Figure 2.

The clustering performances of Miscell versus Scanpy, scVI and Seurat

(A–N) t-SNE visualization of feature representations learned by four methods in the PBMC (A–D), Smart-seq (F–I) and Muris dataset (K–N). The corresponding scatterplots of true labels are presented on the right. The clustering metrics of Miscell versus Scanpy, scVI and Seurat on the PBMC (E), Smart-seq (J) and Muris (O) dataset, respectively. Data in the bar plots are represented as mean ± SEM.