Supplementary Figure 1: Droplet-based scATAC-seq workflow and quality control measurements. | Nature Biotechnology

Supplementary Figure 1: Droplet-based scATAC-seq workflow and quality control measurements.

From: Massively parallel single-cell chromatin landscapes of human immune cell development and intratumoral T cell exhaustion

Supplementary Figure 1

(a) Protocol steps for scATAC-seq in droplets. (b) Genome tracks showing the comparison of aggregate scATAC-seq profiles from A20 B lymphocytes (top panel). scATAC-seq profiles were obtained from four independent experiments, as indicated. The bottom panel shows accessibility profiles of 100 random single A20 cells from two cell mixing experiments. Each pixel represents a 100bp region. (c) Left plot: Pearson correlation heatmaps of log-normalized reads in bulk GM12878 Omni-ATAC-seq peaks in aggregate scATAC-seq profiles generated from varying numbers of single cells, or from published Omni-ATAC profiles5 (n=100,000 ATAC-seq peaks). Right plot: Pearson correlation heatmaps of log-normalized reads in aggregate scATAC-seq profiles from A20 cells (n=100,000 ATAC-seq peaks, identified in an aggregate profile from all cells). Numbers in parentheses indicate the cell loading concentration. (d) Peak recovery analysis with subsampled cells and unique fragments as determined by x-axis and colors, respectively. scATAC-seq cells were subsampled to the indicated unique fragments, and the proportion of peaks recovered from the aggregate profile was calculated as a function of number of cells analyzed. GM12878 cells generated a median of 29,451 unique nuclear fragments per cell (top level of down-sampling was 25,000) while A20 cells generated a median of 20,809 unique nuclear fragments per cell (top level of down-sampling was 20,000). The center line represents the Loess fit, and shaded regions indicate 95% confidence interval (n=16 sub-sampled profiles at each read depth). (e) Pearson correlation analysis with subsampled cells and unique fragments as determined by x-axis and colors, respectively. scATAC-seq profiles were subsampled to the indicated unique fragments, and Pearson correlation to the aggregate profile was calculated as a function of number of cells analyzed (n=16 sub-sampled profiles at each read depth). (f) Analysis workflow for scATAC-seq data in this study.

Back to article page