How do we assess interoperability

[scman1]

Using the example of the Galaxy RO-Crates, I would like to be able to open the Galaxy RO Crate, extract the data and the workflow and try to test different scenarios:

1. Reproduce the results, outside of Galaxy, for instance using Dagster, AiiDA, PWD, pyiron, Nextflow, Jupyter, Larch, or Demeter.
2. Reuse the data, this is determine if intermediate or final data objects are accessible, identifiable and reusable.
3. Reuse the workflow, this is: extract operations, sequencing and parameters to automatically build a “workflow twin” which can execute the operations of the original workflow and provide equivalent outputs.
4. Data driven processing, read the RO Crate, determine the types of data objects contained and which operations can be performed on those outputs (plot, categorise, further processing). This could enable for instance reading the outputs, so we can for instance, plot output data (create default plot dynamically).

Are these scenarios adequate to show interoperability or do we need something else? Is this the same idea of interoperability the group has?

We will try to work on extending the Galaxy RO crates we use and see how far we can take them for supporting these scenarios. I think that the example from Abigail, using frictionless data, is a likely path to explore how far we can use to explore these scenarios. Alternatively, CODATA has created a tool for documenting data using csv tables called Nectar. This is related to the work we are doing on the (CDIF-4-XAS)[https://zenodo.org/communities/cdif-4-xas/records] project, so we already have some work on how to semantically describe the data, and now we are looking for application cases.

[scman1]

We have two deliverables from the OSCARS sponsored project CDIF-4-XAS which we want to use to explore how they could support interoperability between different workflow platforms