Persistent Identifiers for Research Data: A Guide to Selection and Use

Researchers in bioinformatics and the life sciences face a recurring problem: the web addresses that point to their datasets, software, and supplementary materials tend to break over time. A link that resolves today may return a 404 error next year, after a server migration, a journal platform update, or a laboratory website reorganization. Persistent identifiers (PIDs) solve this problem by providing stable, unique references that remain resolvable regardless of where the underlying data are stored. This article explains what persistent identifiers are, how they differ from ordinary URLs, which types of identifiers suit which research outputs, and how to implement them in a data management workflow. The guidance is written for students, researchers, analysts, and life-science professionals who generate or reuse datasets and need to make those datasets citable and discoverable.

What Persistent Identifiers Do

A persistent identifier is a long-lived reference to a digital object, such as a dataset, a publication, a software release, or a person. Unlike a standard URL, which points to a specific location on a specific server, a persistent identifier is designed to remain stable even when the object moves. The identifier itself contains enough information for a resolution service to redirect a user to the current location of the object.

The core functions of persistent identifiers are identification, resolution, and citation. Identification means that each digital object receives a unique string that distinguishes it from every other object. Resolution means that the identifier can be looked up in a registry or resolver to find the object's current location. Citation means that the identifier can be used in reference lists, data management plans, and funding reports to give credit to data producers and to enable verification of research findings.

Persistent identifiers also support the FAIR Guiding Principles, which state that research data should be findable, accessible, interoperable, and reusable. The FAIR principles, published in Scientific Data, emphasize that data need clear and accessible identifiers as part of their findability and accessibility requirements. Without a persistent identifier, a dataset may be technically available but practically undiscoverable because its location changes or its existence is not registered anywhere.

Types of Persistent Identifiers

Several persistent identifier systems are in active use across the sciences. Each has a different governance model, cost structure, and technical implementation. The choice among them depends on the type of object being identified, the requirements of the funding agency or journal, and the infrastructure available at the researcher's institution.

Digital Object Identifiers (DOIs)

The Digital Object Identifier (DOI) system is the most widely recognized persistent identifier for research outputs. DOIs are administered through registration agencies that operate under the International DOI Foundation. A DOI takes the form of a prefix assigned to a registrant, followed by a suffix chosen by the registrant. For example, a DOI might look like 10.1038/sdata.2016.18.

DOIs are commonly assigned to journal articles, datasets, software releases, and other scholarly outputs. Many data repositories issue DOIs automatically when a dataset is deposited. The resolution service for DOIs is operated by the DOI system, and the identifier resolves to a landing page that describes the object and provides access to it.

For bioinformatics researchers, the practical implication is that depositing data in a repository that issues DOIs gives the dataset a stable citation path. The FAIR Guiding Principles paper itself carries a DOI, which demonstrates how the system works for scholarly literature.

Archival Resource Keys (ARKs)

The Archival Resource Key (ARK) is a persistent identifier scheme developed by the California Digital Library. ARKs are designed to identify information objects of any type, and they include a metadata component that describes the object. An ARK takes the form of a URL that contains the string ark: followed by a Name Mapping Authority Hostport, a namespace, and a name.

ARKs are often used by libraries, archives, and cultural heritage institutions. They are also used in some scientific contexts where a lightweight, no-cost identifier is needed. ARKs can be minted by any organization that operates a Name Mapping Authority, which gives institutions flexibility in managing their own identifiers.

Handle System

The Handle System is a general-purpose persistent identifier infrastructure developed by the Corporation for National Research Initiatives. Handles are used for a variety of purposes, including DOIs, which are built on the Handle System. A handle takes the form of a prefix and suffix separated by a slash, such as 20.1000/1234.

The Handle System provides a resolution mechanism that can be used independently of the DOI system. Some repositories and institutions operate their own handle servers, which gives them control over the identifiers they assign. The Handle System is also used in contexts where identifiers need to be resolved across multiple locations or where the identifier must support complex metadata.

