Skip to main content
Fig. 2 | BMC Bioinformatics

Fig. 2

From: DrImpute: imputing dropout events in single cell RNA sequencing data

Fig. 2

DrImpute significantly improved the performance of the existing tools for cell type identification. (a) The average adjusted Rand index (ARI) of 100 repeated runs of pcaR_M (pcaReduce with the M option), pcaR_S (pcaReduce with the S option), SC3, t-SNE/kms (t-SNE followed by k-means), CIDR, scImpute and MAGIC, on seven scRNA-seq datasets. For Zheng and Hrvatin datasets, 1000 cells were randomly sampled from the full datasets and used for the clustering analysis for each method. Black interval represents one plus or minus standard error of the category. Wilcoxon rank sum test was utilized to compare the ARIs from different tools (∗∗: 0.01 ≤ p value < 0.001, ∗∗∗ p value < 0.001). b-c The confusion matrix for (b) iN reprograming using pcaReduce (option S) and (c) mouse preimplantation embryo using t-SNE followed by k-means. Y axis represents ground truth cluster groups reported in the original study and X axis represents predicted groups. Left and right panels, respectively, represent the confusion matrix according to the clustering results without and with preprocessing the scRNA-seq data using DrImpute. The ARI was computed between the original and predicted cell groups

Back to article page