def _to_sire_mapping(mapping):
"""Internal function to convert a regular mapping to Sire AtomIdx format.
Parameters
----------
mapping : {int:int}
The regular mapping.
Returns
-------
sire_mapping : {Sire.Mol.AtomIdx:Sire.Mol.AtomIdx}
The Sire mapping.
"""
sire_mapping = {}
# Convert the mapping to AtomIdx key:value pairs.
for idx0, idx1 in mapping.items():
# Early exit if the mapping is already the correct format.
if type(idx0) is _SireMol.AtomIdx:
return mapping
else:
sire_mapping[_SireMol.AtomIdx(idx0)] = _SireMol.AtomIdx(idx1)
return sire_mapping
def _score_sire_mappings(molecule0, molecule1, sire_mappings, prematch,
scoring_function, property_map0, property_map1):
"""Internal function to score atom mappings based on the root mean squared
displacement (RMSD) between mapped atoms in two molecules. Optionally,
molecule0 can first be aligned to molecule1 based on the mapping prior
to computing the RMSD. The function returns the mappings sorted based
on their score from best to worst, along with a list containing the
scores for each mapping.
Parameters
----------
molecule0 : Sire.Molecule.Molecule
The first molecule (Sire representation).
molecule0 : Sire.Molecule.Molecule
The second molecule (Sire representation).
sire_mappings : [{}]
The list of mappings generated by Sire.
prematch : dict
A dictionary of atom mappings that must be included in the match.
scoring_function : str
The RMSD scoring function.
property_map0 : dict
A dictionary that maps "properties" in molecule0 to their user
defined values. This allows the user to refer to properties
with their own naming scheme, e.g. { "charge" : "my-charge" }
property_map1 : dict
A dictionary that maps "properties" in molecule1 to their user
defined values.
Returns
-------
mapping, scores : ([dict], list)
The ranked mappings and corresponding scores.
"""
# Make sure to re-map the coordinates property in both molecules, otherwise
# the move and align functions from Sire will not work.
prop0 = property_map0.get("coordinates", "coordinates")
prop1 = property_map1.get("coordinates", "coordinates")
if prop0 != "coordinates":
molecule0 = molecule0.edit().setProperty(
"coordinates", molecule0.property(prop0)).commit()
if prop1 != "coordinates":
molecule1 = molecule1.edit().setProperty(
"coordinates", molecule1.property(prop1)).commit()
# Initialise a list to hold the mappings.
mappings = []
# Initialise a list of to hold the score for each mapping.
scores = []
# Loop over all of the mappings.
for mapping in sire_mappings:
# Check that the mapping contains the pre-match.
is_valid = True
for idx0, idx1 in prematch.items():
# Pre-match isn't found, return to top of loop.
if _SireMol.AtomIdx(idx0) not in mapping or mapping[
_SireMol.AtomIdx(idx0)] != _SireMol.AtomIdx(idx1):
is_valid = False
break
if is_valid:
# Rigidly align molecule0 to molecule1 based on the mapping.
if scoring_function == "RMSDALIGN":
try:
molecule0 = molecule0.move().align(
molecule1,
_SireMol.AtomResultMatcher(mapping)).molecule()
except Exception as e:
msg = "Failed to align molecules when scoring based on mapping: %r" % mapping
if _isVerbose():
raise _AlignmentError(msg) from e
else:
raise _AlignmentError(msg) from None
# Flexibly align molecule0 to molecule1 based on the mapping.
elif scoring_function == "RMSDFLEXALIGN":
molecule0 = flexAlign(_Molecule(molecule0),
_Molecule(molecule1),
_from_sire_mapping(mapping),
property_map0=property_map0,
property_map1=property_map1)._sire_object
# Append the mapping to the list.
mappings.append(_from_sire_mapping(mapping))
# We now compute the RMSD between the coordinates of the matched atoms
# in molecule0 and molecule1.
# Initialise lists to hold the coordinates.
c0 = []
c1 = []
# Loop over each atom index in the map.
for idx0, idx1 in mapping.items():
# Append the coordinates of the matched atom in molecule0.
c0.append(molecule0.atom(idx0).property("coordinates"))
# Append the coordinates of atom in molecule1 to which it maps.
c1.append(molecule1.atom(idx1).property("coordinates"))
# Compute the RMSD between the two sets of coordinates.
scores.append(_SireMaths.getRMSD(c0, c1))
# No mappings were found.
if len(mappings) == 0:
if len(prematch) == 0:
return ([{}], [])
else:
return ([prematch], [])
# Sort the scores and return the sorted keys. (Smaller RMSD is best)
keys = sorted(range(len(scores)), key=lambda k: scores[k])
# Sort the mappings.
mappings = [mappings[x] for x in keys]
# Sort the scores and convert to Angstroms.
scores = [scores[x] * _Units.Length.angstrom for x in keys]
# Return the sorted mappings and their scores.
return (mappings, scores)
def _score_rdkit_mappings(molecule0, molecule1, rdkit_molecule0,
rdkit_molecule1, mcs_smarts, prematch,
scoring_function, property_map0, property_map1):
"""Internal function to score atom mappings based on the root mean squared
displacement (RMSD) between mapped atoms in two molecules. Optionally,
molecule0 can first be aligned to molecule1 based on the mapping prior
to computing the RMSD. The function returns the mappings sorted based
on their score from best to worst, along with a list containing the
scores for each mapping.
