[PAST EVENT] Mathematics Colloquium and EXTREEMS-QED Lecture: Greg Hunt (University of Michigan)
Access & Features
- Open to the public
Mathematics Colloquium and EXTREEMS-QED Lecture: Greg Hunt (University of Michigan)
Title: Estimating Cell-type Proportions Using Gene Expressions
Abstract: Shifts in cell type composition are hallmarks of developmental processes of organisms including embryogenesis, morphogenesis, cell differentiation and growth. They can also be a sign of disease and dysfunction. Flow cytometry, which researchers typically employ to establish composition, is time consuming and expensive. It has thus rarely been used in combination with techniques measuring gene expression, despite cell type composition having the potential to confound results from differential gene expression analyses. Given the importance of understanding cell type composition, several methods estimating cell type proportions from gene expression data have been developed. We propose dtangle, a new method for estimating cell type proportions from gene expression data. Bench testing shows that dtangle outperforms other published methods. Moreover, owing to its simplicity, dtangle is up to four orders of magnitude quicker than existing methods. Furthermore, dtangle?s accuracy and efficiency are robust to selection of its tuning parameters. As a case study, we investigate patients with Lyme disease. dtangle?s estimated cell type proportions reveal a temporal trend in the Lyme disease immune response that is consistent with previous findings. Furthermore, the proportions are important covariates in differential expression analysis across disease status.