CMStatistics 2022: Start Registration
View Submission - CMStatistics
Title: Robust statistical inference for cell type deconvolution Authors:  Jingshu Wang - University of Chicago (United States) [presenting]
Abstract: Cell type deconvolution is a computational approach to infer individual cell types' proportions from bulk transcriptomics data. Most existing methods for cell type deconvolution only provide point estimation of the cell type proportions without any uncertainty quantification, though the estimates can be very noisy due to various sources of biases and randomness. We will discuss a new statistical framework MEAD for cell type deconvolution to get more efficient estimators and construct asymptotically valid confidence intervals both for each individual's cell type proportion and for quantifying how cell type proportions change across multiple bulk individuals in downstream regression analyses. Our statistical inference takes into account the biological randomness of gene expressions across cells and individuals, gene-gene dependence, and cross-platform biases and sequencing errors, without any parametric assumptions. We also provide identification conditions of the cell type proportions when there are arbitrary platforms-specific biases across sequencing technologies.