Hacking SMILES to get correct Hydrogens on molecule fragments with RDkit

fix_hydrogens.py

import numpy as np
from rdkit import Chem

# examples
# C=[NH+]C
component = "C=[NH+]"
component = "[NH+]C"

# C[S+](C)C
component = "C[S+]"


print(component)

mc = Chem.MolFromSmiles(component)
n_atoms = mc.GetNumAtoms()
n_bonds = len(mc.GetBonds())

charges = np.zeros(n_atoms, dtype=int)

for idx in range(n_atoms):
    atom = mc.GetAtomWithIdx(idx)
    charges[idx] = atom.GetFormalCharge()
    atom.SetNumExplicitHs(0)
    atom.SetFormalCharge(0)

component = Chem.MolToSmiles(mc, canonical=False)
component = component.replace("[", "").replace("]","")

mc = Chem.MolFromSmiles(component)

for idx, charge in zip(range(n_atoms), charges):
    atom = mc.GetAtomWithIdx(idx)
    atom.SetFormalCharge(charge)

component = Chem.MolToSmiles(mc)
print(component)

Here's an alternative that uses RDKit functionality and doesn't rely on SMILES hacking:

In [15]: def adjustMolHs(mol):
    ...:     organicSubset = (5, 6, 7, 8, 9, 15, 16, 17, 35, 53)
    ...:     for at in mol.GetAtoms():
    ...:         if at.GetAtomicNum() not in organicSubset:
    ...:             continue
    ...:         at.SetNoImplicit(False)
    ...:         at.SetNumExplicitHs(0)
    ...:         at.SetNumRadicalElectrons(0)
    ...:     Chem.SanitizeMol(mol)
    ...:     

In [16]: m = Chem.MolFromSmiles('C=[NH+]')

In [17]: adjustMolHs(m)

In [18]: Chem.MolToSmiles(m)
Out[18]: 'C=[NH2+]'

In [19]: m2 = Chem.MolFromSmiles('[NH+]C')

In [20]: adjustMolHs(m2)

In [21]: Chem.MolToSmiles(m2)
Out[21]: 'C[NH3+]'