N NMR Prediction

The resulting analysis gave a correlation coefficient of R = 0.97 over 21 244 points. The presentation shows a plot of a number of chemical shifts vs. the difference between the experimental and predicted chemical shift values. Approximately 250 shifts out of 20,000 have a deviation of 20 ppm. 

Selecting 24 random compounds (none from tables of related structures) from different sections of the authors review [12] on long-range heteronuclear 

Computer-assisted Structure Elucidation (CASE) Applications Employing Chemical Shift Correlation Data 

There have been very few reports of the combined use of computer-assisted structure elucidation methods that have employed any form of data, whether direct or long-range correlation. Two papers appeared in 1999, the first by Kock, Junker, and Lindel [43]. These authors compared the application of computer-assisted structure elucidation methods to the alkaloid oroidin (3) with and without the inclusion of chemical shift correlation data. The tabulated results obtained for oroidin and reported in the study are quite interesting in terms of the impact of the availability of chemical shift correlation data. 

As is readily seen from the data contained in Table 14.2, when there were limited numbers of long-range HMBC correlations provided to the 

---

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. 

It is possible to determine how the chemical shifts were predicted, and the type of structural fragments used to derive the parameters though a Calculation Protocol window which shows a series of points, each representing an individual chemical structure and associated chemical shift that was used to influence the chemical shift prediction. If the prediction was performed on a compound not in the database, then a variety of different structures is shown in the Calculation Protocol and displayed as a histogram plot containing structures which are only fragmentaUy similar to the input structure. The general apphcabUity and success of N NMR prediction will be examined in further detail below. 

---