Error in nwbRead - Undefined variable "types" or class "types.hdmf_experimental.DynamicTableRegion".

[ross-folkard]

Hi there,

I generated a nwb file using the Spike Interface pipeline with Jupyter notebook, where it can be successfully viewed with nwbWidgets. However, when I try to load the same file in to Matlab, it throws an error. This seems to only happen for files that contain Trigger data with spike times, as I am successfully able to load a nwb file without the trigger data and just spike times.

Any help on the matter would be greatly appreciated, as I am much more comfortable working in Matlab compared with Python.

Error:

Same File on Jupyter notebook with nwbWidgets:

Same file minus Trigger Data:

[bendichter]

@rf13734 could you share the file?

[ross-folkard]

@bendichter Sure, it's quite large (1.7gb), have you a preferred email/google drive?

[rly]

@rf13734 Could you also share the Jupyter notebook that you used and which version of PyNWB you used? Thanks.

[ross-folkard]

Hi @rly ,

I had no issues creating a nwb file with just spikes form the 'simple_bin' SI pipeline, but ran into issues with the ced_nwb_conversion_simple pipeline:

I'm using pynwb 2.5.1

I've shared the relevant files with @bendichter, maybe he can add you to the shared folder.

[bendichter]

I have shared the data

[rly]

Hi @rf13734 ,

It looks like you are using PyNWB 2.5.1, which is a broken and removed version of PyNWB (it was released with the wrong version tag). This version should not be downloadable from PyPI. How did you get it?

Can you please pip uninstall pynwb and then pip install pynwb==2.0.0, try to generate the file again, and let us know if the problem persists? Thanks in advance!

[lvsltz]

It looks like you are using PyNWB 2.5.1, which is a broken and removed version of PyNWB (it was released with the wrong version tag). This version should not be downloadable from PyPI. How did you get it?

Quick comment @rly: I've seen this as well multiple times just by doing pip install pynwb

[rly]

Interesting. That's not great. I might have to delete the release or push 2.6 just so that becomes the largest version.
@lvsltz can you reproduce the behavior?

I've found that pip install -U pynwb with pip 20.3.1 would install the yanked version (incorrectly). See pypa/pip#9290 but this is fixed now and @rf13734 is using pip 21.0.1 (at least currently).

[lvsltz]

@rly I can't reproduce it, but I was using the old pip. Sorry for the diversion on this issue.

[ross-folkard]

Apologies for the very late response, I installed pynwb 2.0

I'm now getting the following assertions error for this file:

Any help would be greatly appreciated!

All the best

Ross

[bendichter]

@rf13734 could you open an issue for this in https://github.com/catalystneuro/mease-lab-to-nwb? It looks like your file that we use to extract intervals has a different number of start and stop times for those intervals, but this is not an issue with MatNWB