dandi / example-notebooks Goto Github PK

Discovered while #10

---------------------------------------------------------------------------
ModuleNotFoundError                       Traceback (most recent call last)
/tmp/ipykernel_18732/3880075473.py in <module>
      1 import pynwb
      2 from pynwb import NWBHDF5IO
----> 3 from nwbwidgets import nwb2widget
      4 import requests
      5 

/opt/conda/lib/python3.9/site-packages/nwbwidgets/__init__.py in <module>
      1 import plotly.io as pio
      2 
----> 3 from .view import nwb2widget, default_neurodata_vis_spec
      4 
      5 # from .ephys_viz_interface import ephys_viz_neurodata_vis_spec

/opt/conda/lib/python3.9/site-packages/nwbwidgets/view.py in <module>
      7 import zarr
      8 from ipywidgets import widgets
----> 9 from ndx_icephys_meta.icephys import SweepSequences
     10 from ndx_spectrum import Spectrum
     11 

/opt/conda/lib/python3.9/site-packages/ndx_icephys_meta/__init__.py in <module>
     30 
     31 # Import the files
---> 32 from .icephys import ICEphysFile,  IntracellularRecordings, Sweeps, SweepSequences, Runs, Conditions # noqa E402, F401
     33 from . import io as __io  # noqa E402, F401

/opt/conda/lib/python3.9/site-packages/ndx_icephys_meta/icephys.py in <module>
      3 from pynwb.icephys import IntracellularElectrode
      4 from pynwb.base import TimeSeries
----> 5 from hdmf.common import DynamicTable, DynamicTableRegion
      6 from hdmf.utils import docval, popargs, getargs, call_docval_func, get_docval, fmt_docval_args
      7 import warnings

ModuleNotFoundError: No module named 'hdmf.common'

@bendichter - any immediate idea on how to mitigate?

We have requirements.txt for various notebooks.

• dependabot sends multitude PRs: https://github.com/dandi/example-notebooks/pulls?q=is%3Apr+dependabot+
• @bendichter merged some, some remain open
• there is no guarantee that notebooks would work with more recent versions of dependencies
• there is no guarantee that results would be reproducible with newer versions
• there is no guarantee that new upgraded versions would be coinstallable and compatible with older versions of the others

with that in mind, I think we should just disable dependabot for this repo and close all those PRs unmerged

If there is no feedback/counter arguments, I will do that in a week

https://github.com/datajoint-company/DataJoint-NWB-showcase

with #9 as the motivator, I think it would be important for us to ensure that notebooks we provide "work" on the hub.

• we need a way to test notebooks on an existing instance of the hub, so we could test new/updated/beta version of the hub
  • testing could be just smoke run of the notebook to start with
  • I found this article which introduces nbmake
    • running pytest --nbmake dandi-notebooks/ now, will report below the "results"
  • to complete in reasonable time we might want to allow to run them in parallel (might not do parallel on small instance)
• we should be able to trigger "testing" all notebooks for a specific type of the hub instance (is that "scriptable" @satra to start a specific instance and interact with that environment)?
• similarly to https://github.com/dandi/dandi-api-webshots/ and https://github.com/datalad/datalad-extensions/ we could have a repository (e.g. example-notebooks-dashboard) which would contain the README.md which in a table would list which notebooks are ok or failing on which type of hub instance.

jovyan@jupyter-yarikoptic:~$ pytest  --nbmake dandi-notebooks/
===================================== test session starts =====================================
platform linux -- Python 3.9.6, pytest-6.2.5, py-1.10.0, pluggy-1.0.0
rootdir: /home/jovyan
plugins: nbmake-0.10, anyio-3.3.0
collected 9 items                                                                             

dandi-notebooks/000004_demo_analysis.ipynb F                                            [ 11%]
dandi-notebooks/NWBWidget-demo.ipynb F                                                  [ 22%]
dandi-notebooks/Untitled.ipynb .                                                        [ 33%]
dandi-notebooks/000055/BruntonLab/peterson21/Fig_coarse_labels.ipynb F                  [ 44%]
dandi-notebooks/000055/BruntonLab/peterson21/Fig_pow_spectra.ipynb F                    [ 55%]
dandi-notebooks/000055/BruntonLab/peterson21/Table_coarse_labels.ipynb F                [ 66%]
dandi-notebooks/000055/BruntonLab/peterson21/Table_part_characteristics.ipynb F         [ 77%]
dandi-notebooks/000055/BruntonLab/peterson21/dashboard.ipynb .                          [ 88%]
dandi-notebooks/cosyne_2020_tutorial/NWB_tutorial_2019.ipynb .                          [100%]

edit 1:

• as @bendichter pointed out - some notebooks need environment with nwb extension(s) installed. So we can't just run them all in the same environment. We might need to annotate and then group and run in specific environments
• some notebooks have their own pip install -r requirements.txt which alters the environment thus making order of execution matter etc. Not yet sure what to do about that, besides running each test in a new "clone" of the test environment

@bendichter - would it be possible to create a specific notebook example for DANDI? that includes a cell on top that downloads/instructs to download data using datalad? once we move all the dandisets to AWS, this should become much faster.

just like we pull in the nwb repo, we can pull this repo in whenever a user logs in to the dandi hub, so any changes to this repo will get picked up.

you could in fact create the examples in the hub, and then send a PR to this repo with those notebooks.

I get a validation error for the nwb file I created as part of following through the User Training video (using DANDI User Guide Part 1) and could not upload it onto the staging dandiset I created, the command is dandi upload -i dandi-staging and this is the error message sub-001/sub-001_ecephys.nwb 225.0 kB 1 skipped failed validation

Has there been a recent change that prevents this file from passing validation now? Or is there something else I should be doing?
Thank you!

Notebooks should be usable on any "compatible" system: user laptops, etc.

• Decide on how to instrument a download location for the data in the notebooks .
  • Inspired by mkdir ../data in https://github.com/dandi/example-notebooks/pull/8/files#diff-69cfed3e5f1ce14dd4a958d296e0f62cfee6fb1b4208c2acc6daf1901b3130e4R146
  • Most likely should be a data/ subfolder within this repository as a placeholder (or potentially some data under git-annex) for the location of downloads.
  • Some notebooks suggest/use ~/Downloads as the default one where the browser would download data
• Strive to avoid modification of the environment from within the notebook
  • Some notebooks have !pip install -r ... commands, which alters current environment
  • Although "user friendly" somewhat -- complicates testing (#10) and could lead to surprizes among users since behavior could depend on the sequence in which they opened notebooks

Considering to use 000055 for the workshop demos, trying to run the provided from this repo notebooks. This issue will collect gotchas I encounter, and will be populate with items, checking them out if/when resolved. Notebooks are under 000055/BruntonLab/peterson21/

• Fig_coarse_labels.ipynb first cell %pip install -r requirements.txt leads to ERROR: Failed building wheel for scipy and thus failure

Hi, I just came across this repo. I haven't looked into DANDI in detail but it seems like there's a nice API which can provide lazy loading and random access to files (assuming those file types support lazy loading), correct me if I'm wrong?

Anyways I'm writing a new library to perform very fast visualizations in jupyter notebooks, it can leverage Vulkan/WGPU using an expressive API. I'm curious to see how it would perform with DANDI.

https://github.com/kushalkolar/fastplotlib

@bendichter - I'm thinking about updating the notebooks from this repo, but I wonder if your yesterday notebooks were not related to this repo