- Biological structure, pathway, and interaction databases are essential resources in bioinformatics for understanding how biological molecules are organized, how biochemical processes operate, and how genes, proteins, and other cellular components interact. These databases complement nucleotide and protein sequence databases by providing information about three-dimensional molecular structures, biochemical reactions, signaling pathways, metabolic networks, and molecular interaction networks. Together, they help researchers connect sequence-level information with molecular mechanisms, cellular functions, and complex biological systems.
- Biological structure databases store information about the three-dimensional organization of biological molecules, including proteins, nucleic acids, and molecular complexes. Structural information helps researchers investigate protein folding, active sites, ligand binding, molecular recognition, and interactions between macromolecules. One of the most important resources is the Protein Data Bank (PDB), which archives experimentally determined structures of proteins, nucleic acids, and other biological macromolecules, as well as their complexes. PDB structures are obtained using methods such as X-ray crystallography, nuclear magnetic resonance spectroscopy, and cryo-electron microscopy. Each structure record may include atomic coordinates, experimental methods, resolution or other quality measures where applicable, ligand information, biological assembly details, and references to related publications.
- The Worldwide Protein Data Bank (wwPDB) coordinates the management and distribution of PDB structural data through its participating organizations. The Research Collaboratory for Structural Bioinformatics Protein Data Bank (RCSB PDB), PDBe at the European Bioinformatics Institute, and the Protein Data Bank Japan (PDBj) provide access to the shared archive through different interfaces and analytical services. Researchers can search these resources using protein names, organisms, structural identifiers, sequences, ligands, or experimental methods. Structure records can be downloaded in formats such as PDB and mmCIF, with mmCIF supporting detailed structural information and serving as the preferred format for many modern structural workflows.
- The PDB archive provides important information about protein structure and molecular function, but its records require careful interpretation. Some structures represent only a fragment of a protein, a particular conformational state, or a complex formed under specific experimental conditions. A structure may contain unresolved residues, engineered mutations, bound ligands, or experimental constructs that differ from the natural biological molecule. Researchers should therefore inspect the experimental method, resolution or other relevant quality indicators, biological assembly, missing regions, and experimental context before using a structure to support functional conclusions.
- Predicted structure resources complement experimentally determined structures. The AlphaFold Protein Structure Database provides predicted three-dimensional structures for many proteins, allowing researchers to examine proteins for which experimental structural information is unavailable. Such predictions can support domain analysis, structural comparisons, and the generation of hypotheses about possible functional regions. However, confidence varies across residues and regions, and predicted structures do not automatically establish biological activity, molecular interactions, or the correct conformation under physiological conditions. Structural confidence measures and the biological context of the protein should be considered when interpreting predicted models.
- Other structure-related resources focus on particular aspects of molecular organization. SCOP and CATH classify protein structural domains and evolutionary relationships using different classification frameworks. These resources help researchers identify structural folds, domain architectures, and relationships between proteins that may have limited sequence similarity. InterPro, Pfam, and the NCBI Conserved Domain Database complement structural classifications by identifying protein families and domains from sequence signatures. Structural classification can provide useful evolutionary and functional clues, although proteins with similar folds can have different biochemical activities.
- Biological pathway databases describe the sequences of biochemical reactions and molecular events through which cells perform essential functions. A pathway may involve metabolic reactions, signal transduction, gene regulation, immune responses, cellular transport, or other biological processes. Pathway databases connect genes and proteins with reactions, metabolites, cellular components, and biological outcomes. They help researchers interpret the functional significance of individual genes and understand how molecular components contribute to larger biological systems.
- KEGG, or Kyoto Encyclopedia of Genes and Genomes, is a major resource for biological pathways and molecular networks. It provides pathway maps covering metabolism, genetic information processing, environmental information processing, cellular processes, and other biological categories. KEGG also links genes, proteins, compounds, reactions, and diseases through related database resources. Researchers use KEGG for pathway enrichment analysis, metabolic reconstruction, comparative genomics, and the interpretation of gene expression or proteomics datasets. Pathway maps may represent reference knowledge rather than a complete, experimentally verified pathway operating identically in every organism or condition.
