Extending the Boutiques Tool Description Framework in CBRAIN

2276 

Abstract Submission 

Sergiy Boroday¹, Pierre Rioux¹, Natacha Beck¹, Darcy Quesnel¹, Xavier Lecours-Boucher¹, Najmeh Khalili-Mahani², Tristan Glatard³, Reza Adalat¹, Samir Das⁴, Bryan Caron⁵, Alan Evans⁶

¹McGill Centre for Integrative Neuroscience,Ludmer Centre,MNI,McGill University, Montreal, Canada, ²McGill Centre for Integrative Neuroscience,Ludmer Centre,MNI,Perform Research Centre, Concordia Univ, Montreal, Canada, ³Concordia University (Department of Computer Science and Software Engineering), Montreal, Canada, ⁴McGill Centre for Integrative Neuroscience, Montreal Neurological Institute, McGill University, Montreal, Quebec, ⁵McGill University, Montréal, Quebec, ⁶Montreal Neurological Institute and Hospital, Montreal, QC

Sergiy Boroday  
McGill Centre for Integrative Neuroscience,Ludmer Centre,MNI,McGill University
Montreal, Canada

Pierre Rioux  
McGill Centre for Integrative Neuroscience,Ludmer Centre,MNI,McGill University
Montreal, Canada

Natacha Beck  
McGill Centre for Integrative Neuroscience,Ludmer Centre,MNI,McGill University
Montreal, Canada

Darcy Quesnel  
McGill Centre for Integrative Neuroscience,Ludmer Centre,MNI,McGill University
Montreal, Canada

Xavier Lecours-Boucher  
McGill Centre for Integrative Neuroscience,Ludmer Centre,MNI,McGill University
Montreal, Canada

Najmeh Khalili-Mahani  
McGill Centre for Integrative Neuroscience,Ludmer Centre,MNI,Perform Research Centre, Concordia Univ
Montreal, Canada

Tristan Glatard, Associate Professor  
Concordia University (Department of Computer Science and Software Engineering)
Montreal, Canada

Reza Adalat  
McGill Centre for Integrative Neuroscience,Ludmer Centre,MNI,McGill University
Montreal, Canada

Samir Das  
McGill Centre for Integrative Neuroscience, Montreal Neurological Institute, McGill University
Montreal, Quebec

Bryan Caron, Dr.  
McGill University
Montréal, Quebec

Alan Evans  
Montreal Neurological Institute and Hospital
Montreal, QC

CBRAIN (Sherif et al., 2014) (https://cbrain.ca) is an open source, web-based, collaborative research software platform that addresses major challenges in big data research. CBRAIN allows scientists to launch, monitor and share large-scale big data analyses using advanced scientific tools through an easy to use web-based interface.

The ability to easily integrate new tools and update existing ones is an important feature of CBRAIN. Initially within CBRAIN each tool integration required the web form and related logic to be coded directly within the Ruby on Rails web development framework. That approach has been updated using Boutiques (Glatard et al., 2018) to define the command line and parameters in a JSON format. The use of Boutiques offers multiple parameter validation features and enables clearer form validation messages. Based upon the descriptor, CBRAIN automatically builds the user interface for each tool allowing users to select application parameters while automatically performing parameter validation.

While no formal scheme can possibly foresee all the tool description aspects, the Boutiques format provides a 'custom' section. In this section of the descriptor, CBRAIN-specific properties such as validation of input parameters against a list of CBRAIN supported data types or file naming schemes is possible.

In CBRAIN tools have been integrated via three different methods:
1: A CBRAIN developer created Ruby code and templates (2009 to 2015).
2: Support was added to CBRAIN using Boutiques descriptors for automated tool integration (2015 to 2020). The code to run the tool and the user interface are generated in CBRAIN at boot time based on templates. Even though this method facilitated tool integration, some limitations persisted, including difficulties in developing, maintaining and documenting the logic related to custom Boutiques descriptor properties.
3: CBRAIN uses Boutiques descriptors but with a new integration method performed at runtime using hooks and a library of modules where each module handles one custom property/keyword (2020 to present). With this method, tool developers can easily introduce CBRAIN-specific properties to Boutiques that enhance the standard behavior.

The newest method to integrate scientific tools leveraging the Boutiques descriptor at runtime has the benefit of making the integration more versatile. Each keyword is handled by a dedicated module resting either in the core CBRAIN codebase or in a plugin repository. Each module contains documentation and examples. A module overrides or modifies one or several standard CBRAIN procedures, related to rendering the input web form or tool execution. The new method can also modify the Boutiques descriptor for the duration of some steps, for example, by adding prepackaged inputs or outputs. Some of the CBRAIN modules used in conjunction with the Boutiques descriptors include:

Before running the pipeline:
CBRAIN can verify the type of the input file and the pattern of input file name with BoutiquesFileTypeVerifier and BoutiquesFileNameMatcher modules.
After the pipeline has run:
CBRAIN can specify a set of valid exit codes BoutiquesAllowedExitCodes module.
CBRAIN can specify post-processing cleanup actions with BoutiquesPostProcessingCleaner.

Using the Boutiques tool integration methods in CBRAIN, tool integrators are no longer required to write any Ruby code. Sometimes additional features and customizations not yet supported by Boutiques and CBRAIN are needed. With guidance from the CBRAIN development team, tool integrators can add their own modules. CBRAIN is able to support a variety of users and projects, together with their complex workflows, through customizations that are alone not definable within standard Boutiques. Since 2020 more than ten modules and properties have been added to CBRAIN, demonstrating the adoption and utility of this module-based approach to extending the capabilities of Boutiques.

Image Registration and Computational Anatomy

Databasing and Data Sharing

Workflows ¹

Informatics Other ²

Computational Neuroscience

Computing

Data Organization

Data Registration

Development

Informatics

Neurological

Open Data

Open-Source Software

Workflows