Persistent Uniform Resource Locators (PURLs)

A Persistent Uniform Resource Locator (PURL) is a URL that acts as a redirector. Instead of pointing directly to the resource, the PURL points to a resolver that maintains a mapping between the PURL and the actual location of the resource. If the resource moves, the mapping is updated, and the PURL continues to work.

PURLs are simpler than DOIs or handles because they do not require a registration agency or a formal namespace. However, they depend on the continued operation of the PURL resolver. If the organization that maintains the resolver ceases to operate, the PURLs it manages may stop resolving. PURLs are therefore best suited for situations where the resolver is operated by a stable institution.

Compact Identifiers

Compact identifiers are a lightweight approach to identifying biomedical data. They combine a namespace prefix with a local identifier, such as SRA:SRR1234567. The identifiers.org resolution service maps these compact identifiers to the appropriate databases. This approach is useful for bioinformatics because it allows researchers to refer to database records without embedding a full URL.

The value of compact identifiers lies in their uniformity. A researcher can write UniProt:P12345 or PDB:1ABC, and the resolution service will direct the reader to the correct record. This uniformity supports automated processing and cross-referencing between databases.

At a Glance: Choosing a Persistent Identifier

The following table summarizes the main persistent identifier types and their suitability for different research outputs. Use it as a starting point for deciding which identifier to assign to a dataset, software release, or other digital object.

Selecting an Identifier for Your Data

The choice of persistent identifier should be driven by the requirements of the data, the expectations of the research community, and the infrastructure available to you. The following decision tree can help you work through the selection process.

Step 1: Determine the Object Type

Identify what you are trying to cite. A dataset that will be deposited in a public repository has different requirements than a laboratory protocol posted on a group website. A software release that will be archived in a code repository may need a DOI, while a database record that already has an accession number may only need a compact identifier.

Step 2: Check Repository and Funder Requirements

Many funding agencies and journals require that data be deposited in a repository that assigns persistent identifiers. The NIH Genomic Data Sharing Policy, for example, sets expectations for how genomic data are shared and managed. If your funder or journal specifies a particular identifier type, follow that requirement. Institutional data repositories often issue DOIs as part of their deposit workflow, which simplifies the process.

Step 3: Assess Infrastructure and Support

Consider what identifier infrastructure your institution already operates. If your library runs an institutional repository that mints DOIs, use that service. If your institution operates a handle server, you may be able to assign handles to internal resources. If no infrastructure exists, a generalist repository that issues DOIs may be the most practical option.

Step 4: Consider Long-Term Sustainability

Persistent identifiers are only useful if the resolution service continues to operate. DOIs are backed by a formal registration system with ongoing governance. ARKs depend on the institution that operates the Name Mapping Authority. PURLs depend on the resolver operator. Choose an identifier type whose governance model matches the expected lifespan of the data.

Step 5: Document Your Choice

Record the identifier type, the issuing agency, and the resolution URL in your data management plan. This documentation helps other researchers understand how to cite the data and how to verify the identifier's authenticity.

Implementing Persistent Identifiers in a Data Workflow

Implementing persistent identifiers requires more than minting a string. The identifier must be registered, linked to a landing page, and maintained over time. The following workflow describes the steps involved in assigning and managing persistent identifiers for research data.

Deposit Data in a Repository

The first step is to deposit the data in a repository that supports persistent identifiers. The repository may be a generalist repository, an institutional repository, or a domain-specific database. The NCBI Data Resources, for example, provide access to a wide range of biomedical data, and many NCBI records have stable accession numbers that function as identifiers.

When you deposit data, the repository will typically ask for metadata, including the title, authors, description, and license. The quality of this metadata affects the discoverability of the data and the usefulness of the identifier. Provide complete and accurate metadata at the time of deposit.

Obtain the Identifier

Once the data are deposited, the repository will assign an identifier. For DOIs, the repository registers the identifier with a DOI registration agency. The registration process creates a record that links the DOI to the landing page for the data.

Some repositories issue a draft DOI before the data are finalized. This allows the researcher to include the DOI in a manuscript before the data are publicly released. The DOI becomes active when the data are published.

