list of MyMol.MyMol objects.
:rtype: list
"""
# Get all chiral centers that aren't assigned explicitly in the input
# molecules.
unasignd = [p[0] for p in mol.chiral_cntrs_w_unasignd() if p[1] == "?"]
num = len(unasignd)
# Get all possible chiral assignments. If the chirality is specified,
# retain it.
results = []
if num == 0:
# There are no unspecified chiral centers, so just keep existing.
results.append(mol)
return results
elif num == 1:
# There's only one chiral center.
options = ["R", "S"]
else:
# There are multiple chiral centers.
starting = [["R"], ["S"]]
options = [["R"], ["S"]]
for i in range(num - 1):
if len(options) > thoroughness * max_variants_per_compound:
# Unfortunately, this section lends itself to a combinatorial
# explosion if there are many chiral centers. Necessary to
# control that or it can become problematic. So truncate early
# if you already have a enough (so some will unfortunately
# never be evaluated).
break
options = list(itertools.product(options, starting))
options = [list(itertools.chain(c[0], c[1])) for c in options]
# Let the user know the number of chiral centers.
Utils.log(
"\t"
+ mol.smiles(True)
+ " ("
+ mol.name
+ ") has "
# + str(len(options))
+ str(2 ** num)
+ " enantiomers when chiral centers with "
+ "no specified chirality are systematically varied."
)
# Randomly select a few of the chiral combinations to examine. This is to
# reduce the potential combinatorial explosion.
num_to_keep_initially = thoroughness * max_variants_per_compound
options = Utils.random_sample(options, num_to_keep_initially, "")
# Go through the chirality combinations and make a molecule with that
# chirality.
for option in options:
# Copy the initial rdkit molecule.
a_rd_mol = copy.copy(mol.rdkit_mol)
# Set its chirality.
for idx, chiral in zip(unasignd, option):
if chiral == "R":
a_rd_mol.GetAtomWithIdx(idx).SetChiralTag(
Chem.rdchem.ChiralType.CHI_TETRAHEDRAL_CW
)
elif chiral == "S":
a_rd_mol.GetAtomWithIdx(idx).SetChiralTag(
Chem.rdchem.ChiralType.CHI_TETRAHEDRAL_CCW
)
# Make a new MyMol.MyMol object from that rdkit molecule.
new_mol = MyMol.MyMol(a_rd_mol)
# Add the new molecule to the list of results, if it does not have a
# bizarre substructure.
if not new_mol.remove_bizarre_substruc():
new_mol.contnr_idx = mol.contnr_idx
new_mol.name = mol.name
new_mol.genealogy = mol.genealogy[:]
new_mol.genealogy.append(new_mol.smiles(True) + " (chirality)")
results.append(new_mol)
# Return the results.
return results
def parallel_get_double_bonded(mol, max_variants_per_compound, thoroughness):
"""A parallelizable function for enumerating double bonds.
:param mol: The molecule with a potentially unspecified double bond.
:type mol: MyMol.MyMol
:param max_variants_per_compound: To control the combinatorial explosion,
only this number of variants (molecules) will be advanced to the next
step.
:type max_variants_per_compound: int
:param thoroughness: How many molecules to generate per variant (molecule)
retained, for evaluation. For example, perhaps you want to advance five
molecules (max_variants_per_compound = 5). You could just generate five
and advance them all. Or you could generate ten and advance the best
five (so thoroughness = 2). Using thoroughness > 1 increases the
computational expense, but it also increases the chances of finding good
molecules.
:type thoroughness: int
:return: [description]
:rtype: [type]
"""
# For this to work, you need to have explicit hydrogens in place.
mol.rdkit_mol = Chem.AddHs(mol.rdkit_mol)
# Get all double bonds that don't have defined stereochemistry. Note that
# these are the bond indexes, not the atom indexes.
unasignd_dbl_bnd_idxs = mol.get_double_bonds_without_stereochemistry()
if len(unasignd_dbl_bnd_idxs) == 0:
# There are no unassigned double bonds, so move on.
return [mol]
