Fig. 7: Data requirements for single-cell DA analysis.
From: Best practices for differential accessibility analysis in single-cell epigenomics
a Effect size (Cohen’s d) of downsampling plate-based scATAC-seq data to a mean of 500, 1000, 2000, 5000 counts per cell on the AUCC for single-cell DA methods using matching bulk ATAC-seq as a reference, relative to DA analysis of the same datasets with a mean of 10,000 counts per cell (n = 16 comparisons). Inset text shows the median Cohen’s d. b Effect size (Cohen’s d) of downsampling single-cell multi-omics data to 20, 50, 100, 200, or 500 cells per condition on the AUCC between the ATAC and RNA modalities, relative to DA analysis of the same datasets with 1000 cells per condition (n = 306 comparisons). Inset text shows the median Cohen’s d.