Create a Landing Page

The landing page is the web page that the identifier resolves to. It should describe the data, provide access to the files, and include citation information. The landing page should be stable and should not change its URL. If the data move to a new location, the landing page should remain in place and redirect to the new location.

Link the Identifier to Related Objects

A dataset often relates to other objects, such as a publication, a software tool, or another dataset. Link these objects to the identifier by including the identifier in the metadata of each object. For example, a paper that uses a dataset should cite the dataset's DOI, and the dataset's landing page should list the paper that describes it.

Maintain the Identifier

Persistent identifiers require ongoing maintenance. If the data move, update the resolution record. If the landing page changes, ensure that the old URL still resolves. If the data are updated, decide whether to issue a new identifier for the new version or to update the existing record.

Options and Tradeoffs in Identifier Selection

Each persistent identifier type has strengths and weaknesses that matter in practice. Understanding these tradeoffs helps you make an informed choice.

DOI Tradeoffs

DOIs are the most widely accepted identifier for scholarly outputs. They are recognized by publishers, funders, and citation databases. The main drawback is cost. Registering a DOI requires a fee, which is usually paid by the repository or the institution. For individual researchers, the cost is typically hidden because the repository absorbs it.

DOIs also carry governance requirements. The registration agency must approve the metadata, and the DOI record must be maintained. This is usually handled by the repository, but it means that the researcher depends on the repository's continued operation.

ARK Tradeoffs

ARKs are free to mint, which makes them attractive for institutions that want to assign identifiers without paying registration fees. The institution must operate a Name Mapping Authority, which requires technical infrastructure and ongoing maintenance. ARKs are less widely recognized than DOIs in the scholarly publishing community, which may limit their use in formal citations.

Handle Tradeoffs

Handles offer flexibility because they can be used for any type of object and can be resolved through multiple mechanisms. Operating a handle server requires technical expertise and infrastructure. Handles are not as widely recognized as DOIs in the research community, but they are used in many institutional repositories.

PURL Tradeoffs

PURLs are simple to implement and require no registration fees. They depend on the continued operation of the resolver, which makes them less reliable than DOIs or handles. PURLs are best suited for short-term projects or for objects that are unlikely to move.

Compact Identifier Tradeoffs

Compact identifiers are convenient for referencing database records, but they depend on the identifiers.org resolution service and on the participating databases. If a database changes its accession scheme or ceases to operate, the compact identifier may stop resolving. Compact identifiers are best used as supplementary references, not as the primary identifier for a dataset.

Observations and Measurements in Identifier Use

The practical value of persistent identifiers can be measured in several ways. Tracking these measurements helps you assess whether your identifier strategy is working.

Resolution Success Rate

The resolution success rate is the percentage of identifier lookups that return a valid landing page. A high success rate indicates that the identifiers are well maintained. A low success rate suggests that identifiers are broken or that the resolution service is unreliable.

Citation Count

The number of times a dataset is cited using its persistent identifier indicates the dataset's impact. Citation counts can be tracked through scholarly databases, repository statistics, and altmetrics services. A dataset with a DOI that is cited in multiple papers demonstrates the value of the identifier.

Download and Access Statistics

Repository platforms typically track the number of views and downloads for each dataset. These statistics show how often the data are accessed and can be used to justify continued investment in data management. Restricted datasets may have different access patterns, with access requests and approvals tracked separately.

Data Reuse Indicators

Reuse can be measured by tracking how often a dataset is referenced in subsequent research, how often it is downloaded, and how often it is cited in data repositories. These indicators help you understand whether the persistent identifier is enabling the intended sharing and reuse.

Records and Documentation for Identifier Management

Maintaining persistent identifiers requires careful record keeping. The following records should be maintained for each identifier you manage.

Identifier Registry

Maintain a registry of all identifiers assigned to your research outputs. The registry should include the identifier string, the object it identifies, the date of assignment, the issuing agency, and the current resolution URL. This registry serves as the authoritative record for your identifiers.

