July 1-5, 2018
Nantes Event Center, Nantes, France
Using Predicted 13C NMR Spectra with Open Resources for Structure Dereplication
Dimitris Argyropoulos, Sergey Golotvin, Rostislav Pol, Joe DiMartino, Arvin Moser
Abstract: The ability to quickly and reliably separate and identify active components in natural product mixtures—using bio-assay and/or mass spectrometry guided fractionation—is critical for successful natural product-based drug discovery. Dereplication is the practice of screening active compounds early in the development process, to recognize and eliminate compounds that have been previously studied. This enables scientists to focus on testing truly ‘unknown’ compounds.
For efficient dereplication, one must be able to easily identify characteristic spectral “fingerprints” of compounds in order to identify their structure. The 13C NMR spectrum of a compound can be considered a fingerprint since it is virtually unaffected by conditions such as pH, concentration, and solvent effects. It is also largely magnetic field independent, since there are no couplings that could cause variations in stronger or weaker fields. To determine whether a 13C NMR spectrum has been previously recorded and solved, having access to search databases of 13C NMR spectra is a very powerful and reliable strategy. The search can be enhanced by including search terms such as molecular formula (expanded to cover MF ranges) and by accommodating for missing or extraneous peaks in the NMR spectrum. It is also very common to use such databases to identify structural fragments in the case of genuinely unknown structures.
The next question is whether to use databases of real spectra or predicted spectra. Databases of real spectra usually contain a limited number of structures, and their spectra may not be ideal. On the other hand, there are several “open” databases with chemical structures that could be used to predict 13C spectra. Two examples of this are PubChem and ChemSpider.
Here we explore the possibilities and limitations of using predicted 13C spectra for structures from open databases. The workflow is described together with examples of the results and the potential usefulness of the technique.