Warning

NeuroLibre is at an alpha stage of development, and is not currently open for submissions.

Step 1 - πŸ› Prepare submissionΒΆ

Before submitting to Neurolibre there are several prerequisites that you must complete, those include:

       Hosting
       Coding language
       Repository layout
       Computation and data requirements

πŸ›°οΈ HostingΒΆ

Github is our main hosting service to store, read and interact with any public submission. The submission, which is in fact a github repository, will need to be public.

Where our backend supports other git hosting services, our submission workflow doesn’t. Still if you are elligible, you may use any of those.

Warning

We are not supporting private repositories at the moment.

πŸ‘©β€πŸ’» Coding languageΒΆ

Neurolibre is living under the jupyter notebook ecosystem, hence supporting a broad range of different coding languages. If the one you are using is inside the following list, then you should be ready to go. But please be mindful that the administrator and reviewer team mainly focuses on python 3, if you are using another language don’t expect fine support for debugging.

We only accept submissions which are in notebook .ipynb format. Though less recommended, we also support text notebook representation (like markdown) using jupytext.

Check the following minimalistic example of a jupyter notebook.


πŸ—‚ Repository layoutΒΆ

Our infrastructure heavily relies on binderhub for the interactive compute environment, and jupyter book for the publishing pdf artifact. The backend workflow make use of those dependencies to build your submission and make the data accessible.

We recommend you to use our template repository to give you a head-start. If you prefer to build the submission yourself, you will need to check the specific detailed section on the πŸ—‚ Preprint repository structure.

Either way you need to make sure that you have all the following files ready:

       Binder requirement file - The recipee that describes the software dependencies.
       (if applicable) Data requirement file - A file that describes the data.
       Jupyter book configuration and table of content - To configure the jupyter book build/layout.
       Notebook files - Your collection of executable scripts.
       Static summary - Submission summary document and bibliography for metadata collection.

See also

Take a look at our meta anlaysis on myelin submission which is publically available in this github repository.

πŸ’½ Computation and data requirementsΒΆ

NeuroLibre strives to provides the best experience by giving access to lot of ressources, still we have limited hardware. Taking ressources constraints into account, each submission should:

  • not take more than 8 hours to execute uncached (that includes all notebooks and book build)
  • run with 1 or 2 CPU@3GHz
  • allocate strictly less than ~7.5GB of RAM
  • have few dependencies to keep your installation light (idealy less than 1GB) and fast (<10min)
  • take less than 10GB runtime storage (files generated by your scripts)
  • use no more than 5GB of data (inputs, cached content), downloadable from a trusted source

Warning

A trusted web source is a well known dataset collection (like openneuro) or dataset fetching functions from libraries (for ex. nilearn.datasets.fetch*).

For additional information, our test server (to test and build submissions) uses Intel Xeon Gold 6248, while the production server (where the final submission lives) has Intel Xeon E5-2650.