Fig. 3

A heatmap view of the data representations based on (a) DIPD and (b) Seurat normalized and scaled counts before feature selection. The orders of cells and genes for both panels are based on the hierarchical clustering with Euclidean distance and Ward’s linkage using model departure. The black colored lines in the sidebars on the right represent the top 2000 most variable genes kept by the Seurat pipeline. Visually, both data representations effectively demonstrate the differential expressed genes between the two cell lines. However, highly expressed genes within single cells, as depicted by the bright red spots, may potentially play a role in clustering but many are filtered out by Seurat