Skip to main content
. 2018 Sep 8;35(8):1269–1277. doi: 10.1093/bioinformatics/bty793

Table 1.

Major characteristics of the 12 benchmarking datasets

#Estimated clusters
Organism #Single cells #True clusters SC3 CIDR Seurat t-SNE + k-means SAFE-clustering Ref
Baron_human1 Human 1937 14 23 3 12 11 11 Baron et al., 2016
Baron_human2 Human 1724 14 23 9 10 6 5 Baron et al., 2016
Baron_human3 Human 3605 14 37 5 12 7 14 Baron et al., 2016
Baron_human4 Human 1303 14 19 3 9 4 8 Baron et al., 2016
Baron_mouse1 Mouse 822 13 18 13 9 9 9 Baron et al., 2016
Biase Mouse 49 3 3 5 3 3 4 Biase et al., 2014
Darmanis Human 420 8 11 7 5 5 5 Darmanis et al., 2015
Ting Mouse 187 7 13 10 5 6 9 Ting et al., 2014
Yan Human 90 7 5 5 3 3 3 Yan et al., 2013
Zeisel Mouse 3005 9 32 5 13 6 13 Zeisel et al., 2015
simple case PBMC mixture Human 28 733 3 3 3 16 4 3 Zheng et al., 2017
challenging case PBMC mixture Human 32 695 3 2 10 14 3 3 Zheng et al., 2017

Note. The characteristics include organism origin, number of single cells, the numbers of true and estimated clusters by SAFE-clustering and four individual methods, as well as references.