[IPython-dev] Announcement: Galaxy IPython Integration - interactive programming in Galaxy
Björn Grüning
bjoern.gruening at gmail.com
Thu Aug 7 02:56:54 EDT 2014
Hi Fernando,
>> Disclaimer: Even though the Ipython notebooks can be stored and reused,
>> this plugin will break the Galaxy philosophy of reproducibility, I feel
>> personally bad about that,
>>
>
> Could you unpack that statement a little bit? We think of IPython as a tool
> that can *help* reproducible research, so I'd really like to understand
> better what you mean by the above...
In Galaxy land we are kind of extreme about reproducibility, which is
good. With the IPython container we loose a little bit of this, but we
gain a lot of flexibility for developers.
For example in Galaxy every dataset and tool is versioned. If you
process your data with matplotlib-1.2.1 we are logging this and you can
rerun your analysis with the same settings, versions and so on ...
If we are now using the IPython extension, we don't have such a fine
grained support for packages. We could version the containers and
restrict the container to install packages ... but I feel like the
status quo is acceptable, since I assume our users are mostly
bioinformaticians.
But ... we still maintain the ipynb file. The user can, at any time,
save the file to the history, we will attach the ipynb as metadata to
every dataset that is exported. So that you can use the Galaxy reload
button and you will be back in IPython with all your code.
I think this is already a high level of reproducibility and I'm whining
on a high level :)
Cheers,
Bjoern
> Cheers
>
> f
>
>
>
> _______________________________________________
> IPython-dev mailing list
> IPython-dev at scipy.org
> http://mail.scipy.org/mailman/listinfo/ipython-dev
>
More information about the IPython-dev
mailing list