open-discussion > Documenting, referencing, registering data and processing pipelines: from blessing to intolerable burden?
Showing 1-1 of 1 posts
Display:
Results per page:
May 24, 2022  09:05 PM | Arnaud Delorme
Documenting, referencing, registering data and processing pipelines: from blessing to intolerable burden?
When publishing a neuroimaging paper, ideally, we should
- Publish the data in a standard format and associated DOI
- Publish the code to process the data above, and that generates figures for our the paper
- Version the code, if possible, with a DOI as well
- Provide a Docker container, so the code is not subject to being deprecated
- Maybe publish a Datalad repository that contains both references to the code and the data
- Document each step so other researchers can reprocess the data

In the future, we might have to
- Use specific ontologies to document data and report results, so meta-analysis can be built automatically
- Use only registered tools that use specific standards (which might hinder scientist creativity?)

How much is too much for reproducibility?
Won't AI help extract relevant information from papers and associated data?
Is this level of detail necessary for science to move forward, or is it an unnecessary burden?

I would be interested in hearing your thoughts,

Arno