Weighted inference of gene expression variability in single cell RNAseq data for gene set tests: a replication study

In the gene differential expression analysis of Single Cell RNA-seq data, the aim is to sort out functional gene sets for specific diseases or traits of interest. The mean-variance relationship of the log-counts is also evaluated to generate precision weights for empirical Bayes analysis. We demonstrate through a gene set test study that the functional gene sets identified by the permutation tests on samples of different biological conditions are largely different. The permutation-based test uses the transformation of coefficient of variation (cv) as a novel variability measure and the empirical Bayes statistical methods borrow strength between genes for exact type I error rate control. We also reduce the mean variance trend incurred by scRNA-deq data’s violation of the normal assumption of intermediate linear modeling steps. We apply the methodology to two samples of single cell transcriptomics from lymphoid organs (spleen) of humanized mice with one Mock mouse and one HIV-1 infected mouse. Our results agree with previous findings and in the case of the HIV-infected mouse dataset suggest an additional perspective of interest.Bachelor of Science in Public Healt