Parameters
----------
molecule0 : Sire.Molecule.Molecule
The first molecule (Sire representation).
molecule0 : Sire.Molecule.Molecule
The second molecule (Sire representation).
rdkit_mol0 : RDKit.Chem.Mol
The first molecule (RDKit representation).
rdkit_mol1 : RDKit.Chem.Mol
The second molecule (RDKit representation).
mcs_smarts : RDKit.Chem.MolFromSmarts
The smarts string representing the maximum common substructure of
the two molecules.
prematch : dict
A dictionary of atom mappings that must be included in the match.
scoring_function : str
The RMSD scoring function.
property_map0 : dict
A dictionary that maps "properties" in molecule0 to their user
defined values. This allows the user to refer to properties
with their own naming scheme, e.g. { "charge" : "my-charge" }
property_map1 : dict
A dictionary that maps "properties" in molecule1 to their user
defined values.
Returns
-------
mapping, scores : ([dict], list)
The ranked mappings and corresponding scores.
"""
# Adapted from FESetup: https://github.com/CCPBioSim/fesetup
# Make sure to re-map the coordinates property in both molecules, otherwise
# the move and align functions from Sire will not work.
prop0 = property_map0.get("coordinates", "coordinates")
prop1 = property_map1.get("coordinates", "coordinates")
if prop0 != "coordinates":
molecule0 = molecule0.edit().setProperty(
"coordinates", molecule0.property(prop0)).commit()
if prop1 != "coordinates":
molecule1 = molecule1.edit().setProperty(
"coordinates", molecule1.property(prop1)).commit()
# Get the set of matching substructures in each molecule. For some reason
# setting uniquify to True removes valid matches, in some cases even the
# best match! As such, we set uniquify to False and account ignore duplicate
# mappings in the code below.
matches0 = rdkit_molecule0.GetSubstructMatches(mcs_smarts,
uniquify=False,
maxMatches=1000,
useChirality=False)
matches1 = rdkit_molecule1.GetSubstructMatches(mcs_smarts,
uniquify=False,
maxMatches=1000,
useChirality=False)
# Swap the order of the matches.
if len(matches0) < len(matches1):
matches0, matches1 = matches1, matches0
is_swapped = True
else:
is_swapped = False
# Initialise a list to hold the mappings.
mappings = []
# Initialise a list of to hold the score for each mapping.
scores = []
# Loop over all matches from mol0.
for x in range(len(matches0)):
match0 = matches0[x]
# Loop over all matches from mol1.
for y in range(len(matches1)):
match1 = matches1[y]
# Initialise the mapping for this match.
mapping = {}
sire_mapping = {}
# Loop over all atoms in the match.
for i, idx0 in enumerate(match0):
idx1 = match1[i]
# Add to the mapping.
if is_swapped:
mapping[idx1] = idx0
sire_mapping[_SireMol.AtomIdx(idx1)] = _SireMol.AtomIdx(
idx0)
else:
mapping[idx0] = idx1
sire_mapping[_SireMol.AtomIdx(idx0)] = _SireMol.AtomIdx(
idx1)
# This is a new mapping:
if not mapping in mappings:
# Check that the mapping contains the pre-match.
is_valid = True
for idx0, idx1 in prematch.items():
# Pre-match isn't found, return to top of loop.
if idx0 not in mapping or mapping[idx0] != idx1:
is_valid = False
break
if is_valid:
# Rigidly align molecule0 to molecule1 based on the mapping.
if scoring_function == "RMSDALIGN":
try:
molecule0 = molecule0.move().align(
molecule1,
_SireMol.AtomResultMatcher(
sire_mapping)).molecule()
except Exception as e:
msg = "Failed to align molecules when scoring based on mapping: %r" % mapping
if _isVerbose():
raise _AlignmentError(msg) from e
else:
raise _AlignmentError(msg) from None
# Flexibly align molecule0 to molecule1 based on the mapping.
elif scoring_function == "RMSDFLEXALIGN":
molecule0 = flexAlign(
_Molecule(molecule0),
_Molecule(molecule1),
mapping,
property_map0=property_map0,
property_map1=property_map1)._sire_object
# Append the mapping to the list.
mappings.append(mapping)
# We now compute the RMSD between the coordinates of the matched atoms
# in molecule0 and molecule1.
# Initialise lists to hold the coordinates.
c0 = []
c1 = []
# Loop over each atom index in the map.
for idx0, idx1 in sire_mapping.items():
# Append the coordinates of the matched atom in molecule0.
c0.append(molecule0.atom(idx0).property("coordinates"))
# Append the coordinates of atom in molecule1 to which it maps.
c1.append(molecule1.atom(idx1).property("coordinates"))
# Compute the RMSD between the two sets of coordinates.
scores.append(_SireMaths.getRMSD(c0, c1))
# No mappings were found.
if len(mappings) == 0:
if len(prematch) == 0:
return ([{}], [])
else:
return ([prematch], [])
# Sort the scores and return the sorted keys. (Smaller RMSD is best)
keys = sorted(range(len(scores)), key=lambda k: scores[k])
# Sort the mappings.
mappings = [mappings[x] for x in keys]
# Sort the scores and convert to Angstroms.
scores = [scores[x] * _Units.Length.angstrom for x in keys]
# Return the sorted mappings and their scores.
return (mappings, scores)