Publishing a Dataset

Publishing makes a dataset publicly discoverable and assigns it a DOI. It is a multi-step workflow that lets you review all content before it goes live.

The workflow is designed to make each release both human-readable and machine-actionable for downstream reuse and AI workflows.

Starting the Workflow

From the dataset overview page, click Publish. This opens the publishing workflow at /app/datasets/[datasetId]/publish.

The workflow has five steps:

1. Dataset Metadata - Review all dataset metadata for completeness and accuracy.
2. Study Metadata - Review clinical trial metadata (if applicable).
3. README - Review and edit the auto-generated README for the dataset.
4. Changelog - Review and edit the version changelog.
5. Finalize - Confirm and publish.

Step 1: Dataset Metadata

All required metadata fields are checked before you can proceed. Incomplete required fields are highlighted. Fix any issues by navigating back to the metadata tabs.

Step 2: Study Metadata

If the dataset includes clinical trial data, this step reviews the study metadata for completeness.

Step 3: README

The README is auto-generated from the dataset metadata. You can edit it in the Markdown editor before publishing. The README is displayed to users who view the published dataset.

Step 4: Changelog

The changelog describes what changed in this version compared to previous versions. A default entry is generated for new datasets. Edit it to accurately describe the dataset contents and any changes.

Step 5: Finalize

This is the confirmation step. Review the summary and click Publish to complete the process.

On publication:

• The dataset status changes from draft to published.
• Files are moved from draft storage to published storage.
• A DOI is assigned and recorded.
• The dataset becomes visible in the public discovery interface.
• A stable public landing page is available for citation and sharing.

Publishing Requirements

Before finalizing, the following must be complete:

• At least one title and one description
• Access rights information (access type and rights statement)
• At least one contributor
• README and changelog content

Recommended before finalization:

• Standards and file format details documented (for example CDS, DICOM, OMOP)
• De-identification approach and reuse constraints captured in metadata
• Related identifiers linking publications, source data, or derived outputs

After Publishing

Published datasets are visible at /datasets/[datasetId] on the public site. The dataset can still be updated by creating a new version.

Published entries are intended to support long-term citation, reuse tracking, and contributor recognition.

Creating a New Version

From the dataset overview, use the New Version action. This creates a new draft dataset with the same canonical ID and an incremented version number. All existing metadata is carried over. You can then modify the content and publish the new version independently.