Natural products are derived from natural sources such as herbs. The therapeutic effects of the components of natural products have been empirically demonstrated. Understanding active compound components of natural products can lead to novel combination drug discovery. Many researchers have devoted their endeavors in analyzing compound combinations of natural products. However, systematic analysis of compound combinations in natural products frequently faces hurdles since the information about natural products is widely scattered and has been designed in heterogeneous formats.
Compound combination-oriented natural product database with unified terminology (COCONUT) contains traditional medicine, combination drug, functional food, and molecular drug/target information curated from heterogeneous sources. COCONUT has accomplished standardization and structuralization of the heterogeneous data with unified terminology. The standardization of various data ensures high compatibility, and narrative text was structuralized, ensuring improved accessibility and availability. Furthermore, compound combinations have been extracted by association rule mining, so that the COCONUT database can be used for polypharmacology studies.
COCOA (Compound Combination-Oriented Association) extracts compound combinations that are expected to have therapeutic effects on diseases by the association rule mining analysis with the COCONUT database. It was evaluated by 4-fold cross-validation. For example, cyanocobalamin and vaccenic acid coexist in 12 herbs, including Zea mays, Mangifera indica, and Pisum sativum, which are known to treat diabetes mellitus. As a result, the combination of cyanocobalamin and vaccenic acid can be suggested as a potential therapeutic compound combination for diabetes mellitus.
