Wow. I love seqcis! Rule 3 was trivial
Robert M. Hanson
St. Olaf College Chemistry
from my Windows phone
-----Original Message-----
From: "Robert Hanson" <hans...@stolaf.edu>
Sent: 4/11/2017 7:30 AM
To: "BlueObelisk-Discuss" <blueobelisk-discuss@lists.sourceforge.net>
Subject: Fwd: [BlueObelisk-discuss] Fwd: Cahn-Ingold-Prelog rules into Jmol
[sorry - forgot that this list requires "reply-all"]
---------- Forwarded message ----------
From: Robert Hanson <hans...@stolaf.edu>
Date: Tue, Apr 11, 2017 at 7:29 AM
Subject: Re: [BlueObelisk-discuss] Fwd: Cahn-Ingold-Prelog rules into Jmol
To: John Mayfield <john.wilkinson...@gmail.com>
On Tue, Apr 11, 2017 at 2:37 AM, John Mayfield <john.wilkinson...@gmail.com>
wrote:
On 11 April 2017 at 04:37, Robert Hanson <hans...@stolaf.edu> wrote:
2) What did you get for the other test case, that one checks you have the
ordering ranking for atomic masses.
CC[C@@](CO)([H])[14CH2]C
R.
There you go, that should also be S, ordering is: *CO, *[14CH2]C, *CC, *[H]
https://nextmovesoftware.com/blog/2015/01/21/r-or-s-lets-vote/.
John, what basis in the IUPAC rules leads you to this reading? It suggests that
atoms in the nth sphere cannot be ranked until atoms in the (n+1)th sphere are
checked after application of Rule 1, even if they could be distinguished by
Rule 2. Are you suggesting that after each rule is checked (Rule 1a, Rule 1b,
Rule 2 -- or is it Rule 1(a and b), Rule 2,...?) one must expand to the next
sphere before making a decision? That seems to me (a) unsupportable by the
IUPAC rules and (b) just asking for extremely complex code and a whole lot of
unnecessary checks.
My understanding is that exhaustive application of all rules are done within
the sphere first, then the process is repeated at the next sphere. What I read
is this:
The ranking of each atom in the nth sphere depends in the first place on the
ranking of atoms of the same branch in (n − 1)th sphere, and then by the
application of the Sequence Rules to it; the smaller the number, the higher the
relative ranking. (Ranking Rule 2).
This is certainly my understanding from all the reading I have done. You have
three atoms connected to an atom. You rank those three atoms based on the
rules. Atoms that are tied are taken to the next sphere, but not until that
process is completed.
To me that is pretty clear: We apply all rules to rank all atoms in a single
sphere. Nothing here says, "Atoms in a sphere are compared pairwise, and if
they are identical, then the comparison of this pair is continued to the next
sphere. Once this depth-first relative ranking is determined, the procedure is
repeated with all pairs of the sphere." I can certainly see where that reading
could drive one mad.
Q: Is there software that does a nice job with producing digraphs from SMILES?
I think I added a utility in Centres, however I've barely looked at the code in
5 years - but am planning to brush it off and clean up now though. BTW if you
look closely, Centres is abstract and wraps around existing toolkits - I only
wrapped it around CDK though in theory you could do the same with JMol.
Q: These all implement Rule 1b and the rest of the rules? Have they been
validated in some systematic, common way, so we know they don't have any bugs?
I don't think so. IIRC 1b was introduced to fix this case:
O[C@H](C(CCC1CC1)(CCC1CC1)CCC1CC1)C12CCC(CC1)CC2. If you use that molecule you
can tell whether it does/doesn't implement that rule. Without rule 1b it should
not be possible to label it. In centres you can change the rules of the
ranking: CDKPerceptor.java.
Yes, that's one of the models Mikko sent me. I used it for checking Rule 1b.
Q: Doesn't this argue against the "Why bother doing this -- it's been done
seven times already" argument? Which one is IUPAC-2013-standard?
It wasn't me who said that, I'd only say don't do it because the implementation
will drive you mad :-). The "blessed" version would allow everyone to confirm
against it, as your original question asks - you want to test yours it would be
much simpler just to point to a complete one leave it there. However from my
previous testing I don't know if a complete one exists anywhere (maybe the
LHASA one: http://pubs.acs.org/doi/abs/10.1021/ci00019a004 but of course this
maybe doesn't exist anymore, will ask them).
Supposedly ACD/Labs has a compliant CIP-determining algorithm.
http://bulletin.acscinf.org/PDFs/247nm44.pdf
Is ACD/Labs represented on this list?
I guess this would matter if you had 1,000,000 compounds to check; the 100-line
algorithm (Rules 1 and 2) I wrote seems quite straightforward and suitable for
my purposes. Hard to believe any molecule of interest would push the limits for
such.
CHEBI:51439, whether that's of interest or not is of course subjective
That's a nice test model.
Bob
--
Robert M. Hanson
Larson-Anderson Professor of Chemistry
St. Olaf College
Northfield, MN
http://www.stolaf.edu/people/hansonr
If nature does not answer first what we want,
it is better to take what answer we get.
-- Josiah Willard Gibbs, Lecture XXX, Monday, February 5, 1900
------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
_______________________________________________
Blueobelisk-discuss mailing list
Blueobelisk-discuss@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/blueobelisk-discuss