Frequently Asked Questions¶
I don’t know Python, can I still use the pipeline?¶
Although we like Python and we recommend gaining some understanding of the foundations, we do think it’s possible to use pipeline as-is with minimal knowledge of python.
For start, you can just change dpath under set path and parameters to point it to your data.
Then you should be able to run all the cells in the pipeline without modifying anything.
However, chances are you will like to tweak some parameters. In order to do this all you need to know is how Python dictionary works. Here is a quick tutorial. The general syntax is:
param_stepX["parameter_name"] = value
where value maybe string, numbers, lists or in some cases another dictionary by itself, depending on the parameter you are changing.
Once you run this line the param_stepX dictionary will be updated and you can proceed to the step that use this dictionary.
The pipeline also has a quick real-world example under loading videos and visualization.
Can I get a table of content for the pipeline?¶
Yes! Jupyter has a toc2 extension that should work well with the notebook. You can install it with:
conda install -c conda-forge jupyter_contrib_nbextensions
Can I have a version of pipeline without all the notes?¶
Yes! all the tutorial cells in the pipeline are tagged with “tutorial”.
We can use nbconvert to obtain a copy of the notebook with all those cells removed.
For example, the following command will give you a clean pipeline:
jupyter nbconvert pipeline.ipynb \
--TagRemovePreprocessor.enabled=True \
--TagRemovePreprocessor.remove_cell_tags="tutorial" \
--to notebook \
--output pipeline_clean.ipynb
Just note that since this is a different copy of the pipeline, there might be extra work if you want to merge you edits on this clean copy with updates in the default pipeline notebook.
How can I monitor the progress of some time consuming steps?¶
Dask provide a dash board which contains realtime information about all ongoing computations and cache. By default you can access it by navigating to http://localhost:8787/status in your browser once you have started the cluster. See Monitoring cluster for more detail.
KilledWorker dammit, what do I do?¶
Short answer: sometimes all you need is to try again.
If that doesn’t work or if that’s getting annoying, then increasing memory_limit of the cluster would most likely fix it.
See Dealing with KilledWorker for more detail.
What do I do with the output of MiniAn?¶
Hopefully the tutorials in the pipeline would give you an idea of how to interprete each variable.
Then if you’re happy to work with xarray, then you will feel right at home once you do open_minian.
See Saving to other formats for details on converting the output to other formats.