NPBS Atlas: a comprehensive data resource for exploring the biological sources of natural products

Abstract

Natural products continue to play a pioneering role in drug discovery due to their extraordinary chemical and biological diversity. However, their full therapeutic potential remains largely underutilized, hindered by the fragmented documentation of biological origins in existing data resources. Here, we present natural product and biological source atlas (NPBS Atlas), a data resource covers over 218,000 natural products fully annotated with comprehensive biological sources, bioactivities, and references. The database established through systematic text mining and expert manual curation, places special emphasis on curating source organism data through the information of scientific nomenclature, taxonomic classification, source parts, and the source of Traditional Chinese Medicines. NPBS Atlas represents significant advancement in natural product data resources through its unique content, specialized annotations, and featured data, thereby enabling unprecedented exploration of nature-derived chemical diversity through biological context. The web interface of NPBS Atlas is freely available at https://biochemai.cstspace.cn/npbs/.

Graphical Abstract

Supplementary Information

The online version contains supplementary material available at 10.1186/s13321-025-01116-y.

Scientific Contribution

Records all major biological sources of natural products, including plants, animals, fungi, and bacteria, with dedicated annotation for marine-derived organisms. Unique documentation of source organisms' parts (e.g., roots, stems, leaves, fruits, etc.) and Traditional Chinese Medicine (TCM) applications—featured data absent in similar resources. More than 14,000 natural products annotated with bioactivities (e.g., cytotoxic, antioxidative, antibacterial, antitumor activities, etc.)—curated therapeutic profiles unavailable in comparable repositories.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13321-025-01116-y.

Introduction

Natural products have long served as a cornerstone in drug discovery, offering unparalleled chemical diversity and bioactivity that drive therapeutic innovation [1–3]. Over 50% of approved small-molecule drugs, including landmark therapeutics such as paclitaxel (anticancer), artemisinin (antimalarial), and morphine (analgesic), are derived directly or inspired by natural scaffolds [4, 5]. These bioactive molecules, evolved through millennia of ecological interactions, occupy chemical spaces far beyond the reach of synthetic libraries, yet only a fraction of Earth’s biodiversity has been systematically explored [6, 7]. Recent estimates suggest that only a small portion of plant species and even fewer marine organisms have been studied for their pharmacological potential, underscoring the vast untapped reservoir of natural products awaiting discovery [7, 8]. This chemical richness, coupled with their proven clinical relevance, positions natural products as indispensable assets in addressing emerging health challenges, such as antimicrobial resistance and complex chronic diseases [9–11].

Over the past decades, numerous data resources have been established to catalogue natural products, facilitating cheminformatics-driven drug discovery, including DNP 34.1 [12], Super Natural 3.0 [13], CMNPD 1.0 [14], COCONUT 2.0 [15], LOTUS 1.0 [16] etc. These resources, however, often prioritize natural product data while neglecting the critical information of biological sources [17–19]. For instance, incomplete taxonomic annotations, ambiguous organism-source associations, absent information on source parts (extraction tissues) and references, severely limit the utility of existing repositories. Such gaps hinder the ability to trace bioactive compounds back to their biological origins, impeding efforts to explore structure–activity relationships, ecological roles, or sustainable sourcing strategies [20, 21]. Consequently, the lack of biological sources documentation has emerged as a bottleneck in translating natural product data into actionable insights for natural drug development [22, 23].

To address these limitations, we present natural product and biological source atlas (NPBS Atlas), a comprehensive database integrating rigorously curated biological source annotations for over 218,000 natural products. NPBS Atlas systematically links natural product to its taxonomic origins (e.g., plants, animals, fungi, and bacteria), source parts (e.g., roots, stems, leaves, fruits, etc.), and Traditional Chinese Medicine (TCM) applications. Additionally, it incorporates bioactivity data to enable analyses of the therapeutic profiles of natural products. To facilitate data driven discovery of natural product applications, we deployed the web interface of NPBS Atlas at https://biochemai.cstspace.cn/npbs/, for free access and exploration to the biological context of vast natural molecular space.

