NP-StructurePredictor: Prediction of Unknown Natural Products in Plant Mixtures

Identification
of the individual chemical constituents of a mixture, especially solutions
extracted from medicinal plants, is a time-consuming task. The identification
results are often limited by challenges such as the development of
separation methods and the availability of known reference standards.
A novel structure elucidation system, NP-StructurePredictor, is presented
and used to accelerate the process of identifying chemical structures
in a mixture based on a branch and bound algorithm combined with a
large collection of natural product databases. NP-StructurePredictor
requires only targeted molecular weights calculated from a list of m/z values from liquid chromatography–mass
spectrometry (LC-MS) experiments as input information to predict the
chemical structures of individual components matching the weights
in a mixture. NP-StructurePredictor also provides the predicted structures
with statistically calculated probabilities so that the most likely
chemical structures of the natural products and their analogs can
be proposed accordingly. Four data sets consisting of different Chinese
herbs with mixtures containing known compounds were selected for validation
studies, and all their components were correctly identified and highly
predicted using NP-StructurePredictor. NP-StructurePredictor demonstrated
its applicability for predicting the chemical structures of novel
compounds by returning highly accurate results from four different
validation case studies.