Advancements in the CBRAIN Platform through Integration of Community-Based Tools and Standards

Autor: Shawn T. Brown, Pierre Rioux, Natacha Beck, Najmeh Khalili-Mahani, Candice Czech, Armin Taheri, Gregory Kiar, Xavier Lecours-Boucher, Carolina Makowski, Darcy Quesnel, Jean-Baptiste Poline, Reza Adalat, Tristan Glatard, Samir Das, Alan C. Evans
Rok vydání: 2019
Předmět:
DOI: 10.5281/zenodo.3247776
Popis: Introduction Since 2009, CBRAIN, a collaborative, web-based research platform, has served a broad international community of researchers in performing large-scale data and computational neuroscience (Sherif, 2014). With over 800 users at 193 sites in 32 different countries and hosting over 60 software pipelines, CBRAIN has provided over 35 million CPU hours on computing resources around the world, including Stampede2 at TACC, the world's largest supercomputer dedicated to academic research. CBRAIN is a central component in the Healthy Brains for Healthy Lives infrastructure (HBHL, 2018) and the Canadian Open Neuroscience Platform (CONP, 2018), which require new features connecting existing platforms (e.g. LORIS (Das, 2016), OpenNeuro (Gorgolewski, 2017), and BrainCode (Vaccarino, 2018), utilizing community-based standards (e.g. BIDS (Gorgolewski, 2016), Boutiques (Glatard, 2018), and CARMIN (Glatard, 2015)), and offering new usage modalities and interfaces. Methods CBRAIN's primary purpose is to provide an ecosystem that abstracts away the low-level details of data movement and computational execution on advanced research computing resources. CBRAIN provides an orchestration system consisting of a central control instrument, termed a Portal, which communicates and submit tasks to remote compute servers, called Bourreaux. Portals and Bourreaux access remote data resources through passive DataProviders.The CBRAIN platform provides a unifying service layer for access to remote computing resources around the world (e.g. Compute Canada, XSEDE, and the CCC-Axis). CBRAIN is a Ruby on Rails application, is completely open source (https://github.com/aces/cbrain) and provided as a service free of charge (https://portal.cbrain.mcgill.ca). Requirements from new national and international initiatives lead to developments increasing interoperability, functionality and usability supporting a wider community and integrating with a broader set of community-driven tools and standards. Results New features developed in the CBRAIN platform are: RESTful API: A fully documented and functional CBRAIN RESTful API is now available at https://app.swaggerhub.com/apis/prioux/CBRAIN/5.1.0 and allows projects to utilize CBRAIN as a backend technology. To promote a community standard, CBRAIN will support CARMIN, a common web API for remote pipeline execution, such that any CARMIN compliant tool can use CBRAIN as a backend without rewriting their package. Datalad and S3 Integration: A Datalad DataProvider provides an interface to move data from Datalad (Datalad, 2018) versioned resources into the CBRAIN ecosystem. A new S3 DataProvider provides data movement from cloud-based resources. BIDS Compatibility: Capitalizing on the BIDS standard, an automatic parallelization capability to ensure BIDSApps run efficiently. Provenance and error-handling are also available, and the user only needs to specify the BIDS-formatted input dataset and the pipeline to execute. Boutiques Integration: To enable computational pipelines to be discoverable and shareable, and ease the burden of integration and deployment, we have adopted the Boutiques JSON standard to define our computational pipelines and to pull new pipelines from the Boutiques Repository. New Interactive Interface: A user-focused, interactive and dynamic CBRAIN user-interface has been designed, built with React.js and GraphQL, interacting through the CBRAIN API to launch tasks and manage data, The UI is highly modular and can quickly be adapted to new user and visualization requirements over time. New modular visualizations have also been developed using React.js. Conclusions CBRAIN has for nearly a decade served as a platform for accomplishing large-scale, large-data neuroinformatics. The new implementations provide a connection to a larger community of neuroinformatics and scientific research and promote FAIR standards in computational science.
{"references":["CONP (2018), Canadian Open Neuroscience Platform, https://conp.ca","Das, S. (2016), 'The MNI data-sharing and processing ecosystem', NeuroImage, vol. 124, pp. 1188–1195.","Datalad (2018), available at http://www.datalad.org.","Glatard, T. (2018). Boutiques: a flexible framework to integrate command-line applications in computing platforms. GigaScience, 7, giy016.","Glatard, T. (2015). CARMIN: a common web API for remote pipeline execution. Frontiers in Neuroscience, vol. 9, no. 00053.","Gorgolewski, K.J. (2016). The brain imaging data structure, a format for organizing and describing outputs of neuroimaging experiments. Scientific Data, vol. 3, 160044.","Gorgolewski, K.J. (2017). OpenNeuro—a free online platform for sharing and analysis of neuroimaging data. Organization for Human Brain Mapping. Vancouver, Canada, 1677.","HBHL (2018), Healthy Brains for Healthy Lives, http://www.mcgill.ca/hbhl","Sherif, T. (2014), 'CBRAIN: a web-based, distributed computing platform for collaborative neuroimaging research', Front Neuroinform, vol. 8, pp. 54.","Vaccarino, A. L. (2018). Brain-CODE: A Secure Neuroinformatics Platform for Management, Federation, Sharing and Analysis of Multi-Dimensional Neuroscience Data, vol. 12, pp. 28."]}
Databáze: OpenAIRE
Externí odkaz: