Computational Challenges in Single-Cell Genomics - Confounding Factors, Cell Types and Spatial Transcriptomics
Event details
| Date | 06.10.2014 |
| Hour | 12:15 |
| Speaker | John C. Marioni, PhD, European Bioinformatics Institute (EBI-EMBL) and Wellcome Trust Sanger Institute, Cambridge (UK) |
| Location | |
| Category | Conferences - Seminars |
BIOENGINEERING SEMINAR
Abstract:
Recent technical developments have enabled the transcriptomes of thousands of cells to be assayed in an unbiased manner. These approaches have enabled heterogeneity in gene expression levels across populations of cells to be characterized as well as facilitating the identification of new, and potentially physiologically relevant, sub-populations of cells.
However, to fully exploit such data and to answer these questions, it is necessary to develop robust computational methods that take account of both technical noise and underlying, potentially confounding, variables such as the cell cycle.
In this presentation I will begin by briefly describing how we used spike-ins to quantify technical noise in single-cell RNA-seq data, thus facilitating identification of genes with more variation in expression levels across cells than expected by chance. Subsequently, I will discuss a computational approach that uses latent variable models to account for potentially confounding factors such as the cell cycle before applying it to study the differentiation of Th2 cells. I will show that accounting for cell-to-cell correlations due to the cell cycle allows identification of otherwise obscured sub-populations of cells that correspond to different stages along the path to fully differentiated Th2 cells.
Finally, I will describe how understanding cell type identity in a multicellular organism requires the integration of each cell’s expression profile with its spatial location within the tissue under study. I will describe a high-throughput method that combines in vitro single-cell RNA-sequencing with a gene expression atlas to map single cells to their location within the tissue of interest. The utility of the method will be demonstrated by applying it to allocate cells to their precise location within the brain of the marine annelid Platynereis dumerilii.
Bio:
PhD in Applied Mathematics, University of Cambridge (2008), then Postdoctoral research in the Department of Human Genetics, University of Chicago.
At EBI-EMBL since September 2010.
Abstract:
Recent technical developments have enabled the transcriptomes of thousands of cells to be assayed in an unbiased manner. These approaches have enabled heterogeneity in gene expression levels across populations of cells to be characterized as well as facilitating the identification of new, and potentially physiologically relevant, sub-populations of cells.
However, to fully exploit such data and to answer these questions, it is necessary to develop robust computational methods that take account of both technical noise and underlying, potentially confounding, variables such as the cell cycle.
In this presentation I will begin by briefly describing how we used spike-ins to quantify technical noise in single-cell RNA-seq data, thus facilitating identification of genes with more variation in expression levels across cells than expected by chance. Subsequently, I will discuss a computational approach that uses latent variable models to account for potentially confounding factors such as the cell cycle before applying it to study the differentiation of Th2 cells. I will show that accounting for cell-to-cell correlations due to the cell cycle allows identification of otherwise obscured sub-populations of cells that correspond to different stages along the path to fully differentiated Th2 cells.
Finally, I will describe how understanding cell type identity in a multicellular organism requires the integration of each cell’s expression profile with its spatial location within the tissue under study. I will describe a high-throughput method that combines in vitro single-cell RNA-sequencing with a gene expression atlas to map single cells to their location within the tissue of interest. The utility of the method will be demonstrated by applying it to allocate cells to their precise location within the brain of the marine annelid Platynereis dumerilii.
Bio:
PhD in Applied Mathematics, University of Cambridge (2008), then Postdoctoral research in the Department of Human Genetics, University of Chicago.
At EBI-EMBL since September 2010.
Practical information
- Informed public
- Free