Fig. 5

Quantifying imputed gene expression in scRNAseq. a, b Correlation between imputed scRNAseq data and bulk RNAseq data across different cell subpopulations in the Cell Type data (a) and the Time Course data (b). Correlation is computed between the mean gene expression measurements averaged across cells with a cell type from scRNAseq and the mean gene expression measurements averaged across sample replicates from bulk RNAseq in the same cell type. For scRNAseq, results are shown for unimputed data (red) and imputed data by different methods that include DrImpute (blue), MAGIC (green), SAVER (pink), scImpute (purple), VIPER with elastic net selection (peach), and VIPER with lasso selection (dark blue). c Gene expression variation across cells in imputed scRNAseq data versus that in the raw data for DEC cells in Cell Type data set. For each gene in turn, the coefficient of variation (CV) across all cells after imputation (y-axis) is computed and plotted against the CV of non-zero cells before imputation (x-axis) for different methods. Each dot represents a gene, and the color of the dot represents the mean of non-zero values