# Step-by-Step Workflow for Depositing a Dataset into Dryad By the end of this guide you will be able to move a finished research dataset from your project folder to a published Dryad record with its own DOI, ready to cite in a manuscript and to hand to a funder. The workflow covers file preparation, documentation, privacy review, the submission form, curation, and verification of the final result. Dryad is a generalist research data repository operated by the California Digital Library at the University of California. It accepts data from any discipline, reviews every submission before publication, and issues a persistent Digital Object Identifier (DOI), a permanent web address that keeps pointing to the dataset even if the hosting site changes. Have these on hand before you start: - The cleaned dataset files, in open formats such as CSV or TXT where possible - A plain-text README describing each file and variable - An ORCID iD, the free researcher identifier from orcid.org - The DOI of the related manuscript, if one exists - Funder names and award numbers - A terminal window if you want to generate checksums, the short digital fingerprints used to verify file integrity No special software is required. A browser handles the entire submission. ## Why Researchers Deposit Data in a Repository Research data sharing has shifted from a courtesy to a requirement. Journals ask for public data availability statements, funders ask for [data management](/blog/guides/data-management-basics-principles-processes-and-best-practices) plans, and the FAIR principles define what a well shared dataset looks like [1]. FAIR stands for Findable, Accessible, Interoperable, and Reusable. Findable means the dataset turns up in search, which requires a title people actually type. Accessible means a stranger can retrieve it through a stable address. Interoperable means the formats and vocabularies are ones other tools can read. Reusable means the documentation and license let someone else apply the data to a new question. A repository deposit meets these goals in a way a lab server never will, because it adds a permanent identifier, a public landing page, and long-term preservation. Researcher attitudes back the practice up. In one international survey, 222 traumatic stress researchers from 28 countries completed a questionnaire on sharing and reuse, and the majority held a positive outlook, endorsing scholarly altruism and the ethical case for making data available [2]. Across fields, documented motivations include research integrity, replicability, and reuse, along with a recurring practical barrier, the extra work of preparing data once a paper is finished [3]. Depositing in a public repository also raises transparency and opens the door to secondary analyses the original team never planned [4]. One caution from the same literature: the FAIR principles state goals, not steps. Their implementation depends on the characteristics of the platform you choose, which is why every principle below is paired with a concrete action in the workflow [1]. ## Before You Deposit: Files, Documentation, and Ethics ### Choose and structure the files Decide what a reuser needs, not what looks tidy. Raw instrument output, processed tables, and the scripts that connect them each serve a purpose, and a deposit containing only polished summary tables usually fails the reuse test. Prefer open formats, CSV for tables and TXT or Markdown for documents, because proprietary formats become harder to open over time. Give files descriptive, stable names such as soil_respiration_measurements_2025-04.csv rather than data_final_v3.xlsx. Documentation matters as much as the data. Studies of computational workflows show that uncontrolled changes in input data, configuration parameters, workflow descriptions, and software versions cause workflow decay, the gradual failure of shared analyses when others try to rerun them. Record software names and versions, parameter values, and processing order in your README. Draft the README in a text editor before opening the submission form. The following template fits a small environmental dataset: ```markdown # Dataset: Soil respiration responses to spring thaw in northern # hardwood forests, 2024-2026 ## Overview Hourly soil CO2 efflux measured with closed-dynamic chambers at 12 plots across three sites in New Hampshire, USA, from January 2024 to May 2026. ## Files - soil_respiration_measurements.csv : one row per plot per hour - site_metadata.csv : plot coordinates, elevation, dominant tree species - analysis_scripts.R : code computing daily flux sums, written for R 4.3.2 ## Variables - plot_id : plot label, A1 through D3 - timestamp_utc : ISO 8601, for example 2025-04-02T14:00:00Z - flux_umol_m2_s : soil CO2 efflux in micromoles