Python API: visualizations

[tomasz (Tomasz Kościółek)]

Hello!

how can I visualize visualization objects in Jupyter? For example, I run
feature_table.visualizers.tabulate_seqs(deblur_sequences.representative_sequences)
and all I get is this:

Results (name = value)
-----------------------------------------------------------------------------------------
visualization = <visualization: Visualization uuid:
9486c314-d14a-4e30-b52c-ca97a68af501>

How can I get the visualization I would normally get from using qiime tools view artifact.qzv? Do I have to save it first and then visualize from a saved qzv file?

thanks,
Tomasz

[thermokarst]

Hi @tomaz, we don't have a good workflow developed yet for viewing visualizations within Jupyter - check out this post for a little detail:

Embedding visualizations in jupyter

Not yet, but that would be really nice. The technical limitation right now is figuring out a way out of the jupyter sandbox as index.html files in visualization usually load assets and other files, so we need a way to “hoist” that directory into something that the jupyter notebook would serve (or we start an ephemeral web-server for reprs).

Maybe @uria can provide some more details about their workaround.

Otherwise, we are stuck working within Jupyter's limitations, so some code like this can get you moving:

viz = feature_table.visualizers.tabulate_seqs(rep_seqs)
viz.visualization.save('foo.qzv')
!qiime tools view foo.qzv

This will open the visualization in a new tab, although YMMV depending on how Jupyter NB is configured.

[uria (Uria Mor)]

Hey there @tomasz ,

regarding

@thermokarst

Maybe @uria can provide some more details about their workaround.

Though I really couldn't find a good reason to use this workaround (qiime view works just fine for me...) what I suggested in mentioned thread is to:

1. Extract the visualization object as a directory

out_dir = "/tmp/deblur_sqs_vis"
feature_table.visualizers.tabulate_seqs(deblur_sequences\
                .representative_sequences).export_data(out_dir)

Now you should have (I hope) a directory called deblur_sqs_vis under /tmp. This directory also should contain sub-directory named data in which there's an index.html file.
2. Use the HTML object in IPython.core.display in order to view this directory

from IPython.core.display import  HTML
HTML('<iframe height = 1000, width = 1000, src = "{}/data/index.html"> </iframe>'\
    .format(out_dir))

Hope it'll work for you.