Fig. 5: ELVAR compares favorably to DA-seq and Milo. | Nature Communications

Fig. 5: ELVAR compares favorably to DA-seq and Milo.

From: Cell-attribute aware community detection improves differential abundance testing from single-cell RNA-Seq data

Fig. 5

a Violin plots displaying the z-statistics of associations between cell-counts (y-axis, mature Cd4t-cells and naïve Cd4t-cells) and age derived from negative binomial regressions for ELVAR, Milo and DAseq, as shown. Gray dashed line indicates the threshold of statistical significance P = 0.05. P-values derive from a one-tailed Wilcoxon rank sum test comparing ELVAR derived z-statistics to those from DA-seq and Milo, respectively. b Violin plots displaying the fraction of captured mature and naïve Cd4t cells (fCC[Naïve+Mat], y-axis) for each biological condition (x-axis) and method. c, d As a, b but for alveolar macrophage M1 and M2 subtypes. e, f As a, b, but for stem-cell and T-regulatory cell fractions in colon tissue in relation to colon cancer progression (N = normal, U = unaffected FAP cases, P = polyp, A = adenoma). g, h As a, b but for neutrophil (Neu) and monocyte-derived dendritic cells (moDC) fractions in relation to Covid-19 disease severity. i, j As a, b but for sensory neurons (SensNeu) and microvillar cells (MicVillar) in relation to Covid-19 smell-loss phenotype. Source data are provided as a Source Data file.

Back to article page