per square meter per second - soil_temp_c : soil temperature at 5 cm depth in degrees Celsius ## Methods summary Chamber closed for 90 seconds per measurement, four cycles per hour. CO2 measured with an infrared gas analyzer calibrated weekly against a 487 ppm reference standard. Full details appear in the related manuscript. ## License Data are shared under CC0. Contact: a.researcher@university.edu ``` Generate checksums next, so you can later prove the uploaded files match your originals: ```bash cd ~/projects/thaw_respiration/data_for_dryad md5sum *.csv > checksums.md5 cat checksums.md5 ``` Expected output: ``` 9e107d9d372bb6826bd81d3542a419d6 site_metadata.csv b1946ac92492d2347c6235b4d2611184e soil_respiration_measurements.csv ``` macOS users can run md5 instead of md5sum. ### Clear privacy and ethics review first Sensitive data need attention before any repository sees them. Removing names and obvious dates is a start, but it is rarely enough. Work in traumatic stress research recommends systematic, data analytic assessment of re-identification risk, the chance that de-identified records can be traced back to individuals, guided by a published checklist before datasets go public [5]. Interview and focus group data demand the most care. Researchers who prepared qualitative health data for deposit reported that de-identification is manual, time consuming, and costly, and that few repositories provide field-specific guidance [6]. Clinical and genomic data carry legal limits. Under the HIPAA Privacy Rule, the minimum necessary standard restricts what may be shared, and how a sharing activity is characterized, whether research, public health, or clinical support, changes how the standard applies [7]. Confirm with your institutional review board that public deposit is permitted, and keep a record of the consent language participants agreed to. ## Data Repository Comparison: Is Dryad the Right Home? ![Decision path asking whether a domain repository exists, whether the journal partners with Dryad, and whether the institution requires a repository.](/article-images/illustrations/how-to-deposit-data-in-dryad-repository-choice.webp) A data repository stores datasets, exposes them through search, and mints [persistent identifiers](/knowledge/bioinformatics/persistent-identifiers-in-bioinformatics-selection-criteria-and-implementation-workflow). Three broad types exist, and the best choice depends on your data type and the policies that apply to you. Domain repositories accept one data type and enforce community standards. PKRxiv, a repository for pharmacokinetic data, shows the pattern. It was designed around FAIR principles, supports structured submission of pharmacokinetic, pharmacogenetic, and safety data, issues persistent DOIs for discoverability and citation, and offers three sharing models, unrestricted, noncommercial, and contributor-controlled, with optional embargo periods [8]. By late September 2025 it held more than 5,500 drug concentration-time data points from over 900 participants across three continents [8]. When a repository like this exists for your field and accepts your data, use it, because its templates and review process fit the data natively. Searchable registries of research data repositories can help you locate one. Generalist repositories accept any discipline. Dryad is the leading example connected to journal publishing. Evidence from life cycle assessment research shows why generalists matter: FAIR-compliant deposits in that field happened mainly through generic repositories that provide the supporting tools, while domain-specific sharing guidance was still missing [9]. Institutional repositories are run by universities and serve policy compliance. A survey of 196 librarians at Nigerian federal university libraries found that depositing research outputs into a central institutional repository without restrictions ranked among the top three strategies for improving data sharing [10]. | Option | Best suited for | Access control | Review before release | Cost model | |--|--|--|--|--| | Dryad | Generalist datasets tied to journal articles | Public, with optional embargo | Staff curation | Free through partner journals and institutions, one-time charge otherwise | | Domain repository, for example PKRxiv | [Structured data](/blog/data-analysis/what-is-structured-data) types with community standards | Unrestricted, noncommercial, or contributor-controlled | Discipline-specific | Set by each repository | | Institutional repository | Outputs required by university policy | Managed by the institution | Varies | Usually free to affiliates | | Journal supplementary material | Small files reviewed with the manuscript | Public with the article | Journal peer review | Usually free | The practical rule: check