Artificial intelligence-curated repository of gene-encoded natural diverse components from herbal medicines

Natural components, evolved to help organisms adapt and defend against threats, are also vital sources for drug discovery due to their diverse and potent bioactivities. In the present work, we proposed the Gene-encoded Natural Diverse Components Repository (GNDC, https://cbcb.cdutcm.edu.cn/gndc/), a primary and most extensive database dedicated to cataloging diverse natural components. GNDC currently catalogs over 234 million natural components that are organized into four specialized sub-databases: HerbalMDB for 2.32 million secondary metabolites, HerbalPDB for 229 million small peptides, HerbalRDB for 2.38 million small RNAs, and HerbalCDB for 0.26 million carbohydrates. By leveraging customized pipelines for high-throughput multi-omics data and AI technologies, the GNDC enables large-scale discovery and annotation of natural products from nuclear and organellar genomes of species listed in eight global pharmacopoeias and multi-resource data. Compared to existing resources, GNDC achieves a 10-fold increase in component yield and introduces over 200 million previously unreported components. To support this unprecedented data volume and complexity, state-of-the-art AI tools are seamlessly integrated to decipher and annotate vast data collections, such as classification and gene expression signature generation of millions of secondary metabolites. We envision that the GNDC will drive the transformation of drug discovery from an "experience-driven" approach to a "big data-driven" paradigm.