Skip to main content
NWB Inspector of Best Practices for NWB Files
Cody Baker, Oliver Ruebel, and Benjamin Dichter
Presenting author:
Cody Baker
Neurodata Without Borders (NWB) is a data standard that packages neurophysiology data with the metadata necessary for reanalysis. The NWB format allows data to be human- and machine-readable, and allows data to be aggregated across many labs. However, helping a community with very diverse data converge on a standard is a challenge with many nuances. Through inspecting NWB files published on the DANDI archive, we have found patterns of mistakes and opportunities for improvement. Inspecting this data by hand has several drawbacks: It is time-intensive, it is variable, and it necessarily occurs after data submission. To help data publishers produce the best data possible, we have developed a collection of ‘Best Practices’ to serve as additional guidelines beyond the base schema, as well as a fully automated tool for detecting differences against these guidelines: the NWB Inspector.

In contrast to the NWB validator, which checks for strict schema adherence, NWB Inspector looks at the values, shapes and layout of data and applies heuristics to ensure compliance with the Best Practices. The checks are divided into 3 categories: critical, best practice violations, and best practice suggestions. It also supports dynamic configuration to support alternative or stricter guidelines, such as those enforced by DANDI publication. Finally, NWB Inspector generates a report that communicates what adjustments would allow the data to best follow NWB Best Practices.