Fig. 6: ELVAR is robust to batch correction and false positives in cell-annotation.

a Violin plots of negative binomial regression z-statistics of association of cell-type counts with age for mature and naïve lung-tissue Cd4t cells and for 3 different scenarios: None = ELVAR was run with no batch (sample) correction, Harmony = ELVAR was run data batch-corrected with Harmony, Seurat: ELVAR was run on data batch-corrected with Seurat. Each violin plot contains the values for 100 distinct ELVAR runs. The P-values are derived from two-tailed Wilcoxon rank sum tests comparing “None” to either “Harmony” or “Seurat”. Horizontal dashed lines indicate the P = 0.05 significance level. b As a but for nasopharyngeal neutrophils and monocyte-derived dendritic cells (moDC) cell counts changing with Covid-19 disease severity (mild vs critical). c As a but for microvillar and sensory olfactory neurons in the olfactory epithelium (OE) in relation to their counts changing with long-term smell loss in Covid-19 patients. d As a but for colon enterocyte stem-cell and differentiated cell counts changing with cancer stage progression. e Left panel: Sensitivity to detect a significant (NBR P-value < 0.05) change in the abundance of naïve and mature Cd4T-cells with age in the lung tissue of mice under different rates of false positives (FPR, x-axis). Sensitivity was estimated over 100 distinct runs. Middle left panel: As left panel, but for detecting a significant change in the abundance of neutrophils and monocyte-derived dendritic cells (moDC) with Covid-19 disease severity. Middle right panel: As other panels, but for detecting a significant change in the abundance of olfactory sensory neurons and microvillar cells with Covid-19 smell-loss phenotype. Right panel: as other panels, but for detecting a significant change in the abundance of enterocyte stem and differentiated cells with colorectal adenoma progression. Source data are provided as a Source Data file.