Table 2.

Source network structures and synthetic datasets

Network name		TF-gene networks	Gene No.	Edge No.	Network density	Data generator	Data type	Ref.
GNW100	GNW100_1	DREAM4 in Silico size 100	100	176	0.0356	The TF-gene reference networks were subsets of source networks in GNW. In each dataset, 1/3 genes were randomly selected and perturbed. Each experiment was sampled at 21 time points. 3 replicates were generated by adding different amount of noises. The noises are simulated by GNW. All the parameter settings were defaults in GNW.	Time-series data with multifactorial perturbation	[22, 24]
	GNW100_2		100	249	0.0503
	GNW100_3		100	195	0.0394
	GNW100_4		100	211	0.0426
	GNW100_5		100	193	0.0390
GNW500	GNW500_1	E.coli	500	1365	0.0109
	GNW500_2		500	867	0.0069
	GNW500_3		500	1107	0.0089
	GNW500_4		500	947	0.0076
	GNW500_5		500	1272	0.0102
GNW1000	GNW1000_1	E.coli	1000	2337	0.0047
	GNW1000_2		1000	2455	0.0049
	GNW1000_3		1000	2089	0.0042
	GNW1000_4		1000	2171	0.0043
	GNW1000_5		1000	2249	0.0045
GNW2000	GNW2000_1	Yeast	2000	4738	0.0024
	GNW2000_2		2000	4467	0.0022
	GNW2000_3		2000	5055	0.0025
	GNW2000_4		2000	5283	0.0026
	GNW2000_5		2000	4817	0.0024
GNW3000	GNW3000_1	Yeast	3000	7515	0.0017
	GNW3000_2		3000	7998	0.0018
	GNW3000_3		3000	7626	0.0017
	GNW3000_4		3000	8075	0.0018
	GNW3000_5		3000	7333	0.0016

A number of directed network structures were generated from source networks provided by GNW. The network names, gene and edge numbers for each structure are listed in the table. Network density is defined as the true edges divided by all possible edges. The network structures were used to simulate the time-series datasets using GNW