Mots-Clés
single cell genomics, machine learning, functional data analysis
Description
Massively parallel sequencing and high-throughput cell biology technologies have paved the way for a better investigation of the suspected but unaccessible cell-to-cell variability of molecular profiles, based on DNA, RNA, chromatin states and conformation. Single-cell genomics now allows the study of cell-to-cell variability within a biological sample and investigate new questions that were out of reach for classical bulk genomics. However, the full exploitation of this incredible wealth of data can not be done without new machine learning and computational breakthroughs to handle the unprecedented complexity and volume of data already at hand. The challenge ahead concerns the dimensionality of the data combined with a high variability of biological processes at stake. While concensus methods have already been proposed for dimension reduction based on expression profiles, the spatial organization of genomic data along the DNA molecule in 1D or even 3D raises methodological challenge that have not yet been investigated. However this information is crucial, for instance when considering chromatin accessibility, for which the accessibility at a given position strongly depends on the neighboring states. In this project we will focus on cutting edge single cell replication timing data, whose purpose is to measure the dynamics of replication at the single cell level. Replication Timing is strongly organized along the genome, and the proper description of the inter-cellular variability of this essential genomic process requires dedicated machine learning methods that account for this spatial structure. Among the SingleStatomics consortium, we propose a 18 month postdoc position that will be dedicated to the development of new methodologies based on functional data analysis and dimension reduction. The postdoc will be supervised by a group of specialists in machine learning for single cell genomics (F. Picard LMBC ENS de Lyon, J. Chiquet, AgroParisTech), in collaboration with specialists in the mathematical aspects of statistical learning (V. Rivoirard and A. Roche, CEREMADE, Université Paris Dauphine), and a specialist in replication (M.N. Prioleau, Institut Jacques Monod). The candidate will develop a statistical dedicated to dimension reduction for spatially organized single cell data, and will also analyze some genomic data.
The candidate should have a solid background in bioinformatics and/or statistical learning. Prior knowledge in single cell genomics would be appreciated but not mandatory. The position could start between september 2021 and january 2021 (to be discussed), for 18 months, based in Lyon.
To apply send a CV and the contact details of two referees to Franck Picard franck.picard@ens-lyon.fr