ACD/NNMR

Fig. 14.4 ACD/NNMR chemical shift calculation for the simple alkaloid harmaline (2). The N6 resonance, as expected, has a chemical shift in a range typical for indoles while the N1 resonance exhibits a calculated chemical shift more typical of that of pyridine, which is well outside the Fi window provided for the survey conditions shown in Figure 3A.

The validation of N NMR prediction is best performed by comparing the predicted shifts for compounds not in the database with the experimental shifts available in the literature or measured directly. ACD/Labs have reported [42] a statistical analysis of their N NMR prediction. Using a classical leave-one-out (LOO) approach they predicted the N shifts for >8300 individual chemical stmctures contained within the ACD/NNMR v 8.08 NNMR program database. The resulting analysis gave a correlation coefficient of = 0.97 over 21 244 points. The distribution in deviations between the experimental values and the predicted values using this LOO approach is shown in Figure 14.5. 

Fig. 14.n The database record in ACD/NNMR associated with st chnine. Note that four references are reported with their associated chemical shifts and coupling constants. 

Martin, G. E. and Williams, A. J. (April 18, 2004) Validation of ACD/ NNMR predictor, ACD/Labs User Meeting, ENC 2004, Monterey. 

The ACD/NNMR version 8 content database contains >8300 chemical structures (>21,000 N chemical shifts). These data have been culled from the literature and checked for quality according to a number of stringent criteria prior to adding to the database. The chemical shift reference is homogenized during the process such that all shifts are relative to one reference. A record includes the chemical structure, the original literature reference, the N chemical shift(s) and, where available, associated heteronuclear coupling constants. These data can be searched by structure, substructure. 

Access to a database of chemical structures and associated spectral parameters can accelerate the process of identifying an unknown. This is especially true when the data are utiHzed to produce NMR prediction algorithms and these are now available for a number of nuclei, most commonly and but also including P and F. Electronic content databases of N data are available from a number of sources [26—28], and the largest and most up-to-date source of data is associated with the ACD/NNMR predictor program, presently used in the laboratory of one of the authors (G. E. M.). 

The latest form of the ACD/NNMR version 2012 content database contains 9,780 chemical structures associated with over 22,000 N chemical shifts. Data continue to be extracted from the hterature on an ongoing basis and are checked according to a nrunher of stringent criteria. An individual compound record includes the chemical structure, ordinal hterature reference, one or more N chemical shifts, and, where available, associated heteronuclear-couphng constants. The latter data are, unfortunately, only infrequendy measured for N. These data can be searched hy diflerent parameters including structure, substructure, and chemical shift. This... 

To further evaluate the capabihty of the ACD/NNMR 2012 predictor as applied to the chemical compounds described in this review, we selected representative molecules from each of the individual sections in this work. This provided us with a set of 45 nitrogen-containing structures with experimentally measured N NMR chemical shifts. Of the compounds chosen 11 were present in the NNMR database and these were removed from the... 

Figure 9 The plot of the calculated versus observed chemical shifts for 117 chemical shifts predicted for molecules In this review using the ACD/NNMR program version 2012.

Figure 10 The plot of calculated versus experimental chemical shifts generated using the ACD/NNMR program, version 2012, following editing of structures (e.g., C=NH vs. C—NHJ. The improvement in the quality of the fit when compared to the graph shown in Fig. 9 is obvious.

NMR data and potentiaUy mass spectrometry, and with N chemical shifts in hand, it is possible to search the ACD/NNMR database using a number of flexible searches to identify potential hits or potential classes of compounds that match the experimental data. Various options can be set in the search interface (see Fig. 12). These include the Looseness Factor, the Minimum Number of Query Shifts, and the Hit Quahty Index (HQI). 

Figure 11 Entries contained in the database for the ACD/NNMR program com-moniy inciude data obtained in muitipie soivents.

ACD/NNMR Predictor, Version 12, Advanced Chemistry Development, Inc., Toronto, ON, Canada, 2011. 

---