Skip to main content
Fig. 2 | Genome Biology

Fig. 2

From: TOAST: improving reference-free cell composition estimation by cross-cell type differential analysis

Fig. 2

Results of the simulation study based on gene expression microarray dataset (GSE19830). a The correlations between estimated and true proportions by number of iterations. Left panel of a: boxplot of correlations for four cell types by number of iterations. Right panel of a: mean correlations across four cell types by number of iterations. b The number of overlaps with true cell type-specific (CTS) markers before and after iterations in the top left panel, the root mean squared bias (RMSBias) in the top right panel, the goodness of fit in the bottom left panel, and the root mean squared error (RMSE) in the bottom right panel. Sample size is 100 for b. p values in each panel are obtained using paired t test. Red font indicates being statistical significant. Top 1001–2000 most variable features are selected as initial features. Baseline performance is presented in the “number of iterations = 0” columns in a and “Before” columns in b

Back to article page