Hi James,
I think the problem here is that you are interpreting the index pairs in
each match returned by getSubstructMatches() as if they were the atom
indices that form a bond. Instead, each pair corresponds to a
(queryAtomIdx, molAtomIdx) pair, and has nothing to do with bonds:
http://rdkit.org/docs/cppapi/namespaceRDKit.html#a254b4d13317dfb319a09a8c2c8fbd435
Furthermore, your molecule only has implicit hydrogen atoms, so you
should not expect any bonds involving H unless you add explicit hydrogen
atoms.
HTH, cheers
p.
On 11/29/18 12:01, James Wallace wrote:
I'm trying to use the Java wrapper to do a basic molecule
fragmentation, greedily matching against anything that fits the SMARTS
"[*;R]-;!@[*]" (effectively any regular bond that isn't part of a ring
system)
As a test case, I'm using a structure with a few obvious bonds that
can be broken, namely (in SMILES)
OC1CCC2CCC(OCCCCC3CCC(C4CCCC(Cl)C4Cl)CC3)CC2C1 .
As the Java Wrapper uses some slightly unusual functions for the
regular runtimes, I have included a toy example:
Set<String> output = new HashSet<String>();
RWMol mol2 = RWMol.MolFromSmiles(input, 0, true);
ROMol patt = RWMol.MolFromSmarts("[*;R]-;!@[*]");
Match_Vect_Vect matches = mol2.getSubstructMatches(patt);
for (int i = 0; i < matches.size(); i++)
{
Match_Vect match = matches.get(i);
for (int j = 0; j < match.size(); j++)
{
//lazy cloning
RWMol mol_to_manipulate =
RWMol.MolFromSmiles(input, 0, true);
Int_Pair pair = match.get(j);
// Find bond and remove
mol_to_manipulate.removeBond(pair.getFirst(), pair.getSecond());
System.out.println(mol_to_manipulate.MolToSmiles());
mol_to_manipulate.canonicalizeMol();
String[] temp =
mol_to_manipulate.MolToSmiles().split("\\.");
for (String tmp : temp)
{
if (tmp.length() > 1)
{
if (!tmp.equals(input))
{
output.add(tmp);
}
}
}
}
}
return output;
}
When I run the selection on this molecule, I see the bond between atom
0 and 1 (the terminal OH) from both directions (0,1 and 1,0). However,
as the iteration continues, the matches seem to no longer relate to
meaningful bonds (0,7 , 1,8 , 0, 13 , 1,12 , 0,16 ,1,17 etc), and an
attempt to remove them doesn't work as there is no bond to link them.
Am I making some mistake in my handling of the query, or is there
something going on with the matches (I'm making sure to load in a new
molecule each time for the bond removal, but I'm struggling to see how
the mapping follows).
_______________________________________________
Rdkit-discuss mailing list
Rdkit-discuss@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/rdkit-discuss
_______________________________________________
Rdkit-discuss mailing list
Rdkit-discuss@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/rdkit-discuss
