Packaging a IEEE 2791 described BioCompute Object as RO-Crate, using a workflow run of nf-core/chipseq as example. This composition is colloqually called a BCO RO-Crate.
This example attempts to show the separation of concerns for workflow data packaging when combining these standards:
- BioComputeObject formalized in standard IEEE 2791
- RO-Crate
- BagIt (RFC8493)
For more information, see https://biocompute-objects.github.io/bco-ro-crate/ where this example is explained in detail.
The license of this example is Creative Commons Zero v1.0 Universal which permits free reuse without attribution or license requirements.
Note that some of the data in data/results are derived from the reference data, licensed under the MIT license
The workflow nf-core/chipseq is licensed under the MIT license
This package is therefore a good example of mixed licensing.
This package contains results of an exemplar run of the Nextflow workflow nf-core/chipseq version 1.2.1. This workflow was chosen because it is well documentedf for domain scientists, uses public reference data, and is reproducible to run using BioConda or Docker containers. The Nextflow workflow engine is also open source and runs on most platforms.
Attribution-wise this workflow is open-source, has multiple community authors and a Zenodo DOI for citations https://doi.org/10.5281/zenodo.3966161
The workflow repository in GitHub is bundled with test data which at nf-core is used for automated testing. This exemplar therefore runs the workflow with those inputs by using -profile test.
Nextflow allows execution of the nf-core workflows by their short name, and they are downloaded automatically. Our run.sh therefore refers to the workflow by reference nf-core/chipseq:
cd data
nextflow run nf-core/chipseq -revision 1.2.1 -profile test,conda | tee nextflow.logNote Here we use -revision 1.2.1 to fix the workflow version. Similar measures should be used when executing workflows that are not contained in the BCO folders.
This workflow retrives reference data sets from https://github.com/nf-core/test-datasets/tree/atacseq which again come from ilummina's iGenomes. This is a good example of reuse of reference data sets that have multiple levels of provenance and attributions.
The folder data/results contain the outputs as produced by the Nextflow workflow.
Note that some of these files are large and are stored using Git LFS - if you need all the files, follow the git-lfs install instructions, then try git lfs fetch in a Git checkout of this repository
data/chipseq_20200910.json is a JSON file conforming to IEEE 2791's JSON Schemas https://w3id.org/ieee/ieee-2791-schema/
As is common practice, the BCO file is named chipseq_20200910.json to reflect the workflow name and this particular run. This name is not required to be particularly unique but should be meaningful to users. The RO-Crate metadata file will help programmatic access in locating the correct BCO file. It is RECOMMENDED that a BCO RO-Crate has only a single BCO file.
Here the role of the BCO is to explain the workflow, particularly for a domain expert not familiar with Nextflow.
The JSON can be validated with tools like https://www.jsonschemavalidator.net/ using the JSON Schema:
{ "$ref": "https://w3id.org/ieee/ieee-2791-schema/2791object.json"}The above schema will recursively include the full IEEE 2791 schemas.
data/ro-crate-metadata.json is a JSON-LD file conforming to RO-Crate Metadata specification 1.1-DRAFT
Here the role of the RO-Crate is to type, relate and describe the individual files in this package. This includes external references, licenses and attributions to authors. This may seem like overlap from the BCO, but as it is here per file it can distinguish more complicated situations such as the reference datasets.
bagit.txt, bag-info.txt, manifest-sha512.txt and tagmanifest-sha512.txt conform to BagIt (RFC8493).
The role of the BagIt is mainly ensure all files are present and not modified or corrupted in transfer.