Metadata Records

Keep the metadata associated with each identifier up to date. This includes the title, authors, description, license, and related identifiers. Metadata updates should be submitted to the registration agency or repository as needed.

Change Log

Record any changes to the identifier resolution, such as a change in the landing page URL or a move to a new repository. The change log helps you track the history of each identifier and provides evidence of proper maintenance.

Access Records

For restricted datasets, maintain records of access requests, approvals, and data use agreements. These records demonstrate compliance with data sharing policies and provide a basis for auditing data access.

Quality Controls for Persistent Identifier Implementation

Quality controls ensure that persistent identifiers are assigned correctly and remain functional. The following controls should be part of any identifier management workflow.

Metadata Validation

Before an identifier is registered, validate the metadata for completeness and accuracy. Check that the title is descriptive, the authors are correctly identified, and the license is specified. Incomplete metadata reduces the discoverability of the data and may cause problems with citation.

Resolution Testing

Test each identifier after it is assigned to confirm that it resolves to the correct landing page. Repeat the test periodically to catch resolution failures early. Automated testing tools can check large numbers of identifiers on a regular schedule.

Duplicate Detection

Check that the same object is not assigned multiple identifiers. Duplicate identifiers confuse citation and make it difficult to track data usage. If a duplicate is found, consolidate the records and redirect the duplicate identifier to the primary one.

Version Control

When a dataset is updated, decide whether to issue a new identifier or to update the existing record. If the update is substantial, a new identifier may be appropriate. If the update is minor, updating the existing record may be sufficient. Document the versioning policy in the data management plan.

Common Failure Patterns in Persistent Identifier Use

Persistent identifiers fail in predictable ways. Recognizing these failure patterns helps you avoid them.

Identifier Rot

Identifier rot occurs when an identifier stops resolving. This happens when the registration record is deleted, the landing page is removed, or the resolution service ceases to operate. Identifier rot is most common with PURLs and other lightweight identifier schemes that depend on a single resolver.

Metadata Drift

Metadata drift occurs when the metadata associated with an identifier becomes outdated or inaccurate. This happens when the data are updated but the metadata are not, or when the responsible person leaves the institution without transferring the maintenance responsibility.

Duplicate Identifiers

Duplicate identifiers occur when the same object is assigned multiple identifiers by different systems. This happens when a dataset is deposited in more than one repository or when a researcher mints a new identifier without checking whether one already exists.

Broken Landing Pages

A broken landing page occurs when the identifier resolves to a URL that no longer works. This happens when the landing page is moved without updating the resolution record, or when the server hosting the landing page is decommissioned.

Orphaned Identifiers

An orphaned identifier is one that no one maintains. This happens when the researcher who created the identifier leaves the institution, or when the project that produced the data ends without a plan for ongoing maintenance.

Limitations of Persistent Identifiers

Persistent identifiers are a necessary part of research data management, but they have limitations that researchers should understand.

They Do Not Guarantee Access

A persistent identifier ensures that a digital object can be found, but it does not guarantee that the object is accessible. The data may be behind a paywall, subject to access controls, or stored in a format that cannot be opened. The identifier points to the landing page, not to the data itself.

They Depend on Human Maintenance

Persistent identifiers require ongoing maintenance by people or organizations. If the responsible party stops maintaining the identifier, it will eventually fail. The longevity of an identifier depends on the commitment of the institution that manages it.

They Do Not Solve All Citation Problems

A persistent identifier makes a dataset citable, but it does not ensure that the citation is correct or complete. Researchers must still provide accurate metadata, and citation systems must be configured to display the identifier properly.

They Are Not a Substitute for Data Quality

A persistent identifier does not make bad data good. The identifier ensures that the data can be found and cited, but the quality of the data depends on the methods used to collect, process, and document it.

Safety and Regulatory Context for Data Sharing

Persistent identifiers are often used in contexts where data sharing is governed by policies and regulations. Understanding this context helps you use identifiers appropriately.

