TL;DR
Marine natural products (MNPs) are a rich source for drug discovery, but accessing and organizing this data is challenging. This paper introduces CMNPD, a comprehensive, open-access database that consolidates chemical structures, biological activities, taxonomic information, and literature for MNPs, aiming to accelerate the discovery of new marine-derived drugs.
What Problem Does This Paper Address?
The ocean harbors an immense diversity of chemical compounds with potential therapeutic applications. However, the field of marine natural products (MNPs) chemistry faces significant hurdles in drug discovery. These include the sheer volume of discovered compounds, the difficulty in accessing and integrating disparate data sources (chemical structures, biological activities, organism origins, literature), and the cost associated with existing commercial databases. While some free databases exist, they are often limited in scope, outdated, or lack comprehensive biological activity data. This creates a bottleneck for researchers seeking to identify novel drug leads, understand structure-activity relationships, and explore chemical ecology.
Methods
The CMNPD database was built through a rigorous data extraction and curation process. Key methods include:
- Literature Mining: Data was primarily extracted from seminal reviews on marine natural products and supplemented by searches in Web of Science.
- Structure Recognition and Standardization: The CLiDE tool was used for optical chemical structure recognition, converting graphical representations to machine-readable formats. Structures were standardized, with stereochemistry carefully represented using wedge-dash diagrams. 2D structures were saved in MOL format and converted to SMILES, InChI, and InChIKey.
- Chemical Classification: The ClassyFire web server was used to classify compounds into chemical classes.
- 3D Conformer Generation: OMEGA software was employed to generate 3D conformers for a subset of compounds meeting specific criteria (size, flexibility, atom types) to ensure accurate representation of molecular shapes.
- Property Calculation: Physicochemical properties (molecular weight, ALogP, polar surface area, QED) were calculated using RDKit. Pharmacokinetic properties (BBB penetration, intestinal absorption, solubility, CYP2D6 binding, hepatotoxicity, plasma protein binding) were predicted using Pipeline Pilot ADMET models.
- Taxonomic Standardization: Organism names were normalized to accepted scientific names using authoritative databases like Catalogue of Life and World Register of Marine Species. Sampling locations were converted to coordinates using Google Maps.
- Biological Activity Data Integration: Initial bioassay data was extracted from literature. Crucially, standardized experimental data from the ChEMBL database was incorporated to provide a rich source of high-quality bioactivity information, including target details, activity types, and assay descriptions.
- Data Visualization: Interactive visualizations (sunburst charts, dot distribution maps, network graphs) were developed using Plotly and ECharts to aid user exploration and understanding of the data.
Datasets
- Primary Data Source: Scientific literature and patents, with a focus on reviews by Faulkner, Blunt, and Carroll.
- Integrated Datasets: ChEMBL database (release 27, 2020) for biological activity data.
- Taxonomic Databases: Catalogue of Life, World Register of Marine Species (WORMS), Integrated Taxonomic Information System (ITIS), Index Fungorum.
- Chemical Databases (for cross-referencing): Reaxys, PubChem, ChemSpider.
- Spectral Data: Wiley SpectraBase (for NMR, IR, Raman, UV-Vis, and mass spectra).
The paper introduces CMNPD as a new, comprehensive dataset specifically for marine natural products, differentiating itself by its breadth of integrated data types and its open-access nature.
Key Insights and Ideas
- The Ocean as a Drug Discovery Frontier: Despite extensive research, the ocean remains a vastly underexplored resource for novel drug candidates, with its unique environmental conditions driving extraordinary chemical diversity.
- Data Integration is Crucial: The true power of MNPs research lies in the ability to connect chemical structures with their biological activities, organism origins, and geographical distribution. CMNPD's strength is its comprehensive integration of these disparate data types.
- Open Access Empowers Research: By providing a free, user-friendly platform, CMNPD democratizes access to MNP data, enabling a wider scientific community to participate in drug discovery efforts.
- Standardized Biological Data is Key: The integration of standardized, high-quality biological activity data from sources like ChEMBL is essential for meaningful structure-activity relationship studies and target identification.
- Visualization Enhances Understanding: Interactive data visualizations are vital tools for navigating the complexity of MNP data, revealing patterns in taxonomy, chemical space, and organism distribution.
Key Results
- Database Size: CMNPD release 1.0 contains 31,561 distinct chemical entities of MNPs from over 13,000 source organisms.
- Active Compounds: 15,774 compounds are linked to biological activity data, mapped to 2,652 targets.
- Bioactivities: A total of 72,343 bioactivities are recorded.
- 3D Information: 79.9% of all records have 3D information generated according to specific criteria.
- Literature Coverage: The database includes 128,488 scientific literature and patents.
- User-Friendly Interface: A web interface modeled on ChEMBL provides intuitive browsing, searching, and data exploration capabilities.
Limitations and Future Work
- Data Completeness: While comprehensive, the database acknowledges that some unidentified species are temporarily classified into broader categories, and the species count is a minimum estimate.
- 3D Conformer Generation Limitations: 3D representations are generated only for compounds meeting specific size and flexibility criteria, meaning not all complex MNPs will have this feature.
- Future Work: The authors envision CMNPD growing continuously through extensive data deposition and resource integration, aiming to become an even more comprehensive repository and a driving force for marine drug development. They also encourage data deposition from researchers to further enhance the database's utility.
Summary updated