NMR Molecular Modeling

[Darrell R. Davis]

on 15 Jun 90 19:40:19 GMT,
Kurtis MacFerrin <macferrin%slsvax.harvard.edu at husc6.harvard.edu> said:

Kurtis> In article <9006141833.AA08245 at remote.dccs.upenn.edu>, GOLUB at PENNDRLS.UPENN.EDU (Ellis Golub) writes...
>Does anyone know of a program, running on an IRIS which can determine
>molecular structure from 13C NMR data (chemical shift, Jcc 1,2 & 3 bond)?
Kurtis> Yes, there are several. Two that I know about were written by David Hare of
Kurtis> Hare Research, Washington. They are called ftnmr and felix. I belive BioGraf
Kurtis> in California makes a program called nmrgraf.


First:  FTNMR and FELIX do *not* determine molecular structure based
on 13C NMR data.  These programs are for processing and analyzing NMR
data (they do a superb job BTW), determining structures from this data
is a subsequent step.  They are products of Hare Research Inc., Dennis
Hare, not David, is the president of this company.  These programs do
indeed run on SGI boxes.

Second: It is my opinion that the best you could probably do from 13C
data of the type you describe is access a database and look for known
structures or part structures that have NMR data that match yours.
Something like the Aldrich Library of NMR Spectra. Unfortunately I
don't think it exists in electronic format.  What you are asking is a
pretty tough problem given the rather limited information content of
chemical shift and coupling constants.

Third: There exists several "molecular modeling packages" for
determining structure based on NMR data, but the required input is
different than you describe.  The information needed is either angle
constraints or distance constraints, or both. This should imply that
considerable data analysis is required above and beyond chemical shift
and coupling constant analysis :-).  The methodology falls into two
categories: the use of either restrained molecular dynamics, distance
geometry, or a combination of both.  The NMRGRAF software is real new
and hasn't been used much yet; DSPACE from Hare Research uses distance
geometry and has sophisticated display capability, but is not a
molecular modeling package in the generic sense; other research groups
use Charmm or XPLOR for the calculations and popular modeling software
like Quanta, FRODO, SYBYL, MIDAS, INSIGHT or something like that for
display.

E-mail me if you want any other information about my favorite topic.

--------------------------------------------------------------
                           *                                  
Darrell R. Davis       *   *   *    "Faster, faster, until the
Assistant Professor      * * *       thrill of speed overcomes
Medicinal Chemistry   *A**L**T**A*   the fear of death."
University of Utah       * * *       
                       *   *   *       --H.S. Thompson 
                           *
--------------------------------------------------------------