Are there compounds that have the same SMILES code but are not stereoisomers? I am trying to find inconsistencies with SMILES code and currently I have that many compounds have multiples options for SMILES code but I want to find if there is a SMILES code that can represent multiple compounds.
$\begingroup$
$\endgroup$
2
-
$\begingroup$ Did you check the SMILES to be of same type and origin? The three likely most frequently requested SMILES openbabel assigns are canonical, universal, and InChIfied SMILES (documentation); it only is a question of structure complexity of the molecule structures submitted if they differ (within one program), or in comparison to other programs (GUI like ChemDraw/ChemDoodle/Marvin, CLI like RDKit). Both obabel and RDKit allow SMILES to SMILES conversion for a consistent starting point for such a test, too. $\endgroup$– ButtonwoodJul 31 at 20:26
-
$\begingroup$ Once on a solid ground, comparison with similar benchmarks/checks might be of interest for you, too. There are (at least) the ones by Noel O'Boyle, John Mayfield, Roger Sale presented to 256th ACS National Meeting Boston Aug 2018 (slides), and the blog post Running a SMILES Validation Benchmark by 2020 by Richard L. Apodaca. $\endgroup$– ButtonwoodJul 31 at 20:29
Add a comment
|
2 Answers
$\begingroup$
$\endgroup$
1
Many SMILES for single structures
There are many, many valid SMILES for most organic structures. For example, CCO, OCC, and C(O)C are all valid SMILES for ethanol.
Many structures for single SMILES?
There are probably cases where the same SMILES can represent multiple compounds, but most of these probably involve transition metal complexes, which SMILES is (I think) widely regarded as not handling very well, or else cases where what is meant by "compound" is also ambiguous. E.g. are "ethenol" and acetaldehyde are arguably both represented by C=CO, but in order to see that you need to know about keto-enol tautomerism.
-
2$\begingroup$ The correct (re)assignment of double bonds in organic compounds indeed can be problematic. We need to talk about Kekulization, Aromaticity and SMILES by Noel O'Boyle, John Mayfield presented to the 254th ACS National Meeting in Washington, DC Aug 2017 provides an insight. For an identification of a molecule in a database, given the computational power available today, use of multiple descriptors (SMILES complemented by a second «key») can be helpful (a bit like two-factor authorization on email). $\endgroup$ Jul 31 at 20:41
$\begingroup$
$\endgroup$
The SMILES corresponds to the underlying molecular graph (with bonds being edges and atoms vertices). This means the question you are asking is "are there any molecules that are different, not stereoisomers, and have the same underlying molecular graph".
The answer to this is yes. One typical example is atropoisomers (due to conformational barriers), though this can also be considered stereoisomery depending on how strict you are. This is actually relevant, for example in the kinase field some biaryls will show atropoisomerism. Another example is topological isomery. Imagine a trefoil knot C100 cycloalkane vs the ordinary C100 cycloalkane. Same graph, so same SMILES. However, in most typical uses of SMILES, these type of issues can be considered purely curiosities.
