ConformerGenerator

Example using ETKDG

from confgen.etkdg import ETKDG

# Initialize ConformerGenerator
cgen_etkdg = ETKDG(n_confs=100, pruneRmsThresh=0.1, n_cores=2)

# Initialize rdkit.mol object without any conformers
mol = Chem.MolFromSmiles("C[C@H](N)C(=O)NCC(=O)O")
mol = Chem.AddHs(mol)

# Generate Conformers
mol3d = cgen_etkdg.generate(mol)

mol3d.GetNumConformers()
# >> 14

Atoms can be constrained in the embedding process by passing a list of atom indices to the constrained_atoms argument. This requires the presence of one 3D conformer in the molecule, the atoms will the constrained to their respective position in this conformer.

Alternatively, a coordMap dictionary with atom indices as keys and either lists, numpy arrays or Point3Ds as values can be passed to constrained_atoms.

# constrain atoms with the following indices to their respective position in the first conformer
constrained_atoms = [2,3,4,5,16]

# constrain atoms with the following indices to the positions given in the dictionary
constrained_atoms = {2: [0,0,0], 3: [2,0,0], 5: [2,2,0]}

mol3d = cgen_etkdg.generate(mol, constrained_atoms=constrained_atoms)

Example using CREST

CREST keywords can be found here

from confgen.crest import CREST

# Initialize ConformerGenerator
cgen_crest = CREST(gfn=2, ewin=6, mquick=True, mdlen='x0.1', n_cores=2)

# Initialize rdkit.mol object without any conformers
mol = Chem.MolFromSmiles("C[C@H](N)C(=O)NCC(=O)O")
mol = Chem.AddHs(mol)

# Generate Conformers
mol3d = cgen_crest.generate(mol)

mol3d.GetNumConformers()
# >> 9

Constrains can be parsed to CREST as a list of atom and/or bond indices:

pattern = Chem.MolFromSmarts('NC(=O)CN')
match = list(mol.GetSubstructMatch(pattern))
# >> [5, 3, 4, 1, 2]

mol3d = cgen_crest.generate(mol, constrained_atoms=match, constrained_bonds=[(1,4),(4,5)])

Tools

Geometry Optimization

Availabel methods: uff, mmff and gfnff, gfn1 and gfn2 The method run will return a new rdkit.mol object with the optimized conformers sorted with ascending energies.

from confgen.tools import GeomOptimizer

gfn2_opt = GeomOptimizer(method='gfn2', n_cores=2)

mol3d_opt = gfn2_opt.run(mol3d)

conformer_energies = [conf.GetDoubleProp('energy') for conf in mol3d_opt.GetConformers()]

RMSD Clustering

The keep argument specifies which conformer of each cluster should be retained, either the lowest energy conformer (lowenergy) or the centroid (centroid).

from confgen.tools import Cluster

rmsd_cluster = Cluster(rmsdThreshold=0.5, keep='lowenergy')

clustered = rmsd_cluster.run(mol3d_opt)

clustered.GetNumConformers()
# >> 8

Energy Filter

The ewin argument specifies the energy window in which conformers should be retained (up from the lowest energy conformer found).

from confgen.tools import Filter

filtered = Filter(ewin=2).run(mol3d_opt)
filtered.GetNumConformers()
# >> 6