for a domain repository first, choose Dryad when you need a generalist home that integrates with your journal, and use an institutional repository when your university or funder requires one. ## How to Deposit Data in Dryad, Step by Step The walkthrough below follows one running example, a soil respiration dataset with two CSV tables and one R script. The same steps apply in any field. ### Step 1: Sign in at datadryad.org Open datadryad.org and choose the option to submit or publish data. You can sign in with an ORCID iD, institutional single sign-on, or a Dryad account. ORCID is the strongest choice because it links the deposit to your researcher record without extra steps. ORCID iDs use the pattern 0000-0002-1825-0097, sixteen digits in four groups separated by hyphens. ### Step 2: Start the submission through the right door ![Flowchart showing a direct Dryad submission and a journal-integrated submission both leading to the same prefilled form and curation queue.](/article-images/illustrations/how-to-deposit-data-in-dryad-submission-doors.webp) Two entry points lead to the same form. A direct submission starts on the Dryad site. A journal-integrated submission starts inside the manuscript system of a partner journal, which passes article details such as the title, authors, and DOI to Dryad so fields arrive prefilled and the dataset is linked to the paper automatically. If your target journal offers integration, take that route. It removes retyping and prevents broken links between paper and data. ### Step 3: Complete the metadata fields Metadata, the structured description of a dataset, is what makes it findable and understandable. The form asks for a consistent set of fields, and each one has a job: - Title. State the topic, the data type, and the scope, for example "Soil respiration responses to spring thaw in northern hardwood forests, 2024-2026". A title like "My field data" hides the dataset from search. - Abstract. Two to five sentences covering what was measured, where, when, how many samples, and the general methods. - Authors. Add every contributor with an ORCID iD and affiliation, in the same order as the manuscript. - Keywords. Five to ten terms a stranger in your field would type into a search box. - Related works. The manuscript DOI, and any prior datasets this one builds on. - Funding. Funder names and award numbers, which funders match against their own records. - Methods. How the data were collected and processed, written for someone who has never seen your lab. - Coverage. Collection dates and geographic coverage when the form offers them. A workable abstract for the running example: ``` The dataset contains 214,272 hourly measurements of soil CO2 efflux from 12 forest plots in New Hampshire, USA, collected between January 2024 and May 2026 with closed-dynamic chambers. Files include the measurement table, plot-level site metadata, and R code reproducing the daily flux sums reported in the related manuscript. ``` Provenance, the record of where data came from and what happened to them, raises the interpretability of scientific results and enables reliable collaboration and data sharing [11]. The methods field and the per-file descriptions are where that record lives, so write them for a reader with no access to your notebook. ### Step 4: Upload the files and describe each one Before uploading, confirm the folder contains exactly what you intend to share: ```bash find . -type f -not -name "checksums.md5" | sort ``` Expected output: ``` ./README.md ./analysis_scripts.R ./site_metadata.csv ./soil_respiration_measurements.csv ``` Drag the files into the upload area or browse for them. Upload each file separately where practical, because individual files can carry individual descriptions that search engines and reusers read. A single zip is acceptable for a large directory tree, but say in the description exactly what it contains. After upload, Dryad lists the files with sizes. Spot check integrity by downloading one file, rerunning md5sum on it, and comparing the value against checksums.md5. ### Step 5: Set the license and embargo Dryad publishes datasets under CC0, a public domain dedication that waives copyright restrictions so anyone may reuse the data without asking permission. If the related manuscript is still in press, set an embargo, a hold that keeps the files private until a release date you choose. Match that date to the journal's expected publication date so the data appear when readers can act on them. ### Step 6: Submit for curation Review the preview page, confirm every field, and submit. The submission enters a curation queue, meaning Dryad staff review it for completeness, file integrity, and the presence of sensitive content. Status changes arrive by email. Curation is not