Table 1.

Accuracies of pseudotime recovery on 5 synthetic datasets and 6 experimental datasets using different pseudotime recovery methods. Pearson correlation coefficient between the predicted and observed times is used to evaluate the result. We compared HopLand with 6 other methods, i.e. Monocle, Wanderlust, Topslam, SCUBA, Wishbone and Diffusion map

Method	Dataset	HopLand	Wanderlust	Monocle	Topslam	SCUBA	Wishbone	Diffusion map
Synthetic data (Zwiessele and Lawrence, 2016)	Synthetic data 1	0.8997	0.9224	0.8158	0.8872	0.9069	0.2534	0.8441
	Synthetic data 2	0.9578	0.9627	0.8202	0.9693	0.9396	0.5089	0.9420
	Synthetic data 3	0.9159	0.7527	0.8322	0.8840	0.7947	0.4704	0.8365
	Synthetic data 4	0.9111	0.8988	0.9095	0.9236	0.8772	0.2249	0.7849
	Synthetic data 5	0.9261	0.9100	0.8988	0.9488	0.9205	0.4498	0.9390
qPCR	GUO2010 (Guo et al., 2010)	0.9230	0.8121	0.5796	0.9297	0.9401	0.5476	0.4949
scRNA-Seq	DENG2014 (Deng et al., 2014)	0.8198	0.8879	0.9177	0.9269	0.9655	0.5115	0.8395
	YAN2013 (Yan et al., 2013)	0.9129	0.8426	0.9421	0.9380	0.9776	0.2876	0.8893
	LPS (Amit et al., 2009)	0.6712	0.7902	0.8899	0.7117	0.5307	0.6064	0.5783
	HSMM (Trapnell et al., 2014)	0.5716	0.2810	0.4560	0.1890	0.0397	0.4850	0.1386
	ES_MEF (Islam et al., 2011)	0.8712	0.8919	0.5166	0.9035	0.8518	0.3952	0.8343
	Average of scores	0.8527	0.8138	0.7799	0.8374	0.7885	0.4310	0.7383
	SD of scores	0.1216	0.1874	0.1756	0.2256	0.2990	0.1256	0.2432

Note: The top three scores in each dataset are in bold.