The ability to profile the transcriptomes of thousands of single cells has enhanced our understanding of complex tissues and enabled the discovery of novel cell types. As single cell technologies have matured, applications have shifted from simply characterizing cell types in a sample to comparing cellular composition and gene expression of samples between experimental conditions. However, the analysis of single cell data remains challenging as the data tends to be noisy, sparse, and prone to technical variation such as batch effects. The way that we account for and model variability in single cell data is key to robust and reproducible discoveries. In this talk I will describe some bioinformatics challenges that we faced when analysing heterogeneous samples from human heart biopsies. I will describe a new method, propeller, that we developed for finding statistically significant shifts in cell type composition between groups of samples that leverages biological replication. I will show some benchmarking comparisons with other statistical frameworks, and how we applied propeller to single cell and spatial data.
Dr Belinda Phipson
Lab Head, The Walter and Eliza Hall Institute of Medical Research
Dr Belinda Phipson is a laboratory head in the Bioinformatics Division at WEHI. In 2013 she completed her PhD with Professor Gordon Smyth on empirical Bayes methods for analysing gene expression data. During her postdoc with Professor Alicia Oshlack, she developed methods for analysing DNA methylation arrays and single cell data, and had significant collaborations in the areas of healthy development and stem cell derived kidney organoids. In 2018 she was awarded the inaugural ABACBS early career researcher award, and in 2019 she was awarded an NHMRC Emerging Leader Investigator Grant. The focus of her lab is on developing bioinformatics methods for data from cutting edge single cell assays; including transcriptomics, proteomics and spatial technologies.