def graph(self, m):
from rdkit.Chem import EditableMol, Atom, rdchem
hcount = m.GetNumAtoms(False) - m.GetNumAtoms(True)
# create new molecule using single bonds only
em = EditableMol(Mol())
nbridx = [None] * m.GetNumAtoms()
iatom = 0
for atom in m.GetAtoms():
atnum = atom.GetAtomicNum()
if atnum == 1 and atom.GetIsotope() == 1:
#if atom.GetMass() > 1: pass
hcount += 1
else:
newatom = Atom(atnum)
#if atom.GetTotalDegree() == 0: newatom.SetNoImplicit(True) # otherwise [Na]. becomes [NaH].
#newatom.SetFormalCharge(atom.GetFormalCharge())
newatom.SetFormalCharge(0)
em.AddAtom(newatom)
aidx = atom.GetIdx()
nbridx[aidx] = iatom
iatom += 1
for a2 in atom.GetNeighbors():
a2idx = nbridx[a2.GetIdx()]
if a2idx != None:
em.AddBond(aidx, a2idx, rdchem.BondType.SINGLE)
#cansmi = self.cansmiles(em.GetMol())
cansmi = MolToSmiles(m, isomericSmiles=True)
#cansmi = cansmi.replace('+','').replace('-','').replace('[N]','N').replace('[O]','O').replace('[C]','C').replace('[I]','I').replace('[S]','S').replace('[P]','P').replace('[B]','B').replace('[Br]','Br').replace('[Cl]','Cl')
return "%s%s%d%+d" % (cansmi, ' H', hcount, GetFormalCharge(m))
def to_rdkit_molecule(data):
"""
MoleculeContainer to RDKit molecule object converter
"""
mol = RWMol()
mapping = {}
for n, a in data.atoms():
ra = Atom(a.atomic_number)
ra.SetAtomMapNum(n)
if a.charge:
ra.SetFormalCharge(a.charge)
if a.isotope:
ra.SetIsotope(a.isotope)
if a.is_radical:
ra.SetNumRadicalElectrons(1)
mapping[n] = mol.AddAtom(ra)
for n, m, b in data.bonds():
mol.AddBond(mapping[n], mapping[m], _bond_map[b.order])
conf = Conformer()
for n, a in data.atoms():
conf.SetAtomPosition(mapping[n], (a.x, a.y, 0))
conf.Set3D(False)
mol.AddConformer(conf)
for c in data._conformers:
conf = Conformer()
for n, xyz in c.items():
conf.SetAtomPosition(mapping[n], xyz)
mol.AddConformer(conf)
SanitizeMol(mol)
return mol
def decode(v):
"""Decode a molvector into a molecule
:param v: molvector
:result rdkit.RWMol:
"""
chunksize = atom_size + bond_chunk_size
nchunks = len(v) // chunksize
m = RWMol()
bonds = {}
for i in range(nchunks):
start = i * (atom_size + bond_chunk_size)
el, c, h, b1, o1, b2, o2, b3, o3, b4, o4 = v[start:start + chunksize]
atom = Atom(el)
atom.SetFormalCharge(c)
atom.SetNumExplicitHs(h)
atom_idx = m.AddAtom(atom)
assert atom_idx == i
for b, o in ((b1, o1), (b2, o2), (b3, o3), (b4, o4)):
if o:
to_atom = atom_idx + o
bonds[tuple(sorted((atom_idx, to_atom)))] = b
for (a1, a2), btype in bonds.items():
try:
m.AddBond(a1 % m.GetNumAtoms(), a2 % m.GetNumAtoms(),
BondType.values[btype])
except:
pass
return m
def to_rdkit_molecule(data):
"""
MoleculeContainer to RDKit molecule object converter
"""
mol = RWMol()
conf = Conformer()
mapping = {}
is_3d = False
for n, a in data.atoms():
ra = Atom(a.number)
ra.SetAtomMapNum(n)
if a.charge:
ra.SetFormalCharge(a.charge)
if a.isotope != a.common_isotope:
ra.SetIsotope(a.isotope)
if a.radical:
ra.SetNumRadicalElectrons(a.radical)
mapping[n] = m = mol.AddAtom(ra)
conf.SetAtomPosition(m, (a.x, a.y, a.z))
if a.z:
is_3d = True
if not is_3d:
conf.Set3D(False)
for n, m, b in data.bonds():
mol.AddBond(mapping[n], mapping[m], _bond_map[b.order])
mol.AddConformer(conf)
SanitizeMol(mol)
return mol
def set_atom_feat(a: Atom, key: str, val: int):
if key == 'atomic_num':
a.SetAtomicNum(val)
elif key == 'formal_charge':
a.SetFormalCharge(val)
elif key == 'chiral_tag':
a_chiral = rdchem.ChiralType.values[val]
a.SetChiralTag(a_chiral)
elif key == 'num_explicit_hs':
a.SetNumExplicitHs(val)
elif key == 'is_aromatic':
a.SetIsAromatic(bool(val))
return a
def to_rdkit_molecule(data: MoleculeContainer):
"""
MoleculeContainer to RDKit molecule object converter
"""
mol = RWMol()
mapping = {}
bonds = data._bonds
for n, a in data.atoms():
ra = Atom(a.atomic_number)
ra.SetAtomMapNum(n)
if a.charge:
ra.SetFormalCharge(a.charge)
if a.isotope:
ra.SetIsotope(a.isotope)
if a.is_radical:
ra.SetNumRadicalElectrons(1)
mapping[n] = mol.AddAtom(ra)
for n, m, b in data.bonds():
mol.AddBond(mapping[n], mapping[m], _bond_map[b.order])
for n in data._atoms_stereo:
ra = mol.GetAtomWithIdx(mapping[n])
env = bonds[n]
s = data._translate_tetrahedron_sign(n, [x for x in mapping if x in env])
ra.SetChiralTag(_chiral_ccw if s else _chiral_cw)
for nm, s in data._cis_trans_stereo.items():
n, m = nm
if m in bonds[n]: # cumulenes unsupported
nn, nm, *_ = data._stereo_cis_trans[nm]
b = mol.GetBondBetweenAtoms(mapping[n], mapping[m])
b.SetStereoAtoms(mapping[nn], mapping[nm])
b.SetStereo(_cis if s else _trans)
conf = Conformer()
for n, a in data.atoms():
conf.SetAtomPosition(mapping[n], (a.x, a.y, 0))
conf.Set3D(False)
mol.AddConformer(conf, assignId=True)
for c in data._conformers:
conf = Conformer()
for n, xyz in c.items():
conf.SetAtomPosition(mapping[n], xyz)
mol.AddConformer(conf, assignId=True)
SanitizeMol(mol)
AssignStereochemistry(mol, flagPossibleStereoCenters=True, force=True)
return mol