DrugBank

DrugBank is a curated knowledgebase that brings together detailed chemical, pharmacological, biochemical and clinical information about approved drugs, experimental compounds, biologics and nutraceuticals. Each entry (DrugCard) integrates a drug’s chemical structure and identifiers with pharmacodynamics (targets, mechanism of action), pharmacokinetics and metabolism (ADME enzymes, metabolites), clinical annotations (indications, dosage, contraindications) and safety information (side effects, interactions). The resource is designed to support translational research, drug discovery, pharmacology, and clinical informatics by making richly annotated, machine-readable drug data available for both human inspection and programmatic use. Core capabilities include standardized, cross-referenced DrugCards that aggregate chemical formats (SMILES, InChI, 2D/3D structures), physiochemical properties, mechanism-of-action descriptions, curated lists of protein targets (with links to sequence and structural databases), metabolic enzymes and pathways, and documented drug–drug and drug–food interactions. Entries are cross-mapped to external identifiers and resources such as UniProt, PubChem, ChEMBL, PDB, ATC, RxNorm and common registry numbers, enabling straightforward integration into bioinformatics workflows and knowledge graphs. DrugBank also provides downloadable structure files (SDF), tabular exports (TSV/CSV), and full XML snapshots to support bulk analysis, cheminformatics, and machine-learning model training. Typical use cases span the drug discovery and translational spectrum. In silico drug repurposing workflows use DrugBank’s target–drug links and mechanism annotations to find existing drugs that modulate disease-relevant proteins or pathways. Cheminformaticians and screening scientists use the provided SMILES, InChI and SDF files to annotate compound libraries, compute descriptors, and prioritize hits. Systems pharmacology and network medicine projects use the curated drug–target and interaction data to build drug–target–disease networks for polypharmacology analyses. Clinical researchers and pharmacovigilance teams leverage the clinical annotations and interaction information to support adverse-event signal interpretation and to design safer medication regimens. Data scientists also use DrugBank as a labeled dataset for training models predicting ADMET properties, drug–target affinity, or adverse reactions. DrugBank offers multiple access modes to accommodate different workflows: a web interface for manual search and inspection, programmatic access for automated queries and integration, and bulk download packages for large-scale analyses. The data is provided in formats commonly used in computational biology and cheminformatics (XML, CSV/TSV, SDF) so it can be loaded directly into tools like RDKit, bioconductor packages, or integrated into pipelines that consume UniProt or PubChem identifiers. Because entries include stable identifiers and cross-references, it’s straightforward to map DrugBank data to transcriptomic, proteomic, or pathway datasets for multi-omic annotation or enrichment analyses. Practical tips: use DrugBank IDs or canonical SMILES to maintain unambiguous linking between datasets; combine the curated target lists with UniProt accessions to pull protein sequences or structures for docking and modeling; and pair DrugBank interaction data with real-world evidence (EHRs or adverse event reports) to validate hypotheses. Be mindful of licensing and data-use terms: DrugBank provides academic and commercial licensing models, and bulk redistribution or commercial use typically requires an appropriate agreement. When citing or reusing data, include version or release information to ensure reproducibility and to track provenance across analyses.