Open SPMMER opened 1 month ago
@SPMMER Does your file use any NWB extensions? It looks like there is an issue with one of the cached specs. Would you be able to share the file with us to investigate?
Also, what version of nwbinspector are you using? You can run pip show nwbinspector
to get that.
No extensions that I know of. They are made with the nwb 2.4.0 schema using generateCore('2.4.0') command.
The nwb inspector version is 0.5.2.
I shared the file with @mavaylon1 so maybe he can upload that.
@SPMMER I have the USB flash drive with the data so I can take a closer look after SFN.
@SPMMER Could you provide some additional details about how these files were generated?
I took a closer look at the file you shared, and it looks like there is an additional cached spec in the specifications group (/specifications/mss
) that is causing the error you described. The mss.multishapes schema is used in unit tests in matnwb, but it is not clear to me why it is being included in this file.
After removing this cached schema, there are still a couple of separate errors when reading with pynwb for validation (e.g. the length of the row ids in the general/intracellular_ephys/intracellular_recordings/
and general/intracellular_ephys/intracellular_recordings/protocol_type
groups do not match the length of the their corresponding values).
It would be helpful to know how these files were created so we can figure out where to address these issues.
@stephprince Thank you for looking into that further.
It is not clear to me as to why this schema would be included.
The files were generated with a custom Matlab script that filters our experiment data for a certain subset of recordings and then builds the nwb file based on the documentation provided in the tutorial here (https://neurodatawithoutborders.github.io/matnwb/tutorials/html/intro.html). We are converting the data from cfs files (CED Signal) and DAT files (HEKA Patchmaster).
There is an issue with matnwb where all cached namespaces are included in the file, independent of whether types from that namespace are used or not.
That would explain why the mss is included.
Thanks @ehennestad for the info!
If I understand correctly, it looks like most of the errors I ran into when trying to validate the file in pynwb were fixed in later versions of MatNWB:
@SPMMER is it possible for you to regenerate the files with a newer version of MatNWB and the newer tutorials? I hacked together a way to modify the original file you shared, but it would probably be better to use an updated version of the MatNWB package and tutorial.
As a summary, after I modified the file to address the issues listed below, I was able to successfully run the NWBInspector on the file:
mss
namespace is cached but raises an error.id
column of the intracellular_recordings
table is empty but the category tables are not.id
column of the protocol_type category does not match the length of the values.intracellular_recordings.colnames
specifies ["recordings_tag"]
as a column, but the VectorData
object for that column is missing.CurrentClampSeries
(an issue with pynwb that I have raised here NeurodataWithoutBorders/pynwb#1971).Since I believe any further issues will be addressed on the MatNWB side, I propose we transfer this issue to that repo.
What happened?
We created our NWB files with MATNWB 2.4.0 and now we try to validate it with NWBInspector. It trows the following error:
Steps to Reproduce
Traceback
Operating System
Windows
Python Executable
Conda
Python Version
3.11
Package Versions
No response
Code of Conduct