Skip to main content
Fig. 6 | Genome Biology

Fig. 6

From: PseudotimeDE: inference of differential gene expression along cell pseudotime with well-calibrated p-values from single-cell RNA sequencing data

Fig. 6

Application of PseudotimeDE, tradeSeq, and Monocle3-DE to the mouse bone marrow dataset. a UMAP visualization and inferred pseudotime by Slingshot. Pre-defined cell types are marked by colors. Slingshot returns a bifurcation topology, denoted as lineage 1 (left) and lineage 2 (right). b UMAP visualization and inferred pseudotime by Slingshot on ten random subsamples. Four out of ten subsamples do not yield bifurcation topology but trifurcation topology, where the third lineage mainly contains the cell type “MK” and was reported in [12]. c Histograms of all genes’ p-values calculated by the three DE methods in the first lineage. d Histograms of all genes’ p-values calculated by the three DE methods in the second lineage. e Venn plot showing the overlaps of the significant DE genes (BH adjusted- p≤0.01) identified by the three DE methods in lineage 1. PseudotimeDE and tradeSeq share 77.6% (Jaccard index) DE genes. (f) Numbers of enriched gene sets (q<0.25) by GSEA using the p-values in lineage 1 by the three DE methods. Although the DE genes are similar in e, PseudotimeDE yields 270 enriched gene sets, while tradeSeq only yields 9. g Venn plot showing the overlaps of the significant DE genes (BH adjusted- p≤0.01) identified by the three DE methods in lineage 2. Similar to lineage 1 in g, PseudotimeDE and tradeSeq share 77.2% (Jaccard index) DE genes. h Numbers of enriched gene sets (q<0.25) by GSEA using the p-values in lineage 2 by the three DE methods. PseudotimeDE and Monocle3-DE yield hundreds of enriched gene sets, while tradeSeq does not yield any enriched gene sets

Back to article page