Figure 5.

ISnorm improves performance of HVG detection in the PBMC dataset. (A) The schematic view of motif enrichment analysis. TFs with highly variable activity should be enriched in the regulatory elements of HVGs compared to non-HVGs. (B) The motif enrichment patterns of top 20 TF motifs with highest variability using ISnorm and other normalization methods. The upper part shows the percentage of HVG-associated peaks containing specific motif based on ISnorm and other normalization methods and the lower part shows the point estimates and error bars of accessibility variation of motifs from scATAC data.