NWB Workshops and Hackathons

Back to the projects list

Review and refine ICEphys Metadata extension

Key Investigators

Oliver Ruebel (LBNL)
Ryan Ly (LBNL)
Thomas Braun (byte physics)
Pam Baker (Allen Insititute for Brain Science)
Liviu S. (Laboratory of Neural Microcircuitry, EPFL / Blue Brain Project, EPFL)
Shreejoy Tripathy (University of Toronto)
Benjamin Dichter (Ben Dichter LLC)
…

Project Description

Review and refine the proposed extension for ICEphys.

Objectives

Evalute suitability of the proposed extension for ICEphys for various use cases
Identify challenges
Refine proposal
Fix known issues

Approach and Plan

Implement fixes for reported issues
Convert ICEphys files from different labs using the extension to evaluate suitability of the extension
Have working group to review and refine the proposal
Make extension pip installable
Add extension to the NDX Catalog

Progress and Next Steps

Released ndx-icephys-meta extensions on PyPi https://pypi.org/project/ndx-icephys-meta/
Submitted ndx-icephys-meta extension to the ndx-catalog
Reviewed extension with developers and users as part of the breakout session and closing discussions https://neurodatawithoutborders.github.io/nwb_hackathons/HCK07_2020_Seattle/projects/breakout_icephys/
Created milestones: 1) Release V.0.2.0 that includes all changes targeted before integration with NWB 1) Future which includes proposals for future changes
Identfied several changes for the extensions that should be implemented before merging the extension with NWB, described below.

Changes identified as part of the hackathon

Add separate stimulus and response tables that the intracellular_recordings table in turn references. See https://github.com/oruebel/ndx-icephys-meta/issues/31
Define more descriptive names for the tables to more clearly communicate the purpose of the different tables and avoid existing terms with ambigious definitions in the community. See https://github.com/oruebel/ndx-icephys-meta/issues/32 Specifically:

1) intracellular_stimulus (new, see #31) 1) intracellular_response (new, see #31) 1) intracellular_recordings (remain as is) 1) sweeps —rename to—> simultaneous_recordings 1) sweep_sequence —rename to—> sequential_recordings 1) runs —rename to —> repetitions 1) conditions —rename to—> experimental_conditions
Add stimulus_type coluns to the sequential_recordings (a.k.a, sweep_sequences table)
Refine add_intracellular_recordings function to make arguments optional and add error checks. See https://github.com/oruebel/ndx-icephys-meta/issues/34
Add documentation for entering time slices for add_intracellular_recordings. See https://github.com/oruebel/ndx-icephys-meta/issues/35
Avoid the use of the ic prefix in the API as it is ambigous with terminology used in the ICEPhys community. Use the prefix icephys instead. See https://github.com/oruebel/ndx-icephys-meta/issues/36

Next steps

The goal is to implement all changes for https://github.com/oruebel/ndx-icephys-meta/milestone/1 prior to integration with NWB
Once implemented and tested the extension should be merged with NWB
The goal is to complete all changes within the next 2 weeks and complete merging of the changes with NWB propoer within the next 4 weeks

Follow-up

As part of the discussions several topics for future follow-up proposals to further refine the management of icephys metadata have been identified. These are currently managed as part of https://github.com/oruebel/ndx-icephys-meta/milestone/2

Add the ability to define custom tables that branch of the existing hierarchy. E.g., see https://github.com/oruebel/ndx-icephys-meta/issues/24 This topic introduces a great deal of flexibility and requires further review of use-cases and impact.
Refactor strategy for storing large collections of ICEphys recordings with the goal to reduce the number of objects required in a file. One main goal here would be to optimize performance and reduce the need for iteration over large numbers of objects. This issue is also related to this https://github.com/oruebel/ndx-icephys-meta/issues/17
Wrap stimulus/response columns to improve rendering and potentially improve performance by delaying resolution of references. See laos in part https://github.com/oruebel/ndx-icephys-meta/issues/23
Add stimulus_types table for describing the stimulus types. See https://github.com/oruebel/ndx-icephys-meta/issues/37

Materials

Proposed extension: https://github.com/oruebel/ndx-icephys-meta
Extension proposal document