Single cell RNA sequencing of vascular smooth muscle cells from atherosclerotic lesions of ApoE ko mice

This report details the comprehensive computational analysis of single-cell RNA sequencing (scRNA-seq) data generated from Single cell RNA sequencing of vascular smooth muscle cells from atherosclerotic lesions of ApoE ko mice study. The primary objective of this study was to characterize cellular heterogeneity, identify distinct cell populations, and investigate cell state transitions, which define the biology of the clonally expanding smooth muscle cells (SMCs) that contribute to atherosclerotic plaques.

• The analysis pipeline commenced with raw sequencing reads obtained from NCBI SRA:PRJNA626450. Initial bioinformatic processing involved quantifying transcript abundances for each cell using Salmon against a GENCODE (Release M36 (GRCm39)) reference transcriptome, incorporating decoy sequences for robust alignment and quantification. The commands used for raw data download, reference preparation, index building, and Salmon quantification are documented in Data Processing: Download, Indexing, and Quantification.

• The resulting transcript-level quantifications were imported into R using tximeta to generate a gene-level count matrix, represented as a SingleCellExperiment object. Rigorous quality control (QC) was then performed using the scater package to assess library quality and filter potentially compromised cells based on metrics such as total library size, number of detected genes, and mitochondrial transcript percentage, employing adaptive thresholds based on median absolute deviations.

• Following QC, the filtered dataset underwent standard scRNA-seq downstream analysis steps. Counts were normalized using deconvolution size factors (scran) and log-transformation (scater). Feature selection identified highly variable genes (HVGs) crucial for capturing biological variation. Dimensionality reduction was performed using Principal Component Analysis (PCA) on the HVGs. Subsequently, graph-based clustering (scran) was applied to the principal components to partition cells into distinct groups by (clustering). These clusters were visualized using Uniform Manifold Approximation and Projection (UMAP). To assign biological meaning to the identified clusters, marker genes significantly upregulated in each group were identified using scran::findMarkers. The expression patterns of top markers were visualized via heatmaps and violin plots, facilitating preliminary annotation of cell types, such as progenitor-like, transitional, and fibroblast-like smooth muscle cells

• Finally, to explore potential dynamic processes or developmental relationships between the identified cell populations, trajectory inference was conducted using slingshot. Based on the marker gene expression and the hypothesis that progenitor-like cells differentiate towards a fibroblast-like state via an intermediate transitional population, cellular lineages were constructed, rooted in the Progenitor-like cluster. Pseudotime values, representing cellular progression along the inferred trajectory, were calculated and visualized on the UMAP embedding. The expression dynamics of key genes along pseudotime were examined to understand the molecular changes accompanying the inferred process, and genes strongly correlated with pseudotime were identified as potential drivers of the transition.

This report presents the detailed methodology, results, and visualizations for each stage of this computational workflow, providing insights into the cellular landscape and dynamics.