Identifying time-delayed gene regulatory networks via an evolvable hierarchical recurrent neural network

Table 3 The effect of network size on GRN reconstruction in case of linearity between the genes. Results are the average of the accuracy in terms of the Delay and Effect criterion for GRN reconstruction of 10 different randomly generated synthetic networks

	Methods	P	TP	FP	FN	Precision	Recall	F-score	Nominal p-value	Adjusted p-value
Delay	HRNN	5	3.9	9.3	6.0	0.32	0.40	0.35
		10	7.5	20.1	11.8	0.28	0.39	0.32
		20	11.6	43.9	28.5	0.21	0.29	0.24
		30	19.1	64.9	39.2	0.22	0.33	0.26
Delay	HCC-CLINDE	5	3.2	3.1	6.7	0.52	0.33	0.404	0.52	1
		10	7.7	6.2	11.6	0.57	0.40	0.47	0.06	0.24
		20	11.3	19.1	28.8	0.38	0.28	0.32	0.22	0.88
		30	17.0	20.5	41.3	0.46	0.29	0.35	0.23	0.92
Effect	HRNN	5	4.5	6.3	6.3	0.44	0.43	0.43
		10	8.9	16.2	10.7	0.36	0.46	0.40
		20	13.0	38.7	27.7	0.25	0.32	0.28
		30	20.2	60.2	38.6	0.25	0.346	0.29
Effect	HCC-CLINDE	5	4.0	2.3	5.9	0.67	0.42	0.51	0.32	1
		10	7.9	6.0	11.4	0.58	0.41	0.48	0.33	1
		20	11.5	18.9	28.6	0.39	0.29	0.33	0.44	1
		30	17.4	20.1	40.9	0.47	0.29	0.36	0.31	1

The p-values are for how the Delay and Effect F-scores of HCC-CLINDE method compare with HRNN. P is the number of genes for each of the networks. The variance of the noise is equal to 1

ISSN: 1756-0381