Single cell RNA-seq and ATAC-seq data integration for macrophage gene networks
Chronic wounds represent an escalating health problem around the world, especially in diabetic patients.. Although multiple factors contribute to the development and impaired healing of chronic wounds, a common characteristic of these poorly healing wounds is a persistent inflammatory response, including the accumulation of pro-inflammatory macrophages. Macrophages are capable of performing a diverse array of tasks during wound healing, ranging from destructive killing functions to pro-healing and homeostatic duties. Recent studies have demonstrated that macrophages adopt a spectrum of phenotypes, especially in vivo, which are difficult to fit into the (pro-inflammatory)/M2 (pro-healing) scheme. In addition to the limited and biased characterization of macrophage phenotypes during wound healing, the regulation of macrophage phenotypes remains poorly understood. We have generated single cell RNAseq (scRNAseq) datasets on macrophages isolated from skin wounds to assess heterogeneity of mRNA profiles and are starting to generate parallel single cell ATACseq (scATACseq) datasets to identify locations of accessible chromatin and thus the regulatory landscape of genes. Although these data will provide valuable information on macrophage heterogeneity during wound healing, computational methods for integrating scRNAseq and scATACseq data are currently limited. A critical first step in analysis of scATACseq data is to enrich signals by grouping cells into clusters (pseudo-bulk) of a similar epigenetic profile for downstream analysis such as chromatin accessibility and transcription factor and target identification. Currently, this step of signal enrichment is carried out using data-driven procedures on scATACseq data alone in most of the tools. We hypothesize that generating cell clusters by modeling transcription factor activities from scRNAseq data will be more biologically relevant and could be a better alternative for signal enrichment in scATACseq data analysis to deliver better downstream results. In Aim A, we will use the BITFAM generated clusters as pseudo-bulks to enrich the scATACseq signals. The downstream analysis of scATACseq data will generate context-specific transcription factor gene targets, which in turn will inform BITFAM to generate better clustering of cells and facilitate the new iteration of scATACseq signal enrichment. In Aim B, we will repeat the procedure in Aim A until it converges to a cell clustering result where each cluster is formed from cells with shared transcription factor activities and gene targets supported by the two omics data. Impact: Developing computational methods for integrating single cell RNAseq and ATACseq data will enable a better understanding of the transcription factor networks critical for regulating macrophage heterogeneity over the course of normal and impaired healing, and could be extended to other cell types, tissues and disease states.

Curation Flag Information