Genomic Data Sharing Policies

The NIH Genomic Data Sharing Policy establishes expectations for the sharing of genomic data generated with NIH funding. The policy requires that data be deposited in designated repositories and that data be shared in a manner consistent with informed consent. Persistent identifiers are part of the infrastructure that supports this sharing.

Human Subjects Data

Data from human subjects may be subject to access controls that limit who can view or download the data. Restricted access models allow researchers to share data while protecting participant privacy. Persistent identifiers can be assigned to restricted datasets, with the landing page describing the access request process.

Data Transfer and Use Agreements

When data are shared with other researchers, a data transfer and use agreement may be required. This agreement specifies the terms of use, including any restrictions on redistribution or reuse. Persistent identifiers help track which data are covered by which agreements.

Institutional Review Board Requirements

Research involving human subjects typically requires review by an institutional review board. The board may impose conditions on data sharing, including requirements for de-identification or restrictions on data access. Persistent identifiers should be assigned in a way that is consistent with these conditions.

Professional Escalation Criteria

Some situations require escalation to a data manager, librarian, or institutional official. The following criteria indicate when you should seek professional assistance.

You Cannot Determine the Appropriate Identifier

If you are unsure which identifier type to use, or if your funder's requirements are unclear, consult your institutional data librarian or research data manager. They can help you navigate the requirements and select an appropriate repository.

Your Data Are Subject to Access Restrictions

If your data include human subjects information, controlled access data, or data covered by a data use agreement, consult your institutional review board or data access committee before assigning an identifier. They can help you determine whether a restricted access model is appropriate.

Your Identifier Stops Resolving

If an identifier that you manage stops resolving, contact the repository or registration agency immediately. They can help you diagnose the problem and restore the resolution. Do not attempt to fix the resolution record yourself unless you have the authority to do so.

Your Institution Does Not Have Identifier Infrastructure

If your institution does not operate a repository or identifier service, consult your library or research office. They may be able to recommend a generalist repository or help you establish the necessary infrastructure.

You Are Preparing a Data Management Plan

If you are writing a data management plan for a funding application, consult your institutional research office or data librarian. They can help you describe your identifier strategy in a way that meets funder expectations.

Frequently Asked Questions

What is the difference between a persistent identifier and a URL?

A URL points to a specific location on a specific server. If the resource moves, the URL breaks. A persistent identifier is a stable reference that is resolved through a registry or resolver. The identifier remains the same even when the resource moves, because the resolution record is updated to point to the new location.

Which persistent identifier should I use for my dataset?

The choice depends on where you deposit the data and what your funder or journal requires. Most data repositories issue DOIs, which are the most widely accepted identifier for datasets. If you are depositing data in an institutional repository, check whether it issues DOIs or uses another identifier scheme. If you are referencing a database record, a compact identifier may be sufficient.

How do I get a DOI for my data?

Deposit your data in a repository that issues DOIs. The repository will handle the DOI registration as part of the deposit process. Many generalist and institutional repositories provide this service at no cost to the researcher. You do not need to register a DOI yourself.

Can I assign a persistent identifier to data that are not publicly available?

Yes. Restricted datasets can have persistent identifiers. The identifier resolves to a landing page that describes the data and explains how to request access. The landing page may require authentication or an access request process.

What happens if my data move to a new repository?

If your data move, update the resolution record for the identifier. For DOIs, this means updating the landing page URL with the registration agency. For ARKs and handles, update the resolution record with the Name Mapping Authority or handle server. The identifier itself does not change.

How long do persistent identifiers last?

Persistent identifiers are designed to last indefinitely, but their longevity depends on the continued operation of the resolution service. DOIs are backed by a formal registration system that is designed for long-term operation. ARKs and handles depend on the institution that operates the resolver. PURLs depend on the resolver operator and are less reliable over long periods.

Do I need a persistent identifier for my software?

Software releases can benefit from persistent identifiers, especially if the software is used in published research. Many software repositories and code archives can issue DOIs for specific releases. This allows other researchers to cite the exact version of the software they used.

How do persistent identifiers support data citation?

