Skip to content

Commit

Permalink
Merge pull request #73 from xiaoruiDong/fix_mol
Browse files Browse the repository at this point in the history 
bug fixes and mol_comparison
  • Loading branch information
@xiaoruiDong
xiaoruiDong committed Oct 20, 2023
2 parents ee3663d + cc64388 commit 26c1674
Show file tree
Hide file tree
Showing 8 changed files with 540 additions and 272 deletions.
40 changes: 29 additions & 11 deletions rdmc/fix.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -13,24 +13,42 @@


DEFAULT_REMEDIES = [
# Remedy 1 - Carbon monoxide: [C]=O to [C-]#[O+]
rdChemReactions.ReactionFromSmarts(
"[O+0-0v2X1:1]=[C+0-0v2X1:2]>>[O+1v3X1:1]#[C-1v3X1:2]"
),
# Remedy 2 - Oxygen Molecule: O=O to [O]-[O]
rdChemReactions.ReactionFromSmarts(
"[O+0-0v2X1:1]=[O+0-0v2X1:2]>>[O+0-0v1X1:1]-[O+0-0v1X1:2]"
),
# Remedy 3 - isocyanide: R[N]#[C] to R[N+]#[C-]
rdChemReactions.ReactionFromSmarts(
"[N+0-0v4X2:1]#[C+0-0v3X1:2]>>[N+v4X2:1]#[C-v3X1:2]"
), # R1N#[C.] to R1[N+]#[C-]
),
# Remedy 4 - amine radical: RC(R)-N(R)(R)R to R[C-](R)-[N+](R)(R)R
rdChemReactions.ReactionFromSmarts(
"[N+0-0v4X4:1]-[C+0-0v3X3:2]>>[N+1v4X4:1]-[C-1v3X3:2]"
),
# Remedy 5 - amine radical: RN(R)=C to RN(R)-[C]
rdChemReactions.ReactionFromSmarts(
"[N+0-0v4X3:1]=[C+0-0v4X3:2]>>[N+0-0v3X3:1]-[C+0-0v3X3:2]"
), # R1N(R2)=C(R3)R4 to R1N(R2)-[C.](R3)R4
),
# Remedy 5 - quintuple C bond, usually due to RC(=O)=O: R=C(R)=O to R=[C+](R)-[O-]
rdChemReactions.ReactionFromSmarts(
"[C+0-0v5X3:1]=[O+0-0v2X1:2]>>[C+0-0v4X3:1]-[O+0-0v1X1:2]"
), # R1=C(R2)=O to R1=C(R2)-[O.]
rdChemReactions.ReactionFromSmarts(
"[C+0-0v3X3:1]-[N+0-0v4X4:2]-[C+0-0v4X3:3]=[O+0-0v2X1:4]>>[C-1v3X3:1]-[N+1v4X4:2]-[C+0-0v4X3:3]=[O+0-0v2X1:4]"
), # R1C(R2)N(R3)(R4)C(R5)=O to R1[C-](R2)[N+](R3)(R4)C(R5)=O
),
# Remedy 6 - amine oxide: RN(R)(R)-O to R[N+](R)(R)-[O-]
rdChemReactions.ReactionFromSmarts(
"[N+0-0v4X4:1]-[O+0-0v1X1:2]>>[N+1v4X4:1]-[O-1v1X1:2]"
), # R1N(R2)(R3)[O.] to R1[N+](R2)(R3)[O-]
),
# Remedy 7 - criegee like molecule: RN(R)(R)-C(R)(R)=O to R[N+](R)(R)-[C](R)(R)-[O-]
rdChemReactions.ReactionFromSmarts(
"[N+0-0v4X4:1]-[C+0-0v4X3:2]=[O+0-0v2X1:3]>>[N+1v4X4:1]-[C+0-0v3X3:2]-[O-1v1X1:3]"
), # R1N(R2)(R3)C(R4)=O to R1[N+](R2)(R3)[C.](R4)[O-]
),
# Remedy 8 - criegee like molecule: RN(R)(R)-C(R)(R)=O to R[N+](R)(R)-[C](R)(R)-[O-]
rdChemReactions.ReactionFromSmarts(
"[N+0-0v4X4:1]-[C+0-0v3X3:2]-[O+0-0v1X1:3]>>[N+1v4X4:1]-[C+0-0v3X3:2]-[O-1v1X1:3]"
),
]


