search

ohbm / osr2020

Website for the Open Science Room at the OHBM 2020 meeting
https://ohbm.github.io/osr2020
Other
18 stars 6 forks source link

Past, Present and Future of Open Science (Lightning talk): NeuroLibre : A cloud-based and curated repository for Jupyter Notebooks in neuroscience #15

Open jsheunis opened 4 years ago

jsheunis commented 4 years ago

NeuroLibre : A cloud-based and curated repository for Jupyter Notebooks in neuroscience

By Loïc Tetrel (1), Mathieu Boudreau (2,3), Elizabeth DuPre (4), Agah Karakuzu (2), Félix-Antoine Fortin (5), Jean-Baptiste Poline (4), Samir Das (4), Pierre Bellec (6), Nikola Stikov (2,3), Centre de recherche de l’institut de gériatrie de Montréal, SIMEXP, Montreal, QC, Canada NeuroPoly Lab, Institute of Biomedical Engineering, Polytechnique Montreal, Montreal, QC, Canada, Montreal Heart Institute, Montreal, QC, Canada McGill, Montreal Neurological Institute-Hospital, Montreal, QC, Canada Calcul Québec University of Montreal, Psychology department, Montreal, QC, Canada

Abstract

Fueled by concerns on the reproducibility of science findings, open science practices are gaining wide adoption. Jupyter Notebooks [1] are a popular approach, embedding code and figures in a single document, and offering a standardized interface to reproduce figures from data. However, the neuroscience community still lacks a curated platform to publish such notebooks. We are announcing NeuroLibre, a new curated repository for Jupyter Notebooks, that offers technical review, free long term preservation along with reactive computational resources.

NeuroLibre is a flagship initiative from the Canadian Open Neuroscience Platform (CONP). Anyone with basic development skills can submit a work to NeuroLibre using Jupyter Notebooks. These works are screened for one of two streams: (i) companion to a published (or preprint) article or (ii) training material for an educational workshop. To assess the quality of the work, the notebooks follow a technical review using the GitHub platform, where the published repository will be hosted. Finally, the user is provided a web link listed on the NeuroLibre website, so they can share their work with the community(Fig. 1).

The NeuroLibre infrastructure was most notably built on BinderHub technology [2]. Other dependencies are open technologies that are well-established in the private and academic community: Docker images for the environment reproducibility, Kubernetes for user resource management, Jupyter Notebooks as a development environment (Fig. 2), Jupyter Book front-end to improve user interactivity on the website, and GitHub for the collaborative coding platform and peer reviewing. The computing resources are offered and managed by Compute Canada (https://www.computecanada.ca/), as part of a resource allocation to CONP.

Useful Links

https://www.neurolibre.com/ https://binder.conp.cloud/

Tagging @ltetrel

kimsin98 commented 4 years ago

This reminds me of Distill in machine learning. Could be the first great step toward tackling the research debt in computational neuroscience!

ltetrel commented 4 years ago

@AKSoo Distill was indeed a model for us :)

ltetrel commented 4 years ago

Here is the link to the presentation: https://neurolibre.github.io/neurolibre-presentation/#/