Back to the projects list
Storing Analysis Results and Provenance in NWB
Session Chairs: Ryan Ly, Andrew Tritt
Participants
- Ryan Ly (LBNL)
- Andrew Tritt (LBNL)
Objectives
- Objective A. Discuss usage of NWB for storing intermediate and final analysis results.
- Objective B. Discuss linkage between NWB files on DANDI and code / DataJoint database that generated with the dandiset.
Proposed discussion topics
Participants should add topics of possible interest for discussion here
- What are shortcomings of existing data types? Should NWB support the storage of arbitrarily nested dicts, Python pickles, and other data types within the scratch space and elsewhere?
- How can users store summary statistics, like mean of a value, or the results of a session-level t-test or permutation test, or experiment-level summary statistics? Should they store these data within the NWB file?
- Can DANDI store files that are not NWB but NWB-adjacent like a DataJoint MySQL dump that was used to generate the NWB file? How can users associate a dandiset with a particular release of a codebase used to generate the NWB file?
Approach and Plan
Progress and Next Steps
Materials
Background and References