R in Grenoble DATA CHALLENGES Magali Richard & Florent Chuffart - PowerPoint PPT Presentation

R in Grenoble DATA CHALLENGES Magali Richard & Florent Chuffart

Introduct ction Data challenges in class Data challenges for scientists Tutorial CIT-SAB MEETING 2

What is a data challenge? Ellrott et al. Genome Biology (2019) 20:195 CIT-SAB MEETING 2

The challenge platform Ø Enables participants to submit their codes Ø Automatically rank the participants Alexis Arnaud CIT-SAB MEETING 12

History 2013: Medical data. Result submission. COMPETITION BUNDLES 2014: Computer vision, speech, NLP, IR. MSCOCO: 361 participants. 2015: CODE SUBMISSION AutoML: 687 participants Hackathons. Coopetitions. 2016: ChaLab Wizard. USE IN EDUCATION 2017: 480 challenges, 10000 users. SCALABILITY, REUSABILITY, DYNAMIC COMPETITIONS See.4c: EU prize, 2 million Euros

Introduction Data ch challenges in cl class Data challenges for scientists Tutorial CIT-SAB MEETING 2

A novel pedagogic approache As practical work As homework As final evaluation CIT-SAB MEETING 2

Chagrade: a dedicated tool in codalab CIT-SAB MEETING 2

Introduction Data challenges in class Data ch challenges for sci cientists Tutorial CIT-SAB MEETING 2

Goal of our data challenge • Quantification of tumor heterogeneity Normal cells Tumor cells ? CAF Immune cells CIT-SAB MEETING 3

A challenge for scientists CIT-SAB MEETING 4

2 editions of the data challenge 1 rst edition (2018) 2 nd edition (2019) • Methylation Data • Methylation and transcriptomic Data • One cancer type • Several cancer types • Cell lines • Primary tumors / cell lines CIT-SAB MEETING 4

When and where? • When : 25-29 of November 2019 • Where : Aussois (CAES CNRS), French Alps CIT-SAB MEETING 6

About the participants (n=34) Institut Curie UGA Grenoble Graduate CRC Cordeliers students INSERM France CEA Innate Pharma Engineers Verteego … Byelorussia Bioinformatics Undergrad students Sweden Industry Researchers Norway Germany Postdoc Luxembourg Computer science Statistics Data science 10 teams of 3-4 people Medical science CIT-SAB MEETING 7

Agenda #1 #2 #2 Talks #2 Guidelines PRACTICALS Social event Feed #1 Back Brain Storming #2 Poster session CIT-SAB MEETING 9

Valorization of the first edition Ø Guidelines Ø Article Ø Blog posts In review in BCM bioinformatics Health data challenges organization: feedback, comments and recommendations. Authors: Elise Amblard, Yuna Blum, Jane Merlevede, Magali Richard Ø R package medepir In preparation https://rdrr.io/github/bcm- uga/medepir/man/medepir-package.html M Richard, C Decamps , F Privé, M Blum CIT-SAB MEETING 16

From data challenges to benchmarks

Introduction Data challenges in class Data challenges for scientists Tu Tutorial CIT-SAB MEETING 2

On line tutorial by Alexis Arnaud

It’s up to you now!

Thank you for your attention ! https://app.secure.griffith.edu.au/events/event/61593 CIT-SAB MEETING 20