Table 1.

Clustering algorithms used in benchmarking

	SC3	Seurat	RaceID3	ILoReg	CIDR
Clustering workflow	Feature selection + Distance matrices with three measures + PCA + k-means + CSPA + hierarchical	Feature selection + PCA + graph-based	Feature selection + k-medoids	ICP L times + PCA + hierarchical	Imputation + PCoA + hierarchical
Visualization workflow	None (via Scater R package)	Feature selection + PCA + t-SNE, UMAP etc.	Feature selection + t-SNE or kNN graph	ICP L times + PCA + t-SNE or UMAP	Imputation + PCoA
Method for estimating the number of clusters (k)	Random matrix theory	None (k by default resolution value)	Saturation	Silhouette	Calinski-Harabasz Index
Version	1.12.00	3.0.0	0.1.3	0.1.0 (Git reference ID ‘85196be6’)	0.1.5