Fig. 8
From: The Poisson distribution model fits UMI-based single-cell RNA-sequencing data
The boxplots comparing clustering performance using ARI and purity in different data representations. This demonstrates that DIPD-based matrix D as data representation performs better than the Seurat normalized counts in the Zhengmix4eq (four cell types in equal proportions (3,994 cells and 15,568 genes), (a, d), Zhengmix4uneq (four cell types of unequal proportions as 1:2:4:6 (6,498 cells and 16,443 genes), (b, d), and Zhengmix8eq (eight cell types in equal proportions, (c, f) data sets