RNA-Seq Signatures Normalized by mRNA Abundance Allow Absolute Deconvolution of Human Immune Cells

Abstract

The molecular characterization of immune cell types is important for designing and developing effective strategies to understand and treat diseases. We characterized 29 different immune cell types within the peripheral blood mononuclear cell (PBMC) fraction of healthy donors using RNA sequencing (RNA-Seq) and flow cytometry. Our RNA-Seq resource has been used first to identify set of genes that are specific, co-expressed and have housekeeping role across the 29 immune cell types. Then, we analyzed the RNA composition for each immune cell type in terms of their relative mRNA heterogeneity and abundance. Lastly, we develop a novel method of gene expression normalization in order to perform absolute deconvolution at a fine resolution from both RNA-Seq and microarray datasets. The resources were validated in independent cohorts and benchmarked with different deconvolution and normalization methods. We believe that this work has clinical and diagnostic value by allowing us to attribute observations in bulk RNA data to specific immune cell populations.