RO-Crate

RO-Crate (Research Object Crate) is a lightweight community standard for packaging a set of files together with one metadata file, ro-crate-metadata.json, that describes what each file is, who or what produced it, and how the files relate to each other. Think of it as a folder that documents itself: open it without any prior context, and the metadata file tells you — and any RO-Crate-aware tool — what you’re looking at.

Why not just a folder with a README

A README is prose, meant for a human, saying whatever its author remembered to write. RO-Crate’s metadata file is JSON-LD built on schema.org vocabulary, so a machine can answer the same questions a human would ask: which file is the actual workflow, which are inputs, who authored this, what license does it carry, what tool (and version) produced that particular output file. That’s what lets a repository, a search index, or another program work with the crate’s contents automatically, not only a person reading carefully.

What’s inside

Every RO-Crate has, at minimum:

• A root Dataset entity representing the crate itself, with hasPart links to every file and directory it describes.
• Contextual entities for anything worth naming — a Person, an Organization, a SoftwareApplication — that other entities can reference.

Profiles: RO-Crate is generic, on purpose

Bare RO-Crate says nothing about what kind of thing is being packaged — that’s the job of a profile layered on top, which adds required structure and vocabulary for a specific use case:

• The Workflow RO-Crate profile packages a workflow and its documentation, so the crate itself is a citable, findable unit.
• The Workflow Run RO-Crate family (Process Run, Workflow Run, Provenance Run) packages not the workflow definition but a record of running it: the artifacts used and produced, the commands that ran, the engine that orchestrated them. s4n execute run --rocrate produces a Provenance Run Crate; see Provenance & Publishing for exactly what that contains and how it relates to PROV’s concepts.

How SciWIn uses it

The rocrate crate (ro-crate-lib) reads, writes, and validates RO-Crates on SciWIn’s behalf — it’s what s4n execute run --rocrate calls into. See Architecture for where it sits relative to the rest of the codebase.

Why a researcher should care

An RO-Crate is what turns “I ran this once, here’s a folder of outputs” into a self-describing package: something a repository like WorkflowHub can accept as a citable, FAIR-registered unit, a colleague can download and understand without asking you, and a script (or s4n itself) can inspect years later, without you there to explain what’s in it.