# Throw out any bond that is in a small ring.
unasignd_dbl_bnd_idxs = [
i
for i in unasignd_dbl_bnd_idxs
if not mol.rdkit_mol.GetBondWithIdx(i).IsInRingSize(3)
]
unasignd_dbl_bnd_idxs = [
i
for i in unasignd_dbl_bnd_idxs
if not mol.rdkit_mol.GetBondWithIdx(i).IsInRingSize(4)
]
unasignd_dbl_bnd_idxs = [
i
for i in unasignd_dbl_bnd_idxs
if not mol.rdkit_mol.GetBondWithIdx(i).IsInRingSize(5)
]
unasignd_dbl_bnd_idxs = [
i
for i in unasignd_dbl_bnd_idxs
if not mol.rdkit_mol.GetBondWithIdx(i).IsInRingSize(6)
]
unasignd_dbl_bnd_idxs = [
i
for i in unasignd_dbl_bnd_idxs
if not mol.rdkit_mol.GetBondWithIdx(i).IsInRingSize(7)
]
# Previously, I fully enumerated all double bonds. When there are many
# such bonds, that leads to a combinatorial explosion that causes problems
# in terms of speed and memory. Now, enumerate only enough bonds to make
# sure you generate at least thoroughness * max_variants_per_compound.
unasignd_dbl_bnd_idxs_orig_count = len(unasignd_dbl_bnd_idxs)
num_bonds_to_keep = int(math.ceil(math.log(thoroughness * max_variants_per_compound, 2)))
random.shuffle(unasignd_dbl_bnd_idxs)
unasignd_dbl_bnd_idxs = sorted(unasignd_dbl_bnd_idxs[:num_bonds_to_keep])
# Get a list of all the single bonds that come off each double-bond atom.
all_sngl_bnd_idxs = set([])
dbl_bnd_count = 0
for dbl_bnd_idx in unasignd_dbl_bnd_idxs:
bond = mol.rdkit_mol.GetBondWithIdx(dbl_bnd_idx)
atom1 = bond.GetBeginAtom()
atom1_bonds = atom1.GetBonds()
if len(atom1_bonds) == 1:
# The only bond is the one you already know about. So don't save.
continue
atom2 = bond.GetEndAtom()
atom2_bonds = atom2.GetBonds()
if len(atom2_bonds) == 1:
# The only bond is the one you already know about. So don't save.
continue
dbl_bnd_count = dbl_bnd_count + 1
# Suffice it to say, RDKit does not deal with cis-trans isomerization
# in an intuitive way...
idxs_of_other_bnds_frm_atm1 = [b.GetIdx() for b in atom1.GetBonds()]
idxs_of_other_bnds_frm_atm1.remove(dbl_bnd_idx)
idxs_of_other_bnds_frm_atm2 = [b.GetIdx() for b in atom2.GetBonds()]
idxs_of_other_bnds_frm_atm2.remove(dbl_bnd_idx)
all_sngl_bnd_idxs |= set(idxs_of_other_bnds_frm_atm1)
all_sngl_bnd_idxs |= set(idxs_of_other_bnds_frm_atm2)
# Now come up with all possible up/down combinations for those bonds.
all_sngl_bnd_idxs = list(all_sngl_bnd_idxs)
all_atom_config_options = list(
itertools.product([True, False], repeat=len(all_sngl_bnd_idxs))
)
# Let the user know.
if dbl_bnd_count > 0:
Utils.log(
"\t"
+ mol.smiles(True)
+ " has "
# + str(dbl_bnd_count)
+ str(
# Not exactly right, I think, because should be dbl_bnd_count, but ok.
unasignd_dbl_bnd_idxs_orig_count
)
+ " double bond(s) with unspecified stereochemistry."
)
# Go through and consider each of the retained combinations.
smiles_to_consider = set([])
for atom_config_options in all_atom_config_options:
# Make a copy of the original RDKit molecule.
a_rd_mol = copy.copy(mol.rdkit_mol)
# a_rd_mol = Chem.MolFromSmiles(mol.smiles())
for bond_idx, direc in zip(all_sngl_bnd_idxs, atom_config_options):
# Always done with reference to the atom in the double bond.
if direc:
a_rd_mol.GetBondWithIdx(bond_idx).SetBondDir(Chem.BondDir.ENDUPRIGHT)
else:
a_rd_mol.GetBondWithIdx(bond_idx).SetBondDir(Chem.BondDir.ENDDOWNRIGHT)
