Skip to main content
NWBWidgets: Interactive Visualization for Neurodata Without Borders Files
Benjamin Dichter, Matt McCormick, Luiz Tauffer, Jeremy Magland, Lydia Ng, Oliver Ruebel, Michael Grauer
Presenting author:
Benjamin Dichter
With more and more cellular neurophysiology data being released in Neurodata Without Borders (NWB) format, the field is in need of a tool to easily explore the data within these files. To address this need, we have created the package NWBWidgets, which uses the Jupyter widget library to build interactive visualization of NWB data types.

NWBWidgets inspects the NWB file, and provides a hierarchical view that allows the user to navigate through all of the data objects in that file. Each object is assigned a specific visualization function based on its neurodata type (LFP, spikes, position trace, etc.), which allows NWBWidgets to provide specific views that make sense for each type of data.

NWB files can be 10s of GB, so performance and efficiency are key. All of the visualizations are rendered in a lazy fashion. None of the visualizations functions are called until that visualization is brought into view. In addition, many of the visualizations are windowed. For instance, voltage traces are windowed by time and by channel. Only the data that is shown to the user is loaded, and data outside the window is not.

We are engaging with users to ensure that NWBWidgets meets the needs of the scientific community. We are integrating this technology with the DANDI data archive to offer visualizations to files that users can run without downloading the data. We are also working with the Allen Institute to build interactive visualizations specialized for their Neuropixel data release.