A persistent identifier gives a dataset a stable citation path. When you cite a dataset, you include the identifier in the reference. The identifier allows readers to find the dataset, and it provides a way to track how often the dataset is cited. This supports proper attribution and helps demonstrate the impact of the data.

Related Bioinformatics Guides

• Data Sharing and Privacy in Genomic Research
• How To Use Alphafold To Predict Structure: Structural Analysis and Computational Methodologies in Bioinformatics
• The Protein Data Bank (PDB): Archival Standards, Structural Validation Metrics, and Bioinformatics Integration Protocols
• The History and Evolution of Bioinformatics
• Bioinformatics in Agriculture and Crop Improvement

References and Further Reading

• EMBL-EBI Training. European Bioinformatics Institute.
• NCBI Data Resources. National Center for Biotechnology Information.
• Genomic Data Sharing Policy. National Institutes of Health.
• The FAIR Guiding Principles. Scientific Data.
• Diagnosing Overtraining Syndrome: A Scoping Review.. Sports health, 2022.
• Global prevalence of functional constipation according to the Rome criteria: a systematic review and meta-analysis.. The lancet. Gastroenterology & hepatology, 2021.
• Global prevalence of irritable bowel syndrome according to Rome III or IV criteria: a systematic review and meta-analysis.. The lancet. Gastroenterology & hepatology, 2020.
• Multidisciplinary biopsychosocial rehabilitation for chronic low back pain.. The Cochrane database of systematic reviews, 2014.
• Cognitive function following a major depressive episode: a systematic review and meta-analysis.. The lancet. Psychiatry, 2019.
• Systematic review with meta-analysis: the efficacy of prebiotics, probiotics, synbiotics and antibiotics in irritable bowel syndrome.. Alimentary pharmacology & therapeutics, 2018.
• Transabdominal pre-peritoneal (TAPP) versus totally extraperitoneal (TEP) laparoscopic techniques for inguinal hernia repair.. The Cochrane database of systematic reviews, 2024.
• Efficacy of gut-brain neuromodulators in irritable bowel syndrome: an updated systematic review and meta-analysis.. The lancet. Gastroenterology & hepatology, 2025.
• A generalist institutional data repository offering both open and restricted access to support NIH data sharing compliance.. 2026.
• The accuracy of artificial intelligence in identifying cephalometric landmarks: A scoping review.. 2026.
• RNA-seq dataset of superworm (&lt,i&gt,Zophobas atratus&lt,/i&gt,) feeding on polystyrene foam.. 2026.
• Verticall: A fast and robust tool for recombination detection in large-scale bacterial genomic datasets. 2026.
• An evaluation of methods to handle missing data in the context of latent variable interaction analysis: multiple imputation, maximum likelihood, and random forest algorithm. Japanese Journal of Statistics and Data Science, 2022.
• EMG Pattern Recognition in the Era of Big Data and Deep Learning. Big Data and Cognitive Computing, 2018.
• Quantum Machine Learning: A Review and Current Status. Data Management, Analytics and Innovation, 2020.
• Model-Based Approaches to Handling Uncertainty. Robust Speech Recognition of Uncertain or Missing Data, 2011.
• Enhancing Multimodal Sentiment Analysis via Learning from Large Language Model. IEEE International Conference on Multimedia and Expo, 2024.
• Information Access and Resource Sharing with Persistent Identifiers. LIS TODAY, 2024.
• Implementation of data citations and persistent identifiers at the ORNL DAAC. Ecological Informatics, 2016.
• Service for data retrieval via persistent identifiers. Data 2015 4th International Conference on Data Management Technologies and Applications Proceedings, 2015.
• Uniform resolution of compact identifiers for biomedical data. Scientific Data, 2018.
• Use of persistent identifiers and research blogging by Portuguese language scientific blogs and their Altmetric coverage. Encontros Bibli, 2020.
• Identifiers for earth science data sets: Where we have been and where we need to go. Data Science Journal, 2017.
• Timestamped urls as persistent identifiers. Ceur Workshop Proceedings, 2020.

This article is educational and does not replace validated analysis plans, institutional policy, clinical interpretation, or specialist review.