How the sequencing-depth normalization works when construct Jaccard index matrix

[Meijiao Wang]

Hi all,

Because the Jaccard similarity method is highly sensitive to sequencing depth of cells, so it is a must to normalize window-cell matrix to sequence depth of each cell. But it seems few document in the snapATAC manual according to this issue (or I missed something). Could you please explain some details for this normalization step? Because it is important to understand what happens when dealing with very sparse (low-depth) datasets.

By the way, how did you compose the grip-like figure (heatmap + text table ),like figure 2 b and e in the paper "Single-nucleus analysis of accessible chromatin in developing mouse forebrain reveals cell-type-specific transcriptional regulation"(the nature neuroscience paper)? by python matplotlib? or by python svgwrite? or simply edit and combine in some figure-editor like AI or Indesign ?