Expand Down Expand Up @@ -163,7 +181,7 @@ def fix_mol_by_remedies(


def fix_mol_spin_multiplicity(
mol: 'RDKitMol',
mol: "RDKitMol",
mult: int,
):
"""
Expand All @@ -183,7 +201,7 @@ def fix_mol(
remedies: List["ChemicalReaction"] = DEFAULT_REMEDIES,
max_attempts: int = 10,
sanitize: bool = True,
fix_spin_multiplicity: bool = True,
fix_spin_multiplicity: bool = False,
mult: int = 1,
) -> "RDKitMol":
"""
Expand All @@ -198,7 +216,7 @@ def fix_mol(
Defaults to ``10``.
sanitize (bool, optional): Whether to sanitize the molecule after the fix. Defaults to ``True``.
fix_spin_multiplicity (bool, optional): Whether to fix the spin multiplicity of the molecule.
Defaults to ``True``.
Defaults to ``False``.
mult (int, optional): The desired spin multiplicity. Defaults to ``1``.
Only used when ``fix_spin_multiplicity`` is ``True``.
Expand Down
191 changes: 11 additions & 180 deletions rdmc/mol.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -5,11 +5,12 @@
This module provides class and methods for dealing with RDKit RWMol, Mol.
"""

from collections import Counter
import copy
from itertools import combinations
from itertools import product as cartesian_product
import traceback
from typing import Iterable, List, Optional, Sequence, Union
from typing import Dict, Iterable, List, Optional, Sequence, Tuple, Union
import pathlib

import numpy as np
Expand Down Expand Up @@ -825,6 +826,15 @@ def GetElementSymbols(self) -> List[str]:
"""
return get_element_symbols(self.GetAtomicNumbers())

def GetElementCounts(self) -> Dict[str, int]:
"""
Get the element counts of the molecules.
Returns:
dict: A dictionary of element counts.
"""
return dict(Counter(self.GetElementSymbols()))

def GetAtomMasses(self) -> List[float]:
"""
Get the mass of each atom. The order is consistent with the atom indexes.
Expand Down Expand Up @@ -1940,182 +1950,3 @@ def generate_vdw_mat(rd_mol,
(vdw_radii[atom1.GetAtomicNum()]
+ vdw_radii[atom2.GetAtomicNum()])
return vdw_mat


def generate_radical_resonance_structures(mol: RDKitMol,
unique: bool = True,
consider_atommap: bool = False,
kekulize: bool = False):
"""
Generate resonance structures for a radical molecule. RDKit by design doesn't work
for radical resonance. The approach is a temporary workaround by replacing radical electrons by positive
charges and generating resonance structures by RDKit ResonanceMolSupplier.
Currently, this function only works for neutral radicals.
Known issues:
- Phenyl radical only generate one resonance structure when ``kekulize=True``, expecting 2.
Args:
mol (RDKitMol): A radical molecule.
unique (bool, optional): Filter out duplicate resonance structures from the list. Defaults to ``True``.
consider_atommap (bool, atommap): If consider atom map numbers in filtration duplicates.
Only effective when ``unique=True``. Defaults to ``False``.
kekulize (bool, optional): Whether to kekulize the molecule. Defaults to ``False``. As an example,
benzene have one resonance structure if not kekulized (``False``) and
two resonance structures if kekulized (``True``).
Returns:
list: a list of molecules with resonance structures.
"""
assert mol.GetFormalCharge() == 0, "The current function only works for radical species."
mol_copy = mol.Copy(quickCopy=True) # Make a copy of the original molecule

# Modify the original molecule to make it a positively charged species
recipe = {} # Used to record changes. Temporarily not used now.
for atom in mol_copy.GetAtoms():
radical_electrons = atom.GetNumRadicalElectrons()
if radical_electrons > 0: # Find a radical site
recipe[atom.GetIdx()] = radical_electrons
atom.SetFormalCharge(+radical_electrons)
atom.SetNumRadicalElectrons(0)
# Make sure conjugation is assigned
# Only assign the conjugation after changing radical sites to positively charged sites
Chem.rdmolops.SetConjugation(mol_copy._mol)

# Avoid generating certain resonance bonds
for atom in mol_copy.GetAtoms():
if (atom.GetAtomicNum() == 8 and len(atom.GetNeighbors()) > 1) or \
(atom.GetAtomicNum() == 7 and len(atom.GetNeighbors()) > 2):
# Avoid X-O-Y be part of the resonance and forms X-O.=Y
# Avoid X-N(-Y)-Z be part of the resonance and forms X=N.(-Y)(-Z)
[bond.SetIsConjugated(False) for bond in atom.GetBonds()]
mol_copy.UpdatePropertyCache() # Make sure the assignment is boardcast to atoms / bonds

