Excellent, Otis. Do let me know if Salem finds anything else.

Now we know that the CACTVS SMILES algorithm allows C1(C) both C(C)1, but
in the case where you have /1 (or \1), that MUST be C/1(C), not C(C)/1.
Since Jmol was always generating branches before connections (because it's
the branching that creates the connections in the first place), anything
with double bond stereochemistry and a branch at an atom would fail at NCI.
And then the secondary issue was that the JSON generator in Jmol was not
properly escaping \ in  this case. I only had it escaping " and new-line.
So you were getting garbage in those cases. Could well be the whole issue.

Sorry it took so many years to figure that out. All this is very timely,
though, because I have submitted a paper for publication about SMILES in
Jmol, and I have been actively working on aligning Jmol's SMILES business
with OpenSMILES. It was the OpenSMILES team who pointed out that C1(C) is
correct and C(C)1 is not.

So, hooray for teamwork! Hooray for OpenSource!

Finally, I hope you do explore the possibility of accessing Jmol's data
structure using "Java" instead of "JavaScript." You might remember that in
the old Java-applet days you could do that, too. Do you remember that when
returning a string, it was advisable to use  + ""  with that to be sure to
turn it into a JavaScript variable first? Jmol.getPropertyAsJavaObject()
could actually be very efficient now, since all you are doing with
Jmol.getPropertyAsArray() is converting JavaScript to JSON to JavaScript.

I could argue that one should always use Jmol.getPropertyAsJavaObject now.



Bob
------------------------------------------------------------------------------
Transform Data into Opportunity.
Accelerate data analysis in your applications with
Intel Data Analytics Acceleration Library.
Click to learn more.
http://pubads.g.doubleclick.net/gampad/clk?id=278785351&iu=/4140
_______________________________________________
Jmol-users mailing list
Jmol-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/jmol-users

Reply via email to