name to smiles conversion

28 views
Skip to first unread message

Simon Teague

unread,
Mar 27, 2013, 7:06:54 PM3/27/13
to licss...@googlegroups.com
All 

I have a file of names and wanted to convert them to smiles but found that the name to smiles macro only worked on a very few, does anyone else understand why?

This type of output is fairly typical of academic groups "we tried a range of known drugs against our screen and this was the output. Are there any patterns?"
Usually I convert to smiles calculate logP and charge type (not sure how to do that in LICSS?) then plot the biological result against these parameters before wasting too much time on it. 
If other people have a different modus operandi I would be interested to her about it. However if I can't get to smiles I can't do even this basic analysis.

A good piece of code would be a spreadsheet of all the known drugs with their props ready calculated which could be filtered by a list of compounds that the academic had screened. It would spit out a new 'subset' spread sheet with those props in and the biological variable in a separate column (vlookup?). [the drug name 'keyed variable' would have to be a 'contains' type not 'exact match' though because the academics all use slightly different names, but I recon a string of 6 letters matching would uniquify among the small set of known and available drugs - anyone want to take that on?)
dmdodrugs_1.xlsm

Kevin Lawson

unread,
Apr 11, 2013, 11:33:10 AM4/11/13
to licss...@googlegroups.com

Hi Simon

 

The Names to Smiles routine uses the NIH database lookup to find names.  The documentation is on their search page: http://cactus.nci.nih.gov/chemical/structure.

 

If, on the other hand, you select the option to use OPSIN, this uses the IUPAC to Chemistry package developed at Cambridge: http://opsin.ch.cam.ac.uk/. The website has links to the documentation.

 

We have used the NIH lookup service to get Smiles for natural products using their CAS numbers as a search term – we got better than 60% hit rate.

 

I am not aware of any other (free) lookup site (with associated webservice) for CAS numbers which has anything like the coverage of the NIH resolver but keen to receive suggestions…

 

Best wishes


Kevin

Reply all
Reply to author
Forward
0 new messages