Table 1.

Data sets considered in this work. The ’genes’ column lists the number of non-zero genes detected in the data set. The peripheral blood mononuclear cell (PBMC) data set from [2] appears twice in this table: ZHENGFILT contains a subset of the full data set ZHENGFULL. See experimental data for more information. ZHENGSIM is a collection of simulated data sets created with the Splatter R package; see the generating synthetic data in the Methods sections for more information

Data set	Cells	Genes	Description	Ground truth clusters	Ref.
Zeisel	3005	4999	mouse neurons	9 (well-separated clusters)	[24]
Paul	2730	3451	mouse myeloid progenitor cells	19 (differentiation trajectory)	[25]
ZhengFull	68579	20387	human PBMCs	11 (some clusters overlap)	[2]
ZhengFilt		5000
10 x Mouse	1.3 million	24015	mouse neurons	39 (algorithmically generated)	[3]
ZhengSim	5000	varies	simulated from human CD19+ B cells	2	using [26]