How to calculate the number of hydrogen atoms from SMILES string?
For example,
SMILES String: C1=CC(=C(C=C1C(CN)O)O)O
The known result is C8H11NO3
$\begingroup$
$\endgroup$
3
-
1$\begingroup$ I think you wish to know how to add explicit hydrogens to a structure. The fact that your structure is originally a SMILES is not particularly important. $\endgroup$– gilleainJan 6, 2016 at 16:25
-
$\begingroup$ I think basically you just want to calculate a molecular formula from the SMILES? $\endgroup$– Geoff HutchisonJan 6, 2016 at 18:40
-
$\begingroup$ @GeoffHutchison I just want to calculate a molecular formula from the SMILES. Because, there is no universal SMILES format handled by any chemical database. So, matching exact SMILES string is very difficult. Instead matching molecular formula is easier. $\endgroup$– ak-SEJan 6, 2016 at 20:48
Add a comment
|
3 Answers
$\begingroup$
$\endgroup$
0
Other answers have done a good job of providing a description of the concepts and algorithms needed to answer the question. Here I'll demonstrate the opposite: a black-box empirical approach.
In Python using rdkit:
from rdkit import Chem
from rdkit.Chem import rdMolDescriptors
my_smiles_string = 'C1=CC(=C(C=C1C(CN)O)O)O'
my_mol = Chem.MolFromSmiles(my_smiles_string)
print(rdMolDescriptors.CalcMolFormula(my_mol))
This shows the result C8H11NO3. You could parse that further to get only the number of hydrogens, if you need to do so. Alternately:
my_mol_with_explicit_h = Chem.AddHs(my_mol)
print(my_mol_with_explicit_h.GetNumAtoms() - my_mol_with_explicit_h.GetNumHeavyAtoms())
displays 11. Slightly more cumbersome but it has the output you want. If you take the latter approach you don't even need to import rdMolDescriptors.
$\begingroup$
$\endgroup$
3
Information on the system can be found here: https://en.wikipedia.org/wiki/Simplified_molecular-input_line-entry_system http://www.daylight.com/dayhtml/doc/theory/theory.smiles.html
The correct number of hydrogen atoms to fulfill the bond count around the other elements is assumed unless placed in square brackets. The bonds can be read from left to right and assumed to be a chain, unless placed in braces which represent a branch.
So there would be 1H around the first C (going from left to right) this is because it is c1 indicating it's one half of a ring closure so will have an extra bond attached, there is one H on the second C. The third C has a branch and an O (the chain continues after the braces close) which would have a single H attached, the third carbon has no H as the branch is attached via a double bond. (so far we have 3H)
The first C in the branch has no H and attaches to another branch and another O with an H. The first C in the second branch has a single H. The second C in the second branch is important because it has 1 after it like the first C. This indicates that they should have a bond between them forming a ring closure. It continues down the branch which now looks like a substituent on the ring. So it has no H. The third carbon in this branch has an O (with an H) and a new branch, so it has a single H. (So we have 7H so far) The third (and final 'branch' has a C attached to a N and two H, the N has two H attached. which adds 4H making 11H.
-
$\begingroup$ Thank you for your information. You have given an explanation for constructing the SMILES string from the chemical structure. But, I need to calculate atoms from the SMILES string (ie., parsing the SMILES, especially H). The problem is -- in SMILES representation, Hydrogen (H) is omitted. $\endgroup$– ak-SEJan 6, 2016 at 16:14
-
$\begingroup$ But you know the valency of carbon and nitrogen (and other organic components) You just need to work out how many other elements are attached to the carbon and whether they are attached via double bond or single and take them away from the overall valency. It effectively displays the skeletal formula in text form. $\endgroup$– amlJan 8, 2016 at 11:39
-
$\begingroup$ Wherever the chemical formula comes from, it looks quite wrong. $\endgroup$– Martin - マーチン ♦Mar 20, 2021 at 1:31
$\begingroup$
$\endgroup$
You have to convert the SMILES to a graph model. Most everything you want to do with SMILES requires you construct the graph model, i.e., nodes (atoms) and arcs (bonds). Once you have that, you can go through all nodes, and see how many of their sites are occupied by explicit bonds, the rest you hydrogenize.
If you do chemical informatics, you really need to learn these principles and learn to program them yourself.
But taking this obvious solution aside for a moment, with SMILES the question is always: can you simplify the solution so as to work purely text string based? Might you be able to use regular expression replace pattern matching to come up with your solution? And the answer is categorically no! Because regex patterns do not work for nested parenthesis as you use in SMILES to write out branches.
Could you use a parser to do it? May be. But if you go through this effort, you might just as well reconstruct your nodes and arc model, which then you can use to do so much more than to just hydrogenize.