instant, so build it into your manuscript timeline rather than the acceptance week. ### Step 7: Answer curator questions Expect short, specific requests. Typical examples include expanding the methods text, splitting an oversized zip file, adding missing ORCID iDs, or confirming consent language for human subjects data. Reply through the submission interface. Each round is usually a small edit, and prompt replies keep the deposit moving. ### Step 8: Verify the published DOI Once curation clears the submission, Dryad publishes the dataset and issues a DOI under the 10.5061/dryad prefix. Confirm the DOI resolves before you cite it, replacing the example DOI with your own: ```bash curl -sI "https://doi.org/10.5061/dryad.example" | head -n 6 ``` Expected output: ``` HTTP/1.1 302 Found location: https://datadryad.org/dataset/doi:10.5061/dryad.example ``` A 301 or 302 status with a location pointing back to datadryad.org confirms the DOI is live. Paste the DOI into your data availability statement and add the dataset to your ORCID record. ### Common Errors During Submission and Their Fixes 1. "Abstract is required." The abstract field was left empty. Write two to five sentences covering what was measured, where, when, and how. 2. "Please provide a valid ORCID iD." The identifier is missing hyphens or contains a typo in its check digit. Copy it directly from your orcid.org profile rather than typing from memory. 3. "Your session has expired. Please log in again." The form times out after long idle periods. Draft the title, abstract, and methods in a text editor first, then paste them in. 4. An upload stalls or fails on a large file. Oversized single files are the usual cause. Compress the directory into a zip, split the dataset into logical parts, and retry on a stable connection. 5. A related-work DOI is rejected. The string is malformed, often missing the https://doi.org/ prefix. Enter the DOI exactly as the publisher displays it. 6. The embargo date arrives while the article is still in press. Contact Dryad support through the help form to adjust the release date so the data do not surface before the paper. ## Tips for Depositing Data in Dryad - Generate metadata from the files themselves. Work on spectroscopic data publication showed that producing both human-readable and machine-readable supporting information directly from primary instrument files, together with the metadata, cuts the extra effort that normally blocks data publication [3]. Your export scripts can emit the CSV files and the variable list in one pass. - Run quality control before you upload. In metabolomics, rigorous QA/QC practices that minimize batch effects are treated as a prerequisite for depositing data in public repositories, and the same discipline strengthens any dataset [4]. - Work from a FAIRification checklist. Life cycle assessment researchers found that adherence to FAIR principles in practice is rare without a deliberate workflow, and proposed a FAIRification workflow as the remedy [9]. Walk your dataset through findability, accessibility, interoperability, and reusability one item at a time before submitting. - Submit alongside the manuscript, not after acceptance. Parallel submission keeps curation off the critical path. - Ask your library for help. Supporting researchers in data management planning ranks among the top strategies for improving data sharing practice [10], and many university libraries offer deposit assistance at no charge. ## Common Mistakes and Limitations Avoid these errors, which curators see repeatedly: 1. Depositing summary tables only. Without raw measurements or the code that produced the summaries, a reuser cannot check or extend the work. 2. Vague titles and abstracts. Search depends on the words you type, and "project data" matches nothing. 3. Missing methods. The FAIR principles set the destination, but reuse fails when the implementation, the actual description of how data were produced, is left out [1]. 4. No README. File names such as data_final_v3.xlsx force every reuser to guess. 5. Uploading sensitive data without a risk assessment. De-identification must be systematic rather than a quick find-and-replace, and re-identification risk deserves explicit analysis before any public release [5]. 6. Ignoring the embargo after journal delays. If publication slips past your release date, contact Dryad support to move it. 7. Treating the deposit as a full software archive. A dataset plus a script with a recorded version is a strong start, but uncontrolled software changes are a documented cause of workflow decay, so record versions rather than assuming the environment survives. 