MIG Seminar – Davis McCarthy – 5th April, 2019

More Information

Andrew Siebel

asiebel@unimelb.edu.au

T: +61 3 8344 0707

Davis McCarthy

Head, Bioinformatics & Cellular Genomics Laboratory, St. Vincent's Institute
Group Leader, Melbourne Integrative Genomics

Friday 5th April
12-1pm
Agar Theatre, BioSciences 4 Building, The University of Melbourne

The effects of DNA variation on gene expression in single cells

How can we understand the effects of DNA variation on gene expression in single cells? What sort of studies might we like to conduct and what computational tools do we need for them to succeed? I will describe some of the opportunities available to improve expression quantitative trait locus mapping with single-cell data and to gain new views of the structure within clonal cell populations at unprecedented resolution. I'll also discuss the challenges inherent in scaling up single-cell data generation to assay cells from tens to hundreds of individuals. Throughout, I will introduce some of the hierarchical and variational Bayes approaches we have used to tackle these problems. Most specifically, I will demonstrate approaches for inferring clonal structure in a cell population by combining bulk exome and scRNA-seq data. I will present a two-step statistical method for the assignment of cells to a clonal tree structure, utilising detected mutations in sparse single-cell expression data.

We have applied these three approaches to inferring clones and assignment cells to clones with bulk exome and scRNA-seq data from 32 healthy human fibroblast lines derived from distinct human donors. Even with great variation in clonal structure across populations, we show that we are able to robustly identify clonal populations and assign a majority of cells to clones, even from a limited number of somatic mutations sparsely detected in scRNA-seq data. Assigning cells to clones using scRNA-seq data instead of scDNA-seq data reduces the cost-per-cell of studies such as this, while also enabling interrogation of the underpinning biology. We can analyse mutations across clones as when using scDNA-seq data, but the scRNA-seq data additionally provides a readout of the transcriptional state of each cell, allowing comparison of gene expression and transcriptional behaviour between cells assigned to different clonal populations. This approach has the potential to enrich our view of mutational processes at single-cell resolution, applicable in both cancerous and healthy tissues.

Enquiries: Andrew Siebel (asiebel@unimelb.edu.au)