def get_fingerprint_from_mol(mol: oechem.OEMol) -> Tuple[float]:
"""Utility that retrieves a molecule's fingerprint and returns it as a tuple.
Refer to :meth:`~assign_fingerprint` for how the fingerprint is stored
in the molecule.
Parameters
----------
mol : oechem.OEMol
The molecule from which to retrieve the fingerprint.
Returns
-------
Tuple[float]
A tuple containing the fingerprint.
Raises
------
ValueError
If the molecule does not contain fingerprint data.
"""
if not mol.HasData(DancePipeline.FINGERPRINT_LENGTH_NAME):
raise ValueError("Could not retrieve fingerprint length for molecule.")
length = mol.GetIntData(DancePipeline.FINGERPRINT_LENGTH_NAME)
def get_fingerprint_index(i):
name = f"{DancePipeline.FINGERPRINT_VALUE_NAME}_{i}"
if not mol.HasData(name):
raise ValueError(f"Unable to retrieve fingerprint value at index {i}")
return mol.GetDoubleData(name)
return tuple(get_fingerprint_index(i) for i in range(length))
def write_mol_to_fingerprint_file(
mol: oechem.OEMol,
properties: [danceprops.DanceProperties],
select_output_dir: str,
select_bin_size: float,
wiberg_precision: float,
):
"""Writes a molecule to its appropriate SMILES fingerprint file"""
# Some of the molecules coming in may be invalid. DanceGenerator may find
# there was an error in charge calculations, in which case the charged
# copy was not assigned to the molecule. This function checks for that.
is_valid_molecule = \
lambda mol: mol.HasData(danceprops.DANCE_CHARGED_COPY_KEY)
if not is_valid_molecule(mol):
logging.debug(f"Ignored molecule {mol.GetTitle()}")
return
charged_copy = mol.GetData(danceprops.DANCE_CHARGED_COPY_KEY)
for atom in charged_copy.GetAtoms(oechem.OEIsInvertibleNitrogen()):
tri_n = atom
break
fingerprint = danceprops.DanceFingerprint(tri_n, wiberg_precision)
# Retrieve the total bond order around the trivalent nitrogen
bond_order = danceprops.get_dance_property(mol, properties).tri_n_bond_order
# Round the total bond order down to the lowest multiple of bin_size. For
# instance, if bin_size is 0.02, and the bond_order is 2.028, it becomes
# 2.02. This works because (bond_order / self._bin_size) generates a
# multiple of the bin_size. Then floor() finds the next integer less than
# the multiple. Finally, multiplying back by bin_size obtains the nearest
# actual value.
bond_order = math.floor(bond_order / select_bin_size) * select_bin_size
filename = f"{select_output_dir}/{bond_order},{fingerprint}.smi"
with open(filename, "a") as f:
f.write(f"{oechem.OEMolToSmiles(mol)} {mol.GetTitle()}\n")
logging.debug(f"Wrote {mol.GetTitle()} to {filename}")