biocompute-objects / BCO_Documentation

Repository for documentation to support the IEEE 2791-2020 standard. Please see our home page for communications/publications:
http://biocomputeobject.org/
BSD 3-Clause "New" or "Revised" License
16 stars 12 forks source link

separation of input Files and other input parameters clashes with CWL's object model #63

Closed mr-c closed 2 years ago

mr-c commented 5 years ago

In CWL inputs can be of type File or Directory, as well as string, or various types of numbers. Additionally we have complex types (lists, records/maps).

In BCO there is currently a mixed "IO" domain for input/output files and a seperate parametric_domain that has only inputs of type "string", so no room for other types of inputs.

This means that auto-generating a BCO from the execution of a CWL workflow will be difficult, and potentially impossible. It is likely that other workflow models will have a similar problem.

HadleyKing commented 5 years ago

Suggestions on how to address this? Is there an example we can use that does fit the current BCO model?

HadleyKing commented 5 years ago

My thoughts: The binning of these values is for conceptual understanding of how a pipeline was implemented (communication purposes), not explicitly to RUN the pipeline. A CWL workflow would be PART of the BCO and as such would still contain all the info represented by the CWL object in the CWL model.

I would like to try two possible examples if some one can point me to them: 1) CWL object that fits the BCO model very well and 2) CWL object that does not