- Reactome is a curated pathway knowledgebase that describes biological reactions and their relationships in a structured, computationally accessible form. It covers processes such as signal transduction, metabolism, DNA replication and repair, immune responses, and cell-cycle regulation. Reactome connects proteins, complexes, small molecules, reactions, and cellular locations, enabling researchers to examine molecular events at different levels of biological organization. Its curated structure supports pathway visualization, gene set analysis, and the interpretation of omics data. Because pathway knowledge is continually updated, researchers should record the resource release used in an analysis.
- WikiPathways is a community-curated pathway resource that supports the creation, sharing, and reuse of biological pathway diagrams. Researchers can explore pathways related to metabolism, disease, signaling, and other biological processes, and can contribute corrections or new pathway knowledge. Its collaborative model supports the development of pathway collections that complement more centrally curated databases. The quality and evidence supporting individual pathways can vary, so users should inspect pathway descriptions, references, and provenance when using them in research.
- BioCyc provides collections of organism-specific and general biological pathway databases, including resources used for metabolic pathway analysis and genome interpretation. Some BioCyc databases focus on particular organisms, while others provide broader pathway knowledge. These resources are useful for exploring metabolic reactions, enzymes, compounds, and predicted biochemical capabilities. Organism-specific pathway assignments should still be interpreted carefully because genome annotations and pathway completeness may vary among species and strains.
- Other resources contribute specialized pathway and network information. The Human Metabolome Database (HMDB) provides information about metabolites, their chemical properties, biological roles, and associations with diseases and biological processes. BRENDA focuses on enzyme information, including biochemical properties and experimental observations. MetaCyc provides curated information about metabolic pathways and enzymes across diverse organisms. These resources can be used together to investigate biochemical reactions, identify candidate enzymes, and interpret metabolic changes observed in metabolomics or other experimental datasets.
- Biological interaction databases describe relationships between molecular components, including protein–protein interactions, protein–DNA interactions, protein–RNA interactions, genetic interactions, and interactions between proteins and small molecules. Interaction data may be obtained from laboratory experiments, computational predictions, curated literature, or high-throughput screening. These records help researchers investigate molecular complexes, signaling networks, regulatory mechanisms, and the organization of cellular systems. Interaction databases vary in their experimental coverage, evidence representation, and criteria for including records.
- STRING is a widely used resource for known and predicted protein–protein functional associations. Its evidence can include experimental data, curated pathway knowledge, genomic context, co-expression, text mining, and other computational sources. STRING combines evidence into association scores that estimate the strength of support for functional relationships. A STRING association does not necessarily mean that two proteins bind directly; the relationship may instead reflect participation in the same pathway or biological process. Researchers should examine the contributing evidence channels and distinguish functional association from experimentally demonstrated physical interaction.
- IntAct is a molecular interaction database that stores interaction evidence curated from scientific literature and submitted datasets. It includes information about interacting molecules, experimental methods, biological roles, and supporting publications where available. IntAct is useful for investigating protein interactions and identifying evidence underlying reported molecular relationships. Its records can be linked with protein accessions and other biological identifiers, supporting integration with sequence and functional annotation resources.
- BioGRID is a curated resource for protein and genetic interactions collected from published research. It contains interaction records across multiple organisms and supports the analysis of molecular and genetic networks. Genetic interactions can reveal relationships between genes that affect a shared phenotype or biological process, even when their encoded proteins do not physically interact. BioGRID is useful for studying genetic networks, functional relationships, and molecular interaction systems, but each record should be interpreted in the context of its experimental method and organism.
- Other interaction resources include the Database of Interacting Proteins (DIP), the Molecular INTeraction database (MINT), and specialized collections of protein interaction and complex data. Resource coverage and maintenance status can change over time, and some historical databases may have been integrated into broader platforms or no longer receive independent updates. Researchers should verify the current availability, release information, and scope of a resource before incorporating it into a new workflow.
