Unraveling the timeline of gene expression: a pseudo-temporal trajectory analysis of single-cell RNA sequencing data

Abstract

There has been a rapid development in single cell RNA sequencing (scRNA-seq) technologies in recent years. Droplet-based single cell platforms such as the 10x Genomics Chromium system enable gene expression profiling of tens of thousands of cells per sample. The goal of a typical scRNA-seq analysis is to identify different cell subpopulations and their respective marker genes. Trajectory analysis can also be used to infer the developmental or differentiation trajectories of cells by ordering them along a putative lineage tree based on their gene expression profiles. This analysis positions cells and cell clusters along a pseudotime trajectory that represents a biological process such as cell differentiation, development, or disease progression. Here we demonstrate a time-course analysis to identify genes that are significantly associated with pseudotime. The article demonstrates a comprehensive workflow for performing trajectory inference and time course analysis on a multi-sample single cell RNA-seq experiment of the mouse mammary gland. The workflow uses open-source R software packages and covers all steps of the analysis pipeline, including quality control, doublet prediction, normalization, integration, dimension reduction, cell clustering, trajectory inference, and pseudobulk time course analysis. Sample integration and cell clustering follows the Seurat pipeline while the trajectory inference is conducted using the monocle3 package. The pseudo-bulk time course analysis uses the quasi-likelihood framework of edgeR.