SMILES extension and radicals

In your description of SMILES extensions, it has been suggested to explicitly include information about radicals (or, I suppose, free electrons in general). This is not really necessary since it is implied by the number of atoms connected to the atom with the free electron(s). For example, methyl radical would be represented as [CH3], carbene as [CH2] and carbon monoxide as [C]=O. Thus, there is only the limitation that you can't specify whether a diradical is a singlet or triplet.

Hi!





Thank you for the suggestion.


This is exactly how we import/export SMILES strings. If you found something different, that is a bug.


As you pointed out, SMILES has the limitation that you can't specify whether a diradical is a singlet or triplet. This is solved in the cxsmiles format.


An other problem, that this method works in SMILES, but not in SMARTS. As far as I know it is not possible to store radical information in SMARTS. Please correct me if I'm wrong.


That's why you should use cxsmarts if you want to use radical information also.

The point was that SMILES already accommodates radicals and the only necessary extension is for specifying the singlet/triplet state of a diradical. Since this is not relevant for most purposes, it hardly seems worth bothering about.

Hi,





If you have radicals you can export it to SMILES without any problem, it is not necessary to use the extended SMILES format.





All the best


Andras

Another concern is when a radical is put on a nonstandard valence atom. In this case you cannot decide from the smiles alone if a radical is present or not. Please find attached a few such compounds from the NCI dataset.





Best regards,





Szabolcs

Charlesb,


Do you know if there is any tricky way to store the radicals in SMARTS?


As SMARTS is an extension of SMILES I would expect that if a molecule can be exported to SMILES, it should be also exportable to SMARTS. But my feeling is that this is not true for radicals.





All the best


Andras