Using similarity searches over databases of estimated 13C NMR spectra for structure identification of natural product compounds

Structure elucidation for natural product compounds is assisted by making similarity comparisons between the uncharacterized experimental 13C NMR spectrum with relevant databases of estimated spectra. Databases of estimated spectra are deduced from a small set of assigned structures using HOSE codes. Using spectra estimated from structures circumvents problems of inconsistent, incomplete, missing or irrelevant data. It also enables rapid generation of reasonably sized databases that are unavailable from commercial sources. We validate the similarity method theoretically by analyzing what can be best expected from a match of an estimated set of peaks to the experimental spectrum. We also show by example that the method is successful when used in the laboratory.