Nice Genomics, Statistics and Learning Day

Deep Learning, Learning and Statistics to study the biology of human diseases

Date and Location

May 28th, 2018

9:30 - 16:00

 

Laboratoire de Mathématiques J.A. Dieudonné

UMR n° 7351 CNRS UNS

Université de Nice - Sophia Antipolis

06108 Nice Cedex 02

Salle de Conférence, Rez-de-Chaussée


 

Organizers

Elisabeth Pécou (LJAD, MSI). Contact: epecou@unice.fr

Laurent Counillon (LP2M)

Charles Bouveyron (LJAD, Inria Sophia Antipolis Méditerranée

Context

Physiological fundamental processes like development and aging and pathologies are strongly correlated to cells' fundamental processes and their disruption, including transcription, translation, signaling and metabolism. Understanding the environmental and genetic perturbations and the processes by which they induce a disease at the cellular level in humans and model organisms opens the way to new personalized, targeted therapeutic strategies. This field of research is benefiting from the great technological advances brought by Next Generation Sequencing tools by giving access to an unprecedented level of details on the components of the cell and their dynamics.

Large datasets are produced that report on:

(1) structure and activity of the genome, e.g. DNA sequences, chromatin accessibility, genome-wide transcriptional profiles;

(2) structure and activity of proteins at the scale of the cell, e.g.  proteomic, metabolomic and lipidomic profiles. 

The growing size and heterogeneity of those datasets is a challenge for the task of modeling and analysis.  In response, plenty of dedicated "Next Generation Algorithms", mainly based on deep learning approaches, are flourishing, some of them establishing themselves as new gold standards, while others fall into forget. In practice, they are often used as black-boxes, their parameterization is left to the intuition of some expert, and without a clear view on the validity domain.

Objectives

The objective of the meeting is a trans-disciplinary workshop to build Biological Data Analysis Challenges:

  • A Challenge is a scientific question related to a biological topic that can be answered by analyzing a given set of experimental data with a statistical learning algorithm.
  • 
Data will come from biology labs (those could be new or already exploited datasets, e.g. for publication).
  • The scientific purpose and analysis algorithmic tools will be discussed during the meeting, to evaluate the feasibility of the Challenge.
  • Challenges will be classified following their degree of difficulty: (I) master, (II) PhD, (III) Post-Doc.

  • Challenges (I) and (II) will be proposed next year to students of the EUR "NGLS (Next Generation Life Scientists)"
, and other suited Master and Doctorate programs.
  • Financial support from the EUR NGLS will be considered for students training, master stipend and/or mobility.
  • Financial support from the UCA Center of Modeling, Simulation and Interaction will be considered for challenges (III) if sustained by a trans-disciplinary project involving a biological team and a mathematics/computer science team. 

     

Call for projects

Send by e-mail (to: epecou@unice.fr) a brief description of the data:

  1. biological information : type of data (e.g. transcriptome), experimental model (e.g. cell culture), number of conditions, acquisition techniques (e.g. RNA seq) …
  2. statistical information:
  •  Type of data: quantitative, qualitative, text, network, functional, …
  •  Volume: number of observations, number of variables
  • Availability: yes total/ partial/not yet (we encourage already available datasets)
  • Straight/Pretreated datasets?
  • Already existing analyses on the dataset?
  • Storage format?

and add a scientific question about the data prior to the meeting (Deadline: May 24th).

Program

First part: A time slot will be attributed to each project, including (1) a presentation of the project (biological model, experimental context, nature and size of the data, format availability, scientific motivation, etc), and (2) a round table to discuss the feasibility of the statistical learning analysis and (3) finalization of the challenge, including its grading.

Second part: Presentation of the project of a recurrent workshop in 2018/2019 on the topic of the day. Discussions.

 

Registration

Nous vous rappelons qu’en raison du plan vigipirate attentat et des normes en vigueur dans notre établissement l’inscription à cet événement est obligatoire.Toute personne non identifiée sur un listing d’inscription se verra refuser l’accès à la conférence/colloque/séminaire/événement. Vous devrez être en mesure de justifier votre identité.