Modelization of single-cell gene expression along a timeline

Report abstract

The explosion of genomic data over the past two decades created a need for new and more performing tools. One type of data in particular, Single-cell RNA Sequencing data, allow for a cell-level detection of biological patterns but technical limitations add lots of noise sources to the data. In particular, some counts are dropped and appear as zeros, even thought the gene is expressed in the cell. Here, we present a new method to detect dropouts in cases of development trajectories. Our various model all improve substantially on random guesses and offer a computationally-efficient way to infer correct gene expression patterns over time. The GitHub containing all our scripts is available online at the following address: https://github.com/HectorRDB/CS289_Final_Project

Spring 2018

Group members: Daniel Amar, Ariane Lozac’hmeur, Hector Roux de Bézieux & Alexandre Vincent

• The project proposal is available on Overleaf and may be read here.
• The project report is available on Overleaf and may be read here.