Persistent Identifiers and Accession Numbers

  • Persistent identifiers and accession numbers are essential components of biological data management because they provide reliable ways to identify, retrieve, reference, and connect biological records across databases and research systems. Bioinformatics relies on enormous quantities of information, including DNA and RNA sequences, protein records, genomic assemblies, molecular structures, genetic variants, biological samples, gene expression datasets, and scientific publications. Names and descriptions can change, different databases may use different naming conventions, and multiple biological entities may have similar names. Persistent identifiers and accession numbers help distinguish records, support database interoperability, improve traceability, and make scientific results easier to reproduce and verify.
  • An identifier is a value assigned to an entity or record so that it can be distinguished from other entities or records. In bioinformatics, identifiers may refer to biological entities such as genes, transcripts, proteins, organisms, and variants, or to digital resources such as datasets, database entries, genome assemblies, software packages, and publications. Identifiers may be assigned by individual databases, international data-sharing initiatives, research organizations, or persistent identifier services. Their meaning depends on the system that defines them and the type of entity they represent. A gene identifier, for example, should not be assumed to identify a specific protein sequence or a particular experimental observation.
  • A persistent identifier (PID) is an identifier designed to remain a stable reference to a digital object, biological record, person, organization, or other resource over time. Persistence means that the identifier is intended to remain usable even if the resource’s location, hosting institution, or technical infrastructure changes. Many persistent identifiers can be resolved through a service that directs users to the current record or landing page. This separation between identity and location is important because URLs and web addresses may change, whereas a well-managed persistent identifier can continue to provide a reliable reference. Persistence is supported by governance, maintenance, metadata management, and long-term resolution services; it is not guaranteed merely because an identifier has a particular format.
  • An accession number is a unique identifier assigned by a database or repository to a submitted or maintained record. It allows users to locate and cite a specific entry within that resource. Accession numbers are widely used for nucleotide sequences, protein records, biological structures, sequencing datasets, gene expression studies, and other research objects. In many databases, an accession number remains associated with the record even when some of its annotations are updated. However, accession numbers do not all follow the same persistence rules. Some identify a record whose content can change, some distinguish particular sequence versions, and others may be replaced or retired according to database policies. Users should therefore consult the relevant resource’s documentation when the distinction matters.
  • Persistent identifiers and accession numbers overlap, but they are not interchangeable terms in every context. An accession number is assigned within a particular database or repository, while a persistent identifier describes the intended long-term stability of a reference. An accession number may function as a persistent identifier when it is maintained reliably and provides durable access to the record. Conversely, a persistent identifier may be assigned by a cross-disciplinary service rather than by the database that stores the underlying biological data. For example, a DOI can identify a research dataset, a publication, or another scholarly object, while a nucleotide sequence may have an accession assigned by a sequence archive. The identifier type should be selected according to the object being identified and the intended use.
  • Biological databases use several identifier systems to organize different types of records. GenBank, the European Nucleotide Archive (ENA), and the DNA Data Bank of Japan (DDBJ) assign accession identifiers to nucleotide sequence records. The International Nucleotide Sequence Database Collaboration coordinates data exchange among these archives. UniProt provides accessions for protein records, while the Protein Data Bank (PDB) uses identifiers for macromolecular structure entries. NCBI Gene assigns Gene IDs to gene records, and Ensembl uses identifiers for genes, transcripts, proteins, and other genomic features. The Sequence Read Archive (SRA) uses distinct accession prefixes for studies, samples, experiments, and sequencing runs. These examples illustrate why researchers must identify both the accession value and the database namespace in which it is defined.
  • Identifier namespaces are important because the same character string may have different meanings in different systems. A namespace specifies the context in which an identifier is unique and interpretable. For example, a gene identifier from one database cannot automatically be treated as equivalent to a similarly formatted identifier from another database. Reliable citation and data exchange should therefore include the database name or namespace, and where appropriate, the entity type and version. In structured data, identifiers are often represented as separate fields for the identifier value, source database, and version so that software can interpret them correctly.
  • Sequence accessions provide a clear example of the relationship between identifiers and versions. A nucleotide sequence record may have an accession followed by a version suffix, such as a hypothetical identifier AB123456.2. The base accession identifies the record, while the suffix distinguishes a particular version of its sequence. If the sequence changes, the version number may increase according to the database’s rules. Annotation changes that do not alter the sequence may be handled differently, depending on the resource. Recording the full versioned accession is important when the exact sequence used in an analysis must be reproduced. A reference to the base accession alone may not be sufficient to establish which sequence version was used.
  • Genome assemblies also require careful identification. Human reference genome assemblies are commonly referred to by names such as GRCh37 and GRCh38, but the assembly name alone may not capture every relevant detail. Researchers may need to record the assembly accession, patch level, annotation release, and source of the sequence files. Genome annotations can change independently of the underlying assembly, and different providers may offer different gene models for the same assembly. Because genomic coordinates depend on the reference sequence and annotation conventions, reporting only a gene name or a coordinate without the relevant reference information can lead to ambiguity. Version-aware identifiers help preserve the exact genomic context of an analysis.
  • Persistent identifiers are also used for biological samples and experimental datasets. In sequencing repositories, separate identifiers may represent a project or study, a biological sample, an experiment, and an individual sequencing run. These are distinct levels of organization. One biological sample may have multiple experiments, and one experiment may generate multiple runs or files. Keeping these identifiers distinct helps researchers trace raw sequencing data back to the sample and experimental design. Similar principles apply to gene expression repositories, where a study, sample, assay, and data file may each have their own record. A reliable workflow should preserve these relationships rather than treating every accession as if it identified the same object.
