fragmenter.py

from rdkit import Chem
import itertools

atoms = ["C","N","O","F","Si","P","S","Cl","Br","I"]
bonds = ["","=","#"]

raw_smiles = []
mols = []
smiles_list = []

atom_pairs = list(itertools.product(atoms, repeat=2))
for a,b in atom_pairs:
    for x in bonds:
        raw_smiles.append(a+x+b)
atom_triples = list(itertools.product(atoms, repeat=3))
bond_pairs = list(itertools.product(bonds, repeat=2))
for a,b,c in atom_triples:
    for x,y in bond_pairs:
        raw_smiles.append(a+x+b+y+c)

atom_quads = list(itertools.product(atoms, repeat=4))
bond_triples = list(itertools.product(bonds, repeat=3))
for a,b,c,d in atom_quads:
    for x,y,z in bond_triples:
        raw_smiles.append(a+x+b+"("+y+c+")"+z+d)

atom_pents = list(itertools.product(atoms, repeat=5))
bond_quads = list(itertools.product(bonds, repeat=4))
for a,b,c,d,e in atom_pents:
    for x,y,z,zz in bond_quads:
        raw_smiles.append(a+x+b+"("+y+c+")"+"("+z+d+")"+zz+e)

for smiles in raw_smiles:
    try:
        mol = Chem.MolFromSmiles(smiles)
    except:
        continue
    if mol != None:
        smiles = Chem.MolToSmiles(Chem.MolFromSmiles(smiles))
        if smiles not in smiles_list:
            smiles_list.append(smiles)
            mols.append(mol)

print len(mols)