Fig. 2: ROGUE use and performance.
From: An entropy-based metric for assessing the purity of single cell populations
a The ROGUE index (reference factor K = 45) decreases monotonically with increasing varied genes in each simulated mixture consisting of two cell types (1:1). The center line indicates the median ROGUE value of n = 50 repeated simulations. The lower and upper hinges represent the 25th and 75th percentiles respectively, and whiskers denote 1.5 times the interquartile range. b The ROGUE values (reference factor K = 45) for the simulated mixtures with cell-type sizes ranging from 1:100 to 1:1. In each mixture, the number of varied genes was 1% of the total gene number (n = 20,000). The center line indicates the median ROUGE value of n = 50 repeated simulations. The lower and upper hinges represent the 25th and 75th percentiles respectively, and whiskers denote 1.5 times the interquartile range. c Pearson correlations of S between the randomly down-sampled datasets (n = 50 runs for each) and the entire datasets (2000 cells) simulated from both NB and ZINB distribution. The center line indicates the median correlation value. The lower and upper hinges represent the 25th and 75th percentiles respectively, and whiskers denote 1.5 times the interquartile range. d Sequencing depth distribution (total UMI counts across cells) for two simulated replicates. The replicate 2 has a sequencing depth ten times that of replicate 1. e The S–E plot of the mixture of replicates 1 and 2 is shown in d. f ROGUE values of n = 100 mixtures versus the silhouette values for every two replicates within individual mixtures. A high silhouette value indicates a substantial difference in sequencing depth between two replicates. g, h The S–E plots and corresponding ROGUE values of 10 cell populations from the PBMC dataset24. i Purity assessment of six human T-cell populations. j Purity evaluation of lung-cancer infiltrating DCs, with each point representing a patient. The center line indicates the median ROUGE value. The lower and upper hinges represent the 25th and 75th percentiles, respectively, and whiskers denote 1.5 times the interquartile range.
