Combinatorial NMR - Applications of NMR Prediction Algorithms
May 6th, 1998
Dear Barry,
As a result of the shift towards high throughput screening and
combinatorial analytical methods, we at Advanced Chemistry Development are pursuing any
possibility to support this exciting field. In collaboration with Varian NMR Instruments,
specifically Evan Williams and Paul Keifer, we have enhanced our NMR software offerings to
support Combinatorial NMR applications. ACD/Combi NMR presently allows the user to access
NMR data directly from the spectrometer and process using a Group Macro processing
feature. We have used this to process the data obtained from a 96-well plate but have
tested the bulk processing on over 200 FIDs. An alternative of course is to obtain the
appropriate set of phasefiles from the spectrometer. Following processing the spectra
automatically populate a database identical to those generated in our ACD/NMR Manager product.
The Combi/NMR module can be used to allow spectral, subspectral or
multi-subspectral searching with the "hits" displayed on a 96-well plate format
using color coding to show responses to the search. An example is shown below where the
multi-subspectral search was performed on the spectrum immediately below and matches,
spectra containing the highlighted subspectra, were located within the 96-well plate as
shown at right. Green highlighted buttons indicate the hits with the lower screen
indicating one of these hits and the upper screen showing the pattern for matching.
 ACD/Combi NMR therefore allows the user to remove the data from the
spectrometer and perform necessary data handling at the desk. Following this data handling
the Combinatorial plate data sets can be stored as individual databases or merged into a
larger overall database containing up to 500,000 spectra. Each entry in the database can
be associated with up to 16,000 textual user data fields that are fully searchable through
the database interface. These of course could include Notebook #'s, Chemist's
Name, Solvent details, Synthesis ID # or whatever.
 Following acquisition of a Combi data set the user has the opportunity
to add suggested structures and associate them with particular spectra in the dataset.
Presently this can be done directly through our integrated ChemSketch structure drawing
package (which can read ChemDraw, ISIS.skc, molfiles or many other formats). Using the ACD
H1 NMR prediction algorithms we generate predicted spectra for each of the suggested
structures and display them on screen for direct visual comparison with the experimental
spectra. This comparison could obviously be performed just by manually screening for
matches but we perform a statistical analysis based on the differences in shifts between
the experimental and predicted spectra and produce a Combi-result factor which varies
between 0 and 1, 1 being a perfect match. The obtained values are then displayed using
color coding to display ranges for the match factors. Databasing these spectra including
structures allows future searches of groups of such databases by structure or substructure
as well as the user definable textual data.
Our future improvement plans for this development project include
quantitation capabilities, ignoring exchangeable protons as an option during the spectral
matching procedure and utilizing user databases containing structures and assignments
pertinent to the chemistry of the user or laboratory in question. We look forward to our
continuing collaborations with Varian as well as suggestions from your readers regarding
possibly useful features that could be incorporated into ACD/Combi NMR.
Best wishes Barry!
Tony Williams
|
