Skip to main content

Generating ASE Atoms object from SMILES

  • Updated

One method to generate 3D structures of organic molecules is by converting molecular information written in SMILES notation. This article explains how to execute this process in the Matlantis environment.

  1. Using Matlantis Features
  2. Using autodE for creating 5-coordinate and 6-coordinate structures

1. Using Matlantis Features

In the Matlantis environment, it is possible to generate 3D structures using Matlantis Features as shown below:

# Example of creating a Testosterone molecule
from matlantis_features.atoms import MatlantisAtoms
molec = MatlantisAtoms.from_smiles("O=C4\C=C2/[C@]([C@H]1CC[C@@]3([C@@H](O)CC[C@H]3[C@@H]1CC2)C)(C)CC4").ase_atoms
molec

The output result is as follows:

Atoms(symbols='OC9OC10H28', pbc=False)

To display the 3D structure, execute the following:

from pfcc_extras import view_ngl
view_ngl(molec, representations=["ball+stick"])

Note: However, it will fail for 5-coordinate and 6-coordinate structures.

molec = MatlantisAtoms.from_smiles("[P1](F)(F)(F)(F)(F)(F)").ase_atoms

The following error will be obtained as the result:

[00:14:42] SMILES Parse Error: syntax error while parsing: [P1](F)(F)(F)(F)(F)(F)
[00:14:42] SMILES Parse Error: Failed parsing SMILES '[P1](F)(F)(F)(F)(F)(F)' for input: '[P1](F)(F)(F)(F)(F)(F)'

---------------------------------------------------------------------------
ArgumentError Traceback (most recent call last)
/tmp/ipykernel_3167/1113753141.py in <cell line: 1>()
----> 1 molec = MatlantisAtoms.from_smiles("[P1](F)(F)(F)(F)(F)(F)").ase_atoms

~/.py39/lib/python3.9/site-packages/matlantis_features/atoms.py in from_smiles(cls, smiles, method, seed)
     73 """
     74 mol = Chem.MolFromSmiles(smiles)
---> 75 mol = Chem.AddHs(mol)
     76 
     77 if method not in ["etkdg_v2", "etkdg", "etdg", "kdg", "2d"]:

ArgumentError: Python argument types in
    rdkit.Chem.rdmolops.AddHs(NoneType)
did not match C++ signature:
    AddHs(RDKit::ROMol mol, bool explicitOnly=False, bool addCoords=False, boost::python::api::object onlyOnAtoms=None, bool addResidualInfo=False)

Please refer to the following for methods to resolve this issue.

 

2. Using autodE for Creating 5-coordinate and 6-coordinate Structures

When creating 5-coordinate and 6-coordinate structures from SMILES notation, it has been confirmed that using autodE can resolve the issue.

First, install autodE by following these steps:

!git clone https://github.com/duartegroup/autodE.git
!cd autodE && pip install .

After that, define the following functions. Using these functions, you will be able to generate 3D structures even for 5-coordinate and 6-coordinate molecules.

import autode as ade
from ase import Atoms, Atom
import numpy as np
from pfcc_extras.visualize.view import view_ngl

def smiles_to_ase_atoms(smiles):
    ade_mol = ade.Molecule(smiles=smiles)
    ase_atoms = ade_to_ase_atoms(ade_mol.atoms)
    return ase_atoms

def ade_to_ase_atoms(ade_atoms):
    ele = np.array([atom.atomic_symbol for atom in ade_atoms])
    pos = np.array([atom.coord for atom in ade_atoms])
    return Atoms(ele, pos)

molec = smiles_to_ase_atoms("[P](F)(F)(F)(F)(F)(F)")
view_ngl(molec, representations=["ball+stick"])

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Related to

Related articles