Jochen Weber (Zuckerman Research Computing)
This project is meant to demonstrate to the institute’s faculty and their labs, associated with one of the two U19 projects (Costa and Miller), that data routinely used in data processing pipelines in the labs can be readily converted into and store as NWB:N files.
Identify meta data mapping. Investigate each of the provided files/datasets w.r.t. what meta data is provided, and create a brief text file for each.
Create initial pass of (stub) code. Read in the data (either in Python or MATLAB, depending on the input format!), and using pynwb or matnwb respectively convert into NWB:N format, while preserving as much of the meta data from (1) as possible.
Test formats. Read in the written out NWB:N objects in both Python and MATLAB to confirm that data can be read back and interchanged between environments (taking care of indexing differences, in terms of dimension order and 0- vs. 1- based indexing!).
Initial pass of loading each of the input formats into memory/inspecting existing fields, and writing down what meta data exists; this is mainly an information gathering mission–some formats will probably lack a lot of additional information to fully appreciate the meaning of fields, just the situation that NWB:N sets out to remedy!
Using existing format conversion projects (from github repositories or the pyNWB tutorial projects page), adapt those code files for use with the datasets provided by Zuckerman researchers for this hackathon.
Ensure that the files (objects) can be read back into Python/MATLAB, and compare their contents with the objects prior to writing to disk (ensuring stability), and then perform cross-platform comparisons.
The following data objects were provided