Construction and content

Collection of raw materials and text mining

The raw materials were derived from scientific literatures on natural product research, systematically retrieved from specialized journals in natural product chemistry and literature databases including PubMed, Web of Science, and the China National Knowledge Infrastructure (CNKI). To ensure comprehensive coverage, a targeted search strategy was implemented using keywords such as "isolation of", "isolated from", "produced by", "derived from", and "originated from", which are frequently employed in describing natural product discovery. Subsequently, a computer-assisted text-mining system was deployed to extract information from abstracts and main texts, including natural product names, source organism’s names, taxonomic information, source parts, specific area (marine organisms), bioactivity descriptions, and TCM applications. The chemical structures of natural products were either manually drawn using ChemDraw (v22.0) based on textual (systematic nomenclature such as IUPAC names) or graphical descriptions, or retrieved from chemical databases such as PubChem and ChEMBL [24, 25].

Data annotation and quality control

To ensure the accuracy and interoperability of the curated dataset, a systematic annotation pipeline was implemented, combining computational workflows with expert validation. Chemical structure data of the natural products were processed using RDKit (v2024.3.5), where molecular representations were first standardized using the Chem.MolStandardize module to handle charges, fragments, and tautomers. Subsequently, we generated unique identifiers, including the canonical SMILES (Chem.MolToSmiles) and InChIKey (Chem.MolToInchiKey), and calculated key drug-like properties such as molecular weight (Descriptors.MolWt), logP (Crippen.MolLogP), the molecular formula (rdMolDescriptors.CalcMolFormula), and the Quantitative Estimate of Drug-likeness (QED, using Chem.QED.qed). We initially utilized the widely-used SAscore [27] algorithm to evaluate the synthetic accessibility of the molecules. In future updates to the database, we plan to integrate newer methods, including AiZynthFinder [28], to provide more comprehensive and accurate information on the molecular properties of synthetic accessibility. Structures with RDKit parsing errors (e.g., incomplete valence, undefined stereodescriptors) were reconstructed manually using ChemDraw, aligning with the original literature descriptions. Chemical classification and biosynthetic pathway annotations of the natural products were performed using NPClassifier [29]. Taxonomic data for the source organisms, including scientific nomenclatures, classifications, and species catalogue links, were retrieved programmatically from the Catalogue of Life (CoL) [30]. We utilized the COL's public API to query the organism names (e.g., Agrimonia eupatoria), which then returned the complete taxonomic information (classification, nomenclature, and resource links). The World Register of Marine Species (WoRMS) webservice was integrated to annotate the marine-derived organisms [31]. The names of biological sources not matched in CoL or WoRMS were manually verified or corrected by referencing the original text. Annotating TCM applications of the source organisms was also referred to TCMBank by matching the scientific names of source organisms with herb names in TCMBank database [32]. The final datasets were spot checked (5%) by experts to ensure data quality.

Web interface implementation

The web interface of NPBS Atlas is built on a backend architecture powered by the Django (v4.2.6) framework [33]. Data is stored in a PostgreSQL (v15.5) database, which is extended with the Bingo (v1.92.0) to enable efficient chemical structure searching. On the frontend, the Ketcher (v2.20.0) is used for drawing and inputting molecular structures on the search page. For visualization, we integrated the JavaScript-based molecular editor JSME (v2024-04–29) [35], which enables interactive display of chemical structures and allows users to easily copy and download (right-click) the structures in various formats such as SMILES, Molfile, and InChI.

Utility and discussion

NPBS Atlas covers over 218,000 fully annotated natural products, records all major biological sources of natural products, including plants, animals, fungi, and bacteria. The systematic analysis of natural product origins within NPBS Atlas reveals distinct taxonomic and ecological patterns with implications for natural drug discovery and biosynthetic research (Fig. 1).

Fig. 1.

Systematic analysis of the natural products (NPs) in NPBS Atlas. a Total of 218,941 NPs with annotations of biological sources including plants, animals, fungi, and bacteria in NPBS Atlas. b Intersection analysis of the NPs with taxonomic categories. c Coverage of the NPs derived from marine-derived organisms (Marine) and TCMs, annotated with bioactivities (Bioactivities), QED > 0.5, and SA Score > 5.

