[Bug]: Error when validating (dandi validate) file created with matnwb

[dustinf1989]

What happened?

I created .nwb files using MATLAB but get this error when I try to validate:
D:\emotional_memory_neuronal_analyses\emotional_memory_neuronal_dandi\000887>dandi validate

[pywnb.GENERIC] D:\emotional_memory_neuronal_analyses\emotional_memory_neuronal_dandi\000887\sub-001\sub-001_ses-ENC_behavior+ecephys.nwb — DtypeError.init: incorrect type for 'expected' (got 'RefSpec', expected 'dtype, type, str or list')

What is field the incorrect type?

I ran nwbinspector on the same file and copied the error message to this txt file:
error_message.txt

Can you help me figure this out please?

Thanks!

Steps to Reproduce

I ran nwbinspector on the same file and copied the error message to this txt file:
[error_message.txt](https://github.com/NeurodataWithoutBorders/matnwb/files/14192478/error_message.txt)

Error Message

No response

Operating System

Windows

Matlab Version

R2022b

Code of Conduct

• I agree to follow this project's Code of Conduct
• Have you ensured this bug was not already reported?

[lawrence-mbf]

Hi @dustinf1989 , Is it possible for you to provide the source file so we can take a look? If not, we would want a minimal example that recreates this issue with the inspector.

[dustinf1989]

Hi @lawrence-mbf thank you for the quick response. Sure, I can share the file. You should be able to download it here: https://drive.google.com/file/d/1qkzeef1o8dnvgSvsIQZvwRrijuSEBd5w/view?usp=sharing

[lawrence-mbf]

Hi @dustinf1989 ,

It looks like the error has something to do with electrode_group in the file; But as far as I can tell, there does not appear to be anything wrong with the type.

@rly @bendichter any thoughts about this?

[dustinf1989]

Hi @lawrence-mbf

Thanks for checking this out. I edited the size of the 'electrode_group' variable so it is the same size as the 'spike_times' variable but I still get the same error.

Could it be related to the dimensions of waveforms'? The dimensions are [143756×64 single]

The new file is uploaded here: https://drive.google.com/file/d/1n60v9kG25oxMP3Waz1AeTmk3o9UUwhiM/view?usp=drive_link

[ehennestad]

I was browsing around, could it be related to this issue in PyNWB?
NeurodataWithoutBorders/pynwb#1692

[dustinf1989]

I can use this Python code to load the file and create a few basic plots, so the file seems to work, but I still get the same error when trying to validate. Any ideas on how to get around this error?

load_nwb_files.txt

[ehennestad]

Open question: is this due to an issue in matnwb or in pynwb / nwb validator?

@stephprince, would you know?

[stephprince]

Looking at the file and where it fails during validation, it looks like the error is coming specifically from units/electrode_group. The error message is confusing in this situation because there is a separate bug in HDMF in which the DtypeError message fails for reference data types. I believe the more accurate error message should be:

Units/electrode_group (units/electrode_group): incorrect type - expected 'object reference', got 'float64'

So to answer your question @ehennestad , there is a separate bug in HDMF in displaying the error message, but I believe the file also fails validation because the units/electrode_group is the incorrect type. The schema defines units/electrode_group as a vector of references similar to the group column in the electrodes table, but in the file that was shared it is a vector of float values.

[ehennestad]

Thanks for looking into this @stephprince !

Todo:
Is there a bug in the matnwb api leading to the units/electrode of these files being of wrong type?