Table 3

Uni-variate Cox proportional hazards analysis results for right Markov clustering modules

Nejm

GSE4922

GSE3143

GSE18229

GSE1456


Module

Size

Hazard ratio

P-value

Hazard ratio

P-value

Hazard ratio

P-value

Hazard ratio

P-value

Hazard ratio

P-value


2

31

1.30E+00

1.75E-10

1.84E+00

5.84E-05

1.55E+01

6.70E-06

1.12E+00

1.64E-04

2.46E+00

8.76E-05

18

9

9.44E-01

3.99E-09

1.35E+00

1.86E-04

2.61E+00

1.42E-01

7.23E-01

2.59E-03

2.19E+00

1.10E-04

4

17

1.14E+00

1.35E-08

2.46E+00

2.95E-04

2.18E-01

9.23E-01

1.08E+00

3.81E-03

3.38E+00

8.42E-04

13

11

1.02E+00

7.49E-07

1.84E+00

2.02E-04

6.20E-01

3.28E-01

8.66E-01

3.31E-04

2.51E+00

2.37E-04

21

8

1.75E+00

5.21E-05

2.08E+00

6.61E-02

-2.14E+00

2.48E-01

1.31E+00

3.44E-02

3.06E+00

2.93E-02

12

11

-6.59E-01

3.59E-04

-7.18E-01

3.40E-01

3.07E+00

8.43E-02

-8.81E-01

1.26E-02

-3.32E+00

1.55E-02

29

8

7.90E-01

9.21E-04

1.80E+00

5.03E-02

-1.08E-01

4.89E-01

1.25E+00

2.41E-03

3.84E+00

2.58E-03

1

70

-6.68E-01

2.05E-03

2.96E+00

2.95E-02

-1.41E-01

2.62E-01

-1.75E-01

6.38E-01

-2.04E+00

2.76E-01


The eight MCL modules were selected by the trained superpc model when the Nejm data set was used as the training data set. Modules are sorted based on P-values. The sole module that is significant across five breast cancer data sets is in bold.

Wu and Stein Genome Biology 2012 13:R112   doi:10.1186/gb-2012-13-12-r112

Open Data