Fig. 2: Recursion-to-completion in real datasets.
From: Anti-correlated feature selection prevents false discovery of subpopulations in scRNAseq
a A schematic of sub-clustering is shown in the form of UMAP projections of the original dataset (left panel), and a sub-clustering iteration of a population found in the first round of feature selection and clustering (right panel). b, c In real datasets of varying technologies, status quo algorithms fail the recursion-to-completion problem while the anti-correlation-based approach prevented recursion-to-completion. Recursive clustering plots where each point indicates a cluster at a given recursive clustering recursion-depth as denoted in successive rings and color. d Boxplots of the mean recursion depth for each of the final sub-clusters for each noted method (1-way ANOVA with 2-sided TukeyHSD poshoc). e Boxplots of the total number of groups obtained through iterative sub-clustering (1-way ANOVA with 2-sided TukeyHSD poshoc). Boxplots show lines that extend to minimum and maximum, with the box bounds from 25th to 75th percentile, and center denoting the median. (d, e: n = 4 datasets). Exact p-values for all pairwise comparisons are availabe in Source Data file. Source data are provided as a Source Data file.
