Machine Learning for Single-Cell Biology
Title : Machine Learning for Single-Cell Biology
Asbtract :
Early developments in sequencing technologies have paved the way to the revolution of medicine, by providing a new framework for the molecular bases of disease, based on the understanding of between-patient genetic variability. There have been emblematic successes, like the characterization of subtypes of cancers and their molecular profiles, the elucidation of the genetic bases of rare diseases, among others. However, facing inefficiencies of treatments, it was rapidly shown that the cellular heterogeneity between and WITHIN patients should be accounted for in treatment strategies. In the late 2010s, technological shifts converged to give rise to single-cell genomics that have the unique capacity to provide high-throughput molecular portraits of individual cells, and to identify predictive biomarkers of a disease trajectory. Single-cell genomics has the potential to produce the so-called cell-based interceptive medicine that could revolutionize healthcare in cancer, cardiovascular, neurological, inflammatory, and infectious diseases.
The full understanding of this ever-increasing complexity of disease will be possible only by fully exploiting the ultra-large complexity of the produced data. The development of machine learning methods for dimension reduction, clustering and statistical testing appears particularly adapted to this very competitive field. In this talk, I will focus on graph-based non-linear dimension reduction on vectorial and functional single-cell data, and on the development of kernel-based methods to compare distributions, with an extention to the emerging field of spatialized genomic data. The idea of the presentation is to provide an overview of targeted machine learning questions that we currently and will investigate in the future !
More information :