8. Choosing Dryad when a domain repository fits better. Structured data types with community standards belong in a domain repository that can validate them natively [8]. Limitations to keep in view: - Curation takes days, not minutes, and timing varies with submission volume. - CC0 means anyone may reuse the data, including for commercial purposes. Make the deposit decision with that in mind. - Dryad is not the right home for every dataset. Continuously updated data streams, extremely large files, and data requiring controlled access usually need a specialist service. - Policies differ. Your funder, journal, and institution may each name a different required repository, so confirm the requirement before you deposit [10]. - Ethics requirements are case specific. Your review board decides what may be shared publicly, and its answer can differ between projects in the same lab. ## After Your Dataset Is Published Cite the dataset in the manuscript's data availability statement using the DOI. Dryad displays a ready-made citation on the dataset page, which follows this pattern: ``` Soil Respiration Working Group (2026), Soil respiration responses to spring thaw in northern hardwood forests, 2024-2026, Dryad, Dataset, https://doi.org/10.5061/dryad.example ``` Add the dataset to your ORCID record so it appears alongside your papers, and link the dataset page back to the article so readers can move in both directions. If errors surface later, Dryad supports versioned revisions, and the DOI continues to resolve to the newest version. Check the dataset page for download statistics when you prepare annual reports or promotion files. ## Frequently Asked Questions ### How much does it cost to deposit data in Dryad? Deposits are free for authors whose journals partner with Dryad and for researchers at member institutions. Other depositors pay a one-time data publishing charge, and the current amount is shown on the Dryad website before you submit. ### How long does Dryad curation take? Curation time varies with submission volume and the condition of your files. Submit early in the manuscript process and expect at least one round of curator questions before the DOI goes live. ### Can I embargo a dataset until my paper is published? Yes. Dryad lets you set a release date so files stay hidden until your article appears, and the landing page remains visible throughout. ### What license will my dataset carry? Dryad publishes data under CC0, a public domain dedication that lets anyone reuse the data without permission, including for commercial purposes. ### Can I include code and software with my dataset? Yes. Dryad accepts code alongside data, and scripts that document their software versions and parameters make your results far easier to reproduce. ### What file formats should I upload? Dryad accepts a wide range of formats. Open formats such as CSV and TXT keep data readable into the future, so convert from proprietary formats when no precision is lost. ### Can I fix or add files after publication? Yes. Published datasets can be updated through a versioned revision, and the DOI keeps pointing to the newest version. ### How do I cite a Dryad dataset? Cite the DOI shown on the dataset page. The page also provides a formatted citation you can copy straight into your manuscript. ## Sources 1. [Data sharing in modeling and simulation of biomechanical systems in interdisciplinary environments](https://doi.org/10.1002/gamm.202370006) 2. [Data sharing and re-use in the traumatic stress field: An international survey of trauma researchers](https://pubmed.ncbi.nlm.nih.gov/37703089/) 3. [Workflows Allowing Creation of Journal Article Supporting Information and Findable, Accessible, Interoperable, and Reusable (FAIR)-Enabled Publication of Spectroscopic Data](https://pubmed.ncbi.nlm.nih.gov/31459544/) 4. [Addressing Pitfalls of Metabolomics for Toxicology: A Call for Standardization, Reproducibility and Data Sharing.](https://pubmed.ncbi.nlm.nih.gov/40619682/) 5. [Sharing traumatic stress research data: assessing and reducing the risk of re-identification](https://pubmed.ncbi.nlm.nih.gov/40387730/) 6. [Responsible Sharing of Qualitative Research Data: Insights From a Pioneering Project in the United States](https://pubmed.ncbi.nlm.nih.gov/41179580/) 7. [Impact of HIPAA’s Minimum Necessary Standard on Genomic Data Sharing](https://pubmed.ncbi.nlm.nih.gov/28914268/) 8. [PKRxiv: A Best Practice Model for Advancing Pharmacoequity Through Open Pharmacokinetic Data Sharing](https://pubmed.ncbi.nlm.nih.gov/41620928/) 9. [Can LCA be FAIR? Assessing the status quo and opportunities for FAIR data sharing](https://doi.org/10.1007/s11367-024-02280-3) 10. [Enhancing data sharing through digitization and data management practices](https://doi.org/10.18488/104.v11i1.4952) 11. [Provenance Information for Biomedical Data and Workflows: Scoping Review](https://pubmed.ncbi.nlm.nih.gov/39178413/)