  • Digital Object Identifiers (DOIs) are widely used to provide persistent references to publications, datasets, software releases, and other scholarly resources. A DOI typically resolves through a DOI resolution service to a landing page containing descriptive metadata and access information. In bioinformatics, a dataset DOI can help identify a published or archived data collection, while a software DOI may identify a particular release. DOIs complement database accessions rather than replacing them. A paper may describe a sequencing study, the study may have an archive accession, and a DOI may identify the publication or a separate dataset record. Citing the correct identifier for each object helps readers retrieve both the scientific explanation and the underlying evidence.
  • Other persistent identifier systems serve different purposes. ORCID identifiers distinguish researchers and help associate publications, datasets, and contributions with the correct person. Research Organization Registry (ROR) identifiers identify research organizations. Handle identifiers and ARK identifiers can be used for durable references to digital or archival objects, depending on the repository and its infrastructure. International Standard Name Identifiers (ISNIs) identify public identities of people and organizations in relevant contexts. These systems are not all specific to bioinformatics, but they support the research ecosystem by improving attribution, institutional identification, and long-term discovery of scholarly resources.
  • Persistent identifiers support cross-referencing between biological databases. A gene record may link to transcript and protein accessions, a protein record may link to a structural record, and a gene or protein may connect to pathway, disease, and literature resources. Stable identifiers make these relationships more reliable than matching names alone. They also enable automated workflows to retrieve records, merge annotations, and construct integrated biological knowledge graphs. However, cross-references should not be interpreted as proof that two records represent identical biological entities. A gene may encode multiple protein isoforms, and a structure may represent only a fragment or a complex containing several molecular components. The relationship type and supporting evidence should therefore be preserved.
  • Identifier mapping is the process of connecting identifiers from different databases or versions of a resource. For example, a researcher may need to map gene identifiers from an expression dataset to protein accessions for functional analysis. Tools such as UniProt ID mapping and Ensembl BioMart support many types of identifier conversion and cross-resource queries. Mapping may be one-to-one, one-to-many, many-to-one, or unresolved. These patterns can arise because of alternative transcripts, gene duplication, record consolidation, changes in database policy, or differences in how biological entities are represented. Researchers should retain original identifiers, record mapping dates and resources, and investigate ambiguous mappings instead of silently selecting one result.
  • Identifier retirement and replacement require special attention. A database may merge duplicate records, retire obsolete identifiers, or redirect an old accession to a newer record. A replacement does not necessarily mean that every historical annotation remains equivalent to the current one. For example, a gene record may be merged with another record after improved genome annotation, while a sequence record may be updated to reflect corrected experimental information. Researchers should consult database histories, replacement links, and release notes when using older identifiers. Maintaining the original identifier alongside the current mapping preserves traceability and helps explain differences between historical and contemporary analyses.
  • Persistent identifiers are important for data citation and scientific reproducibility. A publication should identify the exact biological records and datasets on which its conclusions depend. For a sequence analysis, this may include accession numbers and sequence versions. For a variant study, it may include the reference genome assembly, variant representation, relevant transcript accession, and annotation database release. For a gene expression study, it may include the dataset accession, sample identifiers, and processing information. If a DOI is available for an archived dataset or software release, it can provide an additional durable citation. Precise references allow other researchers to retrieve the relevant evidence, verify interpretations, and repeat the analysis using the same inputs.
  • Identifier systems also contribute to data provenance. Provenance records describe where data originated, how they were processed, and which resources were used to generate an analytical result. Recording accession numbers, source databases, versions, retrieval dates, and checksums can establish a traceable connection between raw data, intermediate files, annotations, and final outputs. For example, a variant table may be linked to a reference assembly, a particular transcript annotation release, a variant caller version, and a clinical interpretation resource. These details make it easier to determine whether a difference between two analyses arose from the biological data, the reference resource, or the computational method.
  • Persistent identifiers must be managed alongside data quality and access considerations. A stable identifier does not guarantee that the associated record is accurate, complete, current, or scientifically validated. Records may contain errors, conflicting annotations, or incomplete metadata. Some identifiers resolve to a landing page rather than directly to a downloadable file, and access to certain datasets may be restricted by consent, privacy, licensing, or repository policies. Researchers should verify the identity and relevance of a record before using it, check its version and status, and ensure that access and reuse comply with applicable conditions.
  • Automated bioinformatics pipelines should treat identifiers as structured data rather than informal text labels. A robust pipeline can validate identifier formats, check whether records exist, retrieve associated metadata, detect obsolete identifiers, and report unmapped or ambiguous values. It should preserve the original identifiers and record any transformations. When possible, it should use official APIs or documented mapping services instead of relying on manually copied tables. If remote services can change their responses, the workflow should record retrieval dates and preserve the resulting mappings or data snapshots when permitted. These practices reduce silent errors and improve the reproducibility of large-scale analyses.
  • Good identifier management also requires careful handling of versions, namespaces, and metadata. Researchers should distinguish a database accession from a gene symbol, a versioned sequence record from an unversioned record, and a dataset identifier from an individual file identifier. Metadata should state what each identifier refers to and identify the resource that assigned it. Where identifiers are used in a database, they should be stored in suitable fields with appropriate constraints and relationships. Where data are exchanged between systems, the identifier namespace and entity type should accompany the value so that receiving applications can interpret it correctly.
  • Persistent identifiers and accession numbers form a foundation for reliable biological data management. They make records distinguishable, improve cross-referencing, support automated retrieval, preserve scientific citations, and connect research results with their underlying evidence. Their value is greatest when they are used with appropriate version information, provenance records, metadata standards, and validation procedures. As bioinformatics continues to integrate increasingly large and diverse datasets, consistent identifier management will remain essential for interoperability, long-term accessibility, and reproducible biological research.
Author:

Leave a Reply

Your email address will not be published. Required fields are marked *