NWB Workshops and Hackathons
Back to the projects listNWB via the web: Linking data and accessing NWB files via the web
Session Chair: Andrew Tritt (LBNL)
The goal of this session is to discuss needs and approaches for linking data via NWB and access of parts of NWB files via the network.
Participants
- Andrew Tritt (LBNL)
- Oliver Ruebel (LBNL)
- Yaroslav Halchenko (Dartmouth College)
- Pam Baker(Allen Insititute for Brain Science)
- Michael Grauer (Kitware, Inc.)
- Satrajit Ghosh (MIT)
- Ariel Rokem (UW)
- Josh Siegel (Allen Insititute for Brain Science)
- …
Objectives
- Identify use-cases and requirements for accessing parts of NWB files via the network (Local, LAN, and WAN)
- Identify possible approaches for enabling linking of data via NWB
Proposed discussion topics
Participants should add topics of possible interest for discussion here
- How can we support foreign fields in NWB (i.e., support linking to data from NWB)?
- How can we enable web-based access to NWB files?
- Do we need alternative storage backends to facilitate web-based data access (e.g., HDF5, KITA, ZARR, …)?
Approach and Plan
Progress and Next Steps
- Two main use cases determined:
- accessing of partial datasets
- chunks of objects or whole objects
- interlinking between related datasets
- e.g. connect separate files for LFP and spiking data
- accessing of partial datasets
- investigate HDF5 VOL for creating VOL that can access data over the web without downloading entire datasets
- https://github.com/pierlauro/H5PyVOL
- https://github.com/pierlauro/H5PyVOL