# Generate Resonance Structures
flags = Chem.ALLOW_INCOMPLETE_OCTETS | Chem.UNCONSTRAINED_CATIONS
if kekulize:
flags |= Chem.KEKULE_ALL
suppl = Chem.ResonanceMolSupplier(mol_copy._mol, flags=flags)
res_mols = [RDKitMol(RWMol(mol)) for mol in suppl]

# Post-processing resonance structures
cleaned_mols = []
for res_mol in res_mols:
for atom in res_mol.GetAtoms():
# Convert positively charged species back to radical species
charge = atom.GetFormalCharge()
if charge > 0: # Find a radical site
recipe[atom.GetIdx()] = radical_electrons
atom.SetFormalCharge(0)
atom.SetNumRadicalElectrons(charge)
elif charge < 0: # Shouldn't appear, just for bug detection
raise RuntimeError('Encounter charge separation during resonance structure generation.')

# If a structure cannot be sanitized, removed it
try:
# Sanitization strategy is inspired by
# https://github.com/rdkit/rdkit/discussions/6358
flags = Chem.SanitizeFlags.SANITIZE_ALL
if kekulize:
flags ^= (Chem.SanitizeFlags.SANITIZE_KEKULIZE | Chem.SanitizeFlags.SANITIZE_SETAROMATICITY)
res_mol.Sanitize(sanitizeOps=flags)
except BaseException as e:
print(e)
# todo: make error type more specific and add a warning message
continue
if kekulize:
_unset_aromatic_flags(res_mol)
cleaned_mols.append(res_mol)

# To remove duplicate resonance structures
if unique:
cleaned_mols = get_unique_mols(cleaned_mols,
consider_atommap=consider_atommap)
for mol in cleaned_mols:
# According to
# https://github.com/rdkit/rdkit/blob/9249ca5cc840fc72ea3bb73c2ff1d71a1fbd3f47/rdkit/Chem/Draw/IPythonConsole.py#L152
# highlight info is stored in __sssAtoms
mol._mol.__setattr__('__sssAtoms', [])
return cleaned_mols


def _unset_aromatic_flags(mol):
"""
A helper function to unset aromatic flags in a molecule.
This is useful when cleaning up the molecules from resonance structure generation.
In such case, a molecule may have single-double bonds but are marked as aromatic bonds.
"""
for bond in mol.GetBonds():
if bond.GetBondType() != Chem.BondType.AROMATIC and bond.GetIsAromatic():
bond.SetIsAromatic(False)
bond.GetBeginAtom().SetIsAromatic(False)
bond.GetEndAtom().SetIsAromatic(False)
return mol


def has_matched_mol(mol: RDKitMol,
mols: List[RDKitMol],
consider_atommap: bool = False,
):
"""
Check if a molecule has a structure match in a list of molecules.
Args:
mol (RDKitMol): The target molecule.
mols (List[RDKitMol]): The list of molecules to be processed.
consider_atommap (bool, optional): If treat chemically equivalent molecules with
different atommap numbers as different molecules.
Defaults to ``False``.
Returns:
bool: if a matched molecules if found.
"""
for mol_in_list in mols:
mapping = mol_in_list.GetSubstructMatch(mol) # A tuple of atom indexes if matched
if mapping and not consider_atommap:
return True
elif mapping and mapping == tuple(range(len(mapping))):
# if identical, the mapping is always as 1,2,...,N
return True
return False


def get_unique_mols(mols: List[RDKitMol],
consider_atommap: bool = False,
same_formula: bool = False,
):
"""
Find the unique molecules from a list of molecules.
Args:
mols (list): The molecules to be processed.
consider_atommap (bool, optional): If treat chemically equivalent molecules with
different atommap numbers as different molecules.
Defaults to ``False``.
same_formula (bool, opional): If the mols has the same formula you may set it to True
to save computational time. Defaults to ``False``.
Returns:
list: A list of unique molecules.
"""
# Dictionary:
# Keys: chemical formula;
# Values: list of mols with same formula
# Use chemical formula to reduce the call of the more expensive graph substructure check
unique_formula_mol = {}

for mol in mols:

# Get the molecules with the same formula as the query molecule
form = 'same' if same_formula else Chem.rdMolDescriptors.CalcMolFormula(mol._mol)
unique_mol_list = unique_formula_mol.get(form)

if unique_mol_list and has_matched_mol(mol, unique_mol_list, consider_atommap=consider_atommap):
continue
elif unique_mol_list:
unique_formula_mol[form].append(mol)
else:
unique_formula_mol[form] = [mol]

return sum(unique_formula_mol.values(), [])
Loading

0 comments on commit 26c1674

Please sign in to comment.