Open Science

LORIS, EEG-BIDS, Git-Annex

May 14th, 2018
Samir Das - Associate Director of Research Software Development
McGill Centre for Integrative Neuroscience
Montreal Neurological Institute
samir.das@mcgill.ca

Best practices in Data Sharing

Committee on Best Practices in Data Analysis and Sharing (COBIDAS)

Richard Stallman
Visit to the MNI
May 2017

Open (Free) Science

Cyberinfrastructure

Specific Compnonents of Open Science

WHAT IS LORIS?

“LORIS is a modular and extensible web-based data management system that integrates all aspects of a multi-center study: from heterogeneous data acquisition (imaging, clinical, behavior, genetics) to storage, processing and ultimately dissemination.”

LORIS Functionality

LORIS Dashboard

Imaging Data

Cross-modal querying

Real-time Query Results

Genomics Browser

BIDS Format

Centralized or Distributed?

Thank you!Special thanks to Pierre Bellec... for being so geeky. :)... and Cecile Madjar

LORIS team on left

Improved Infrastructure

Databases, NoSQL, APIs, Version Control, Provenance Capture

NoSQL

LORIS API

Instrument Format

NeuroImaging Data Model

API Building and Standardization

Consolidated datasets

ADNI, ICBM, NIHPD, Allen Mouse Brain, IBIS, Generation-R, ABIDE, ABIDE Preprocessed, ADHD 200, ADHD Preprocessed, Human Connectome Project, OMEGA, UK Biobank, Edinburgh Biobank, BigBrain, Talairach, 1000 Functional Connectomes, Colin 27, MNI 305, 1000 Brains, AAL, ANIMAL, MAVAN, PreventAD, PING, MNI 152, MNI 305, FSL...

Not enough datasets!

BigBrain

ADNI

  • Cited 343 times

  • Data-use agreements

  • Used for countless analyses

  • Restrictive
  • Cited 295 times
  • Control data for NDAR
  • Longitudinal study (8 years)
  • 532 subjects
  • 8000 distinct variables
  • 37000 individual assessments
  • T1, T2, PD, DTI, Spectroscopy
  • ∼3TB of imaging data
  • 2000 MRI acquisitions

  • Autism infant brain development
  • 3000+ scans
  • 700+ subjects
  • 7000 distinct variables
  • 150,000+ individual assessments
  • T1, T2, DTI, BOLD
  • ∼5 TBs imaging data
  • Genetic/biospecimen data

CCNA

  • $35 million

  • CIHR funded

  • Pan Canadian Consortium

  • Acquisition in June, 2016

  • 35 collections sites

  • LORIS powered

ABIDE

Quebec Parkinson Network

OMEGA

  • Open MEG Archive

  • First open MEG repository

  • 180 user accounts created

  • Associated structural data

  • 500+ reads on academia.edu

Tools and Environments

Neurovault, NeuroSynth, CIVET, VIP, Boutiques, Git-Annex, SOLID, BIDS, NiDM, DiCAT, DCMTK, NiPype, ITK, Freesurfer, SPM, FSL, Mobile MRI, 1000 Brains, AAL, BrainCode, GitHub, Amazon Cloud, FSL, IDA, BrainVisa, DICOM Confidential, DockerHub, Gate, CMIND...

Too many Tools!

Neurovault Example

BrainBrowser

..a set of web-based 3D visualization tools primarily used for viewing neurological data i.e. MRI scans.

It allows for real-time manipulation and analysis of 3D neuroimaging data through any modern web browser.

Data Publishing

Not Data Sharing

Data Publishing Geeksquad for Persistent DOIs

Hackathons

Important value of hackathons to data sharing initiatives

What’s involved?

Longitudinal Acquisition, Storage and Curation, Interoperability, Reproducibility, Transfer, Anonymization, Security, Privacy, Ethics, APIs, Validation, Quality Control, Protocol Checking, Preprocessing, Analysis, HPC, Provenance, Ontological Standarization, Data Harmonization, Upgrades, Maintenance, Bug Fixes, User Interface, Javascript, Bootstrap, Tracking, Extensibility, Data Management, Summary Statistics, Workflows, Development, Tool Integration, Data Sharing, Download, Multi-Modal Linking, Querying, Image Processing, Visualization, Networking, System Administration, Partnerships, Funding, HR bureaucracies ...No big deal!

Data flow

WHAT IS LORIS?

“LORIS is a modular and extensible web-based data management system that integrates all aspects of a multi-center study: from heterogeneous data acquisition (imaging, clinical, behavior, genetics) to storage, processing and ultimately dissemination.”

LORIS Dashboard

Imaging Data

Cross-modal querying

Real-time Query Results

Genomics Browser

CBRAIN

Several default tools:

  • CIVET
  • CivetCombiner
  • CivetQC
  • Freesurfer
  • SPM-batch
  • NIAK
  • FSL (bedpostx, bet, fast, feat, first flirt, melodic, probtrackx)
  • Numerous converters
  • And many more...

cbrain-support.mni@mcgill.ca

CBRAIN projects

TRY ME!

NOW

611 users; 199 international
191 sites
299 countries

NIAK

Challenges and Hurdles
Future unknowns Fear of getting scooped
Waste/duplication Technical challenges
Decreased exposure Privacy concerns
Changing the Publishing Culture Data Harmonization
Licensing and legal aspects Interoperability
More attrition Reproducibility
Sustainibility Obtaining ethics
Funding Scalability

Privacy Concerns

Adrian Thorogood BIC lecture - March 9, 2016

Buy-in from other researchers!

Open Science Beers

Beers at Else's - Wednesday's at 4:44pm

Thank you!Acknowledgements: Alex Zijdenbos, Dario Vins, Jonathan Harlap, Matt Charlet, Andrew Corderey, Sebastian Muehlboeck, Reza Adalat, Louis Collins, Vladimir Fonov, Marc Rousseau, Mia Petkova, Rathi Gnanasekaran, David Brownlee, Tarek Sherif, Pierre Rioux, Nic Kassis, Leigh MacIntyre, Claude Lepage, Ilana Leppert, Natasha Beck, Tristan Glatard, Bert Vincent, Lindsay Lewis, Najma Mahani, Elodie Portales-Casamar, Alden Woodward, Sylvain Milot, Jean Francois Malouin, Sylvain Baillet, Daniel Kroetz, Martin Weiss, Mathieu Desrosier, Jason Karamchandani, Amit Bar-Or, Ted Fon, John Brietner, Derek Lo, Patrick Bermudez, Chris Steele, Pamela Patterson and one of my favourites: Pierre Bellec!

LORIS team on left
Special thanks to Alan Evans for making all of this possible.