- The Human Protein Atlas and related tissue or cellular expression resources provide information about protein abundance and localization in biological samples. Although these are not exclusively interaction databases, their data can help researchers assess whether a proposed interaction is plausible in a particular tissue or cellular context. Protein interaction hypotheses are generally more convincing when the relevant molecules are expressed in compatible cells and compartments, although expression overlap alone does not prove an interaction.
- Structural, pathway, and interaction databases become especially powerful when used together. A protein structure can reveal a potential binding interface, an interaction database can identify previously reported partners, and a pathway database can place those partners within a biological process. For example, a newly characterized signaling protein might be examined using a structural database to identify conserved domains, an interaction database to identify candidate binding partners, and a pathway resource to investigate whether those partners participate in a known signaling pathway. Combining these sources can help generate hypotheses about the protein’s biological role and guide experiments designed to test them.
- These databases also support multi-omics analysis. Transcriptomics measures RNA abundance, proteomics examines proteins and their properties, metabolomics measures small molecules, and other approaches capture additional biological layers. Pathway databases help connect differentially expressed genes or proteins to biological processes, while interaction networks can reveal groups of molecules that may act together. Structural information can then help investigate the potential mechanisms underlying these associations. However, a change in gene expression does not necessarily produce a corresponding change in protein abundance or activity, and network connectivity alone does not establish causal relationships.
- Pathway enrichment analysis is a common application of biological pathway databases. Researchers identify a set of genes or proteins of interest and test whether particular pathways contain more of these molecules than expected under a defined background model. Methods may use over-representation tests, rank-based enrichment procedures, or other statistical approaches. The choice of background gene set, annotation coverage, pathway size, and correction for multiple testing can affect the results. Enrichment indicates statistical association with a pathway, not necessarily that the pathway is activated, inhibited, or causally responsible for an observed phenotype.
- Interaction network analysis examines molecular components as nodes and their relationships as edges. Network properties such as degree, centrality, clustering, and connectivity can help identify candidate hubs, modules, or influential components. These properties depend on network coverage, edge definitions, evidence thresholds, and the selected database. Highly connected proteins may be biologically important, but they can also appear central because they have been studied more extensively. Network-based conclusions should therefore be evaluated alongside experimental evidence, biological context, and potential biases in the underlying data.
- Data integration is a major challenge because different databases use different identifiers, annotation standards, data models, and update schedules. Protein identifiers may differ between UniProt, NCBI, PDB, and interaction resources, while pathway databases may use their own identifiers for reactions, compounds, and biological events. Cross-references and identifier mapping tools help connect these records, but mappings may be ambiguous when proteins have multiple isoforms, genes have changed names, or identifiers refer to different sequence versions. Researchers should retain source identifiers and document mapping procedures to avoid introducing errors during integration.
- Evidence quality is another critical consideration. A structure may be experimentally determined or computationally predicted, a pathway may be supported by curated publications or computational inference, and an interaction may be observed directly or inferred from indirect evidence. The presence of a record in a database should not be interpreted as proof that every associated biological claim has been experimentally confirmed. Researchers should inspect experimental methods, evidence types, supporting references, confidence scores, and the conditions under which the data were obtained. Conflicting results may reflect differences in experimental systems, biological context, or data interpretation.
- Reproducibility requires careful documentation of the database resources and data versions used. Researchers should record access dates, release numbers, accession identifiers, query parameters, filtering thresholds, and software versions. For large-scale analyses, original downloaded files and analysis scripts should be preserved where licensing and data-use requirements permit. This is especially important because structures, pathway annotations, interaction records, and identifier mappings are updated as new evidence becomes available.
- Biological structure, pathway, and interaction databases provide complementary perspectives on molecular biology. Structure resources explain how biological molecules are organized in three dimensions, pathway resources describe biochemical reactions and cellular processes, and interaction databases capture relationships among molecular components. When integrated with sequence databases, functional annotation resources, and omics data, these resources enable researchers to investigate protein function, molecular mechanisms, disease biology, evolutionary relationships, and complex cellular networks. Their effective use depends on selecting suitable resources, interpreting evidence carefully, and documenting analyses in a reproducible manner.