Fig. 8.

scDesign2 guides the choice of cell number in cell clustering, in the case where the total sequencing depth is kept as fixed. scDesign2 generates synthetic 10x Genomics data with twelve cell numbers. Two cell clustering methods—Seurat and SC3—are applied to each synthetic dataset to partition cells into cell clusters. a t-SNE visualization of four synthetic datasets, where cells are labeled by Seurat clusters (top), SC3 clusters (middle), and annotated cell types (bottom). b Two clustering accuracy measures (AMI and ARI) vs. cell number; left: Seurat; right: SC3. In b, the results of the four cell numbers in a are marked as dots and in the top, and the cell number of the real dataset [83] is marked as vertical dashed lines