LogicNet: probabilistic continuous logics in reconstructing gene regulatory networks

BMC Bioinformatics

Table 6 The LogicNet in comparison with ARACNe, Genie3, Narromi, CN, and GRNTE in reconstructing the directed yeast networks. Two Yeast networks, i.e., Y2 and Y3 with 10 genes and 25 edges (Y2)/22 edges (Y3), are reconstructed by the LogicNet by using 10 gene expression samples from the DREAM3 dataset

Method	TP	FP	TN	FN	TPR	FPR	PPV	ACC	MCC	F-measure
Yeast Network Y2
PC-LogicNet	10	20	45	15	0.40	0.31	0.33	0.61	0.09	0.36
Fuzzy-LogicNet	8	18	47	17	0.32	0.28	0.31	0.61	0.04	0.31
ARACNe	0	1	64	25	0.00	0.02	0.00	0.71	−0.07	–
GENIE3-FR-sqrt	1	5	60	24	0.04	0.08	0.17	0.68	−0.07	0.06
GENIE3-FR-all	1	5	60	24	0.04	0.08	0.17	0.68	−0.07	0.06
Narromi	6	5	60	19	0.24	0.08	0.55	0.73	0.22	0.33
CN	1	5	60	24	0.04	0.08	0.17	0.68	−0.07	0.06
GRNTE	12	17	48	13	0.48	0.26	0.41	0.67	0.21	0.44
Yeast Network Y3
PC-LogicNet	11	16	52	11	0.50	0.24	0.41	0.70	0.25	0.45
Fuzzy-LogicNet	8	19	49	14	0.36	0.28	0.30	0.63	0.08	0.33
ARACNe	1	2	66	21	0.05	0.03	0.33	0.74	0.04	0.08
GENIE3-FR-sqrt	2	2	66	20	0.09	0.03	0.50	0.76	0.13	0.15
GENIE3-FR-all	1	5	63	21	0.05	0.07	0.17	0.71	−0.05	0.07
Narromi	5	7	61	17	0.23	0.10	0.42	0.73	0.16	0.29
CN	6	11	57	16	0.27	0.16	0.35	0.70	0.12	0.31
GRNTE	7	15	53	15	0.32	0.22	0.32	0.67	0.10	0.32

ISSN: 1471-2105