Skip to main content

Table 2 Cross-dataset validations of the subtypes.

From: A gene signature based method for identifying subtypes and subtype-specific drivers in cancer with an application to medulloblastoma

 

Testing sets

Training sets and self-