# Assign the StereoChemistry. Required to actually set it.
a_rd_mol.ClearComputedProps()
Chem.AssignStereochemistry(a_rd_mol, force=True)
# Add to list of ones to consider
try:
smiles_to_consider.add(
Chem.MolToSmiles(a_rd_mol, isomericSmiles=True, canonical=True)
)
except:
# Some molecules still give troubles. Unfortunate, but these are
# rare cases. Let's just skip these. Example:
# CN1C2=C(C=CC=C2)C(C)(C)[C]1=[C]=[CH]C3=CC(=C(O)C(=C3)I)I
continue
# Remove ones that don't have "/" or "\". These are not real enumerated ones.
smiles_to_consider = [s for s in smiles_to_consider if "/" in s or "\\" in s]
# Get the maximum number of / + \ in any string.
cnts = [s.count("/") + s.count("\\") for s in smiles_to_consider]
if len(cnts) == 0:
# There are no appropriate double bonds. Move on...
return [mol]
max_cnts = max(cnts)
# Only keep those with that same max count. The others have double bonds
# that remain unspecified.
smiles_to_consider = [
s[0] for s in zip(smiles_to_consider, cnts) if s[1] == max_cnts
]
results = []
for smile_to_consider in smiles_to_consider:
# Make a new MyMol.MyMol object with the specified smiles.
new_mol = MyMol.MyMol(smile_to_consider)
if new_mol.can_smi != False and new_mol.can_smi != None:
# Sometimes you get an error if there's a bad structure otherwise.
# Add the new molecule to the list of results, if it does not have
# a bizarre substructure.
if not new_mol.remove_bizarre_substruc():
new_mol.contnr_idx = mol.contnr_idx
new_mol.name = mol.name
new_mol.genealogy = mol.genealogy[:]
new_mol.genealogy.append(
new_mol.smiles(True) + " (cis-trans isomerization)"
)
results.append(new_mol)
# Return the results.
return results
def parallel_make_taut(contnr, mol_index, max_variants_per_compound):
"""Makes alternate tautomers for a given molecule container. This is the
function that gets fed into the parallelizer.
:param contnr: The molecule container.
:type contnr: MolContainer.MolContainer
:param mol_index: The molecule index.
:type mol_index: int
:param max_variants_per_compound: To control the combinatorial explosion,
only this number of variants (molecules) will be advanced to the next
step.
:type max_variants_per_compound: int
:return: A list of MyMol.MyMol objects, containing the alternate
tautomeric forms.
:rtype: list
"""
# Get the MyMol.MyMol within the molecule container corresponding to the
# given molecule index.
mol = contnr.mols[mol_index]
# Create a temporary RDKit mol object, since that's what MolVS works with.
# TODO: There should be a copy function
m = MyMol.MyMol(mol.smiles()).rdkit_mol
# For tautomers to work, you need to not have any explicit hydrogens.
m = Chem.RemoveHs(m)
# Make sure it's not None.
if m is None:
Utils.log("\tCould not generate tautomers for " + contnr.orig_smi +
". I'm deleting it.")
return
# Molecules should be kekulized already, but let's double check that.
# Because MolVS requires kekulized input.
Chem.Kekulize(m)
m = MOH.check_sanitization(m)
if m is None:
return None
# Limit to max_variants_per_compound tauts. Note that another batch could
# add more, so you'll need to once again trim to this number later. But
# this could at least help prevent the combinatorial explosion at this
# stage.
enum = tautomer.TautomerEnumerator(max_tautomers=max_variants_per_compound)
tauts_rdkit_mols = enum.enumerate(m)
# Make all those tautomers into MyMol objects.
tauts_mols = [MyMol.MyMol(m) for m in tauts_rdkit_mols]
# Keep only those that have reasonable substructures.
tauts_mols = [
t for t in tauts_mols if t.remove_bizarre_substruc() == False
]
# If there's more than one, let the user know that.
if len(tauts_mols) > 1:
Utils.log("\t" + mol.smiles(True) + " has tautomers.")
# Now collect the final results.
results = []
for tm in tauts_mols:
tm.inherit_contnr_props(contnr)
tm.genealogy = mol.genealogy[:]
tm.name = mol.name
if tm.smiles() != mol.smiles():
tm.genealogy.append(tm.smiles(True) + " (tautomer)")
results.append(tm)
return results