Skip to main content

Table 5 The LogicNet in comparison with PCA-CMI, ARACNe, Genie3, Narromi, CN, and GRNTE in reconstructing the undirected yeast networks (the edge direction is not taken into account in calculating the performance). Yeast networks Y2 and Y3 are reconstructed by using 10 gene expression samples from the DREAM3 dataset. Two types of logics, i.e., the PC and the fuzzy logics, are used separately for reconstructing the GRNs and detecting the logic functions in the LogicNet algorithm. The value of c = α + β is set to 1000. The highest accuracies are indicated in boldface

From: LogicNet: probabilistic continuous logics in reconstructing gene regulatory networks

Method

TP

FP

TN

FN

TPR

FPR

PPV

ACC

MCC

F-measure

Yeast Network Y2

 PC-LogicNet

14

10

10

11

0.56

0.50

0.58

0.53

0.06

0.57

 Fuzzy-LogicNet

14

8

12

11

0.56

0.40

0.64

0.58

0.16

0.60

 PCA-CMI-0.1

5

1

19

20

0.20

0.05

0.83

0.53

0.22

0.32

 PCA-CMI-0.05

5

2

18

20

0.20

0.10

0.71

0.51

0.14

0.31

 ARACNe

1

0

20

24

0.04

0.00

1.00

0.47

0.13

0.08

 GENIE3-FR-sqrt

5

1

19

20

0.20

0.05

0.83

0.53

0.22

0.32

 GENIE3-FR-all

3

3

17

22

0.12

0.15

0.50

0.44

−0.04

0.19

 Narromi

8

2

18

17

0.32

0.10

0.80

0.58

0.26

0.46

 CN

8

5

15

17

0.32

0.25

0.62

0.51

0.08

0.42

 GRNTE

14

9

11

11

0.56

0.45

0.61

0.56

0.11

0.58

Yeast Network Y3

 PC-LogicNet

17

7

16

5

0.77

0.30

0.71

0.73

0.47

0.74

 Fuzzy-LogicNet

14

8

15

8

0.64

0.35

0.64

0.64

0.29

0.64

 PCA-CMI-0.1

14

2

21

8

0.64

0.09

0.88

0.78

0.57

0.74

 PCA-CMI-0.05

15

6

17

7

0.68

0.26

0.71

0.71

0.42

0.70

 ARACNe

3

0

23

19

0.14

0.00

1.00

0.58

0.27

0.24

 GENIE3-FR-sqrt

3

1

22

19

0.14

0.04

0.75

0.56

0.16

0.23

 GENIE3-FR-all

3

2

21

19

0.14

0.09

0.60

0.53

0.08

0.22

 Narromi

6

5

18

16

0.27

0.22

0.55

0.53

0.06

0.36

 CN

17

7

16

5

0.77

0.30

0.71

0.73

0.47

0.74

 GRNTE

10

7

16

12

0.45

0.30

0.59

0.58

0.15

0.51