Figure 1: Overview of the scLVM approach. | Nature Biotechnology

Figure 1: Overview of the scLVM approach.

From: Computational analysis of cell-to-cell heterogeneity in single-cell RNA-sequencing data reveals hidden subpopulations of cells

Figure 1

(a) The observed expression profile of differentiation marker genes (upper panel) is the result of the differentiation process of interest together with the effects of the cell cycle and other confounding sources of variation. After accounting for cell-cycle effects (middle panel), one can uncover gene expression signatures that contribute to the continuous differentiation process more clearly (lower panel). (b) scLVM two-stage procedure. First, in the fitting stage, the cell-to-cell covariance matrix that corresponds to the cell cycle is inferred from the gene expression profiles of genes with cell-cycle annotation (upper panel). The learnt covariance is then used in downstream analyses, including the detection of substructure, the detection of gene-to-gene correlations and the analysis of variance (lower panel). Biol. var., biological variance; Tech. var., technical variance.

Back to article page