openmolecules.org Forum: Cheminformatics » Visualising SMILES string

Home » DataWarrior » Cheminformatics » Visualising SMILES string

Show: Today's Messages :: Polls :: Message Navigator

Visualising SMILES string [message #1256]

Thu, 18 March 2021 10:55

amorrison
Messages: 38
Registered: March 2016

Member

Hi Thomas,

I have a smiles string with an attachment point that has been created in another application, e.g. C1CCC([*:1])CC1. Datawarrior fails to identify it as a valid smiles either on paste or convert structure to name. I'm assuming it's the '*:1' that is the issue. Can I modify the smiles so it can be interpreted by Datawarrior?

many thanks in advance,

Angus

Report message to a moderator

Re: Visualising SMILES string [message #1257 is a reply to message #1256]

Sat, 20 March 2021 15:07

thomas
Messages: 715
Registered: June 2014

Senior Member

Hi Angus,

if you replace [*:1] by *, then the atom is interpreted as wildcard query feature, which is drawn as a '?'.
In this case the new molecule is considered as sub-structure rather than as normal structure. The difference is
that a sub-structure may have query features and all unspecified atom valences are not implicitly considered
to be blocked by hydrogen atoms. Therefore, a single bonded oxygen is shown as -O in contrast to -OH in a normal molecule.

DataWarrior makes this distinction, which is not made by most other software, possibly, because the SMILES definition itself does not distinguish between these two structure flavors.

Thomas

Report message to a moderator

Re: Visualising SMILES string [message #1258 is a reply to message #1257]

Sun, 21 March 2021 05:00

amorrison
Messages: 38
Registered: March 2016

Member

Thanks Thomas,

I've tried this. Doing a find and replace my smiles string becomes - C1CCC([*])CC1. If try a structure to name this is still not recognised as a valid smiles. If I remove the square bracket - C1CCC(*)CC1. I get the structure but lose the attachment point.

Sorry if this trivial I'm a med chemist and not so familiar with smiles syntax.

Thanks,

Angus

Report message to a moderator

Re: Visualising SMILES string [message #1261 is a reply to message #1258]

Tue, 23 March 2021 16:38

thomas
Messages: 715
Registered: June 2014

Senior Member

Dear Angus,

I have extended the SmilesParser within OpenChemLib and used by DataWarrior to support '*' and '?' as pseudo atom symbols. Both symbols can be used inside or outside square brackets. While '*' is inline with the opensmiles.org standard, '?' is not. Nevertheless, because of the missing distinction in the smiles syntax between query fragments and full molecules, I believe that '?' makes sense.

'*' creates a wild card atom, which is an atom query feature and only allowed in query fragments. Therefore, a smiles containing a '*' is atomatically perceived as a query structure rather than a full molecule. Its free valences are not considered as being filled with hydrogen.

'?' is converted into an atomicNo=0, which DataWarrior uses as attachment point, e.g. in R-groups after a SAR deconvolution. These atoms are meant to not exist. Their sole purpose is to carry the bond sticking out of the R-group.

If you paste the these 6 rows of smiles:

C1CCC([*:1])CC1O
C1CCC([*])CC1O
C1CCC(*)CC1O
C1CCC([?:1])CC1O
C1CCC([?])CC1O
C1CCC(?)CC1O

into a new DataWarrior Window (newest dev release), then you get the following table:
Note that I have added an oxygen atom. If you see an 'H' at the oxygen, the structure is considered a molecule. If not, the structure is a query fragment with open valences.

/forum/index.php?t=getfile&id=310&private=0