Cell ranger pipeline

Single cell analysis and cluster identification was performed on 1,654 – 16,903 captured cells per sample using the 10X Chromium single cell system (10X Genomics Inc.) at the University of Oklahoma Health Sciences Campus Genomics Core Facility and sequenced each sample to a read depth of 95-164K reads/cell, yielding 85-92% sequence saturation. Read mapping and expression quantification were performed using a combination of the 10X Cell Ranger pipeline (10X Genomics Inc.) and custom Seurat analytic scripts (74 (link)). Briefly, single-cell reads were mapped to the human genome (GRCH38) and assigned to genes using the standard Cell Ranger pipeline. Normalized gene expression was then used to produce a UMAP plot that provided cell clusters based on similarity of gene expression. Once cells were assigned to a cluster, custom Seurat scripts were used to statistically derive the gene expression differences within and between cell clusters using t-tests in Seurat. Trajectory analysis was then conducted using Monocle (75 (link)). To be statistically similar across the study, the Monocle trajectories were used to guide specific trajectory specific t-tests within Seurat. From these, pathway analyses were performed using Ingenuity Pathway Analysis (IPA, QIAGEN, Hilden, Germany) on the differential transcriptional profiles seen in the cell clusters and trajectory groups.

Base calls were converted to fastq format and demultiplexed using Illumina’s bcl2fastq/2.16.0.10. Demultiplexed reads were aligned to the mouse reference genome (mm10) using Tophat v2.0.14 with default settings^{32 (link)}. For the RNA-seq data, aligned reads were quantified using Cuffquant v2.2.2, and then a normalized gene expression matrix was generated using Cuffnorm v2.2.2^{28 (link),33 (link)–35 (link)}. For the ATAC-seq data, PCR duplicates were removed using Samtools v1.9^{36 (link)}, and data from all alignment files were merged into a single file for peak calling using MACS v2.1.0 (parameters: --nomodel --keep-dup all --extsize 200 --shift -100 -B --SPMR --call-summits) to generate a master list of peaks observed in the experiment^{37 (link)}. A peak count matrix was calculated by using Bedtools v2.28.0^{38 (link)} to compute the intersection between each sample’s aligned reads and the master list of peaks. Expression matrices for the single cell data were generated using the 10X Cell Ranger pipeline (10X Genomics).