Skip to main content
Fig. 3 | Genome Biology

Fig. 3

From: Deep sequencing reveals cell-type-specific patterns of single-cell transcriptome variation

Fig. 3

Cell-type patterns of transcriptome variability. a Distribution of expression variability across the transcriptome by cell type. b Enrichment of Gene Ontology categories among variable genes by cell type [49, 52]. Crosses indicate significance (Fishers exact test Bonferroni p < 0.05). Category abbreviations: carb. carbohydrate; dev. development; gen. generation; metab. metabolites; prec. precursor; proc. process. c, d Partial correlation of F-statistic across species, controlling for gene expression level, for cortical (c) and hippocampal (d) pyramidal neurons. Axes are a measure of variation, controlled for gene expression level (see Materials and methods for details). rho indicates the partial correlation coefficient. P values are from a two-sided t-test of association. Marginal histograms are shown overlaid with a normal curve. e Distribution of expression values by cell type for selected genes demonstrating highly variable expression by the outlier-sum statistic in one cell type and as following a normal distribution across cells in another cell type. Histograms of genes identified as highly variable in a given cell type are colored green; those of genes identified as normally expressed in a given cell type are colored orange. f Contingency table of gene categorization as hypervariable and as fast decaying [34]. Sample sizes and abbreviations: brown adipocyte (n = 13, Adi); cardiomyocyte (n = 19, Car); pyramidal neuron, cortex (mouse n = 19, rat n = 8, Pyc); pyramidal neuron, hippocampus (mouse n = 18, rat n = 8, Pyh); serotonergic neuron, dorsal raphe (n = 22, Ser); all single cells (n = 91, All); 10 pg dilution controls (n = 12, 10p); 50 pg dilution controls (n = 9, 50p); 100 pg dilution controls (n = 9, 100p)

Back to article page