Source data are provided in Additional file 1

Plants emerged as the predominant source of natural products (67% of entries), reflecting their historical prominence in ethnopharmacology and the combinatorial diversity of plant secondary metabolism (Fig. 1a). Notably, cross-taxon analysis identified significant overlap between plant and fungal metabolites (3520 shared natural products), likely attributable to convergent biosynthetic strategies (e.g., polyketide and terpenoid pathways) and ecological interactions such as endophytic symbiosis (Fig. 1b). Marine ecosystems dominated animal-derived natural products (77%), with sponges (Porifera) contributing disproportionately, underscoring their evolutionary innovation in chemical defence mechanisms (Fig. 1c). Traditional medicinal applications were strongly associated with plant-derived compounds, with 53% explicitly linked to documented uses in TCMs (Fig. 1c). Bacterial-derived natural products were annotated with the most bioactivities (17%), including anti-inflammatory, antioxidant, and immunomodulatory bioactivities, providing an empirical foundation for reverse pharmacology approaches (Fig. 1c). Evaluating drug development potential, fungal-derived natural products demonstrated superior QED profiles, with 50% exceeding the QED threshold of 0.5 (Fig. 1c). Bacterial natural products, while structurally complex (46% SA Score > 5), displayed exceptional scaffold diversity, including rare macrocyclic and hybrid polyketide-nonribosomal peptide architectures (Fig. 1c).

The NPBS Atlas web interface offers a comprehensive suite of tools designed for intuitive data search, exploration, and analysis. Users can perform queries using identifiers, names, or chemical structures. To address the particularity of Traditional Chinese Medicines (TCMs), the platform uniquely incorporates search capabilities based on Chinese nomenclature and pinyin. For more advanced queries, the interface provides several chemical structure search options—including exact, substructure, and similarity—that leverage cheminformatics algorithms to discover natural products with specific chemical skeletons.

To further facilitate analysis, a multidimensional filtering system enables users to refine datasets by biological source, biosynthetic pathway, chemical class, and molecular weight, which can accelerate the identification of potential drug candidates. The interface also provides search examples and exploration tools categorized by natural product sources (e.g., plants, animals, bacteria, fungi, and marine organisms) to help guide new users.

Finally, NPBS Atlas supports data-driven discovery by offering seamless cross-linking to taxonomic catalogues and references, and offering downloadable chemical structures and annotations of the natural products, which are compatible with computational tools for machine learning, metabolomics, and virtual screening workflows.

Conclusions

In conclusion, we have developed NPBS Atlas, a comprehensive data resource of over 218,000 natural products. Its primary contribution is the systematic, manually-curated annotation of biological source information for each natural product, thereby addressing a critical gap by linking extensive chemical diversity to its biological context. The resource is designed to facilitate the exploration of natural products, particularly through its specialized annotations for detailed taxonomic data, bioactivities, and TCMs.

While NPBS Atlas provides a robust foundation, we acknowledge its current limitations. The bioactivity annotations often lack quantitative experimental parameters (e.g., IC50, Ki values) and standardized assay metadata, which restricts their utility in quantitative structure–activity relationship (QSAR) modeling. Furthermore, the broad taxonomic categorization may oversimplify the origins of organisms from complex symbiotic systems. Therefore, future iterations of NPBS Atlas will prioritize the systematic extraction and integration of detailed, quantitative bioactivity data from scientific literatures. These enhancements will further solidify the database's role as a valuable resource for natural product research and data-driven drug discovery.

Abbreviations

NPBS Atlas

Natural product and biological source atlas

NPs

Natural products

TCM

Traditional Chinese Medicine

QED

Quantitative estimate of drug-likeness

SA

Synthetic accessibility

CoL

Catalogue of Life

WoRMS

World Register of Marine Species

QSAR

Quantitative structure activity relationship

Author contributions

T.X. conceived, designed, and constructed the database. J.D., Y.L., J.Z, and Y.Z. contributed to data collection and annotation. W.C. contributed to data quality control. X.S.X. supervised the project. All authors read and approved the final manuscript.

Funding

This work was supported by the National Natural Science Foundation of China (22122104), the Informatization Plan of Chinese Academy of Sciences (CAS-WX2021SF-0307), the CAS Project for Young Scientists in Basic Research (YSBR-095 and YSBR-052), and the Strategic Priority Research Program of the Chinese Academy of Sciences (XDB0590000).

Availability of data and materials

Source data of the figures are provided in Additional file 1. The data of NPBS Atlas is freely available at https://biochemai.cstspace.cn/npbs/) under Open Source license of Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0).

Declarations

Ethical approval

This study did not require ethical approval as it did not involve human participants, animal experiments, or sensitive personal data.

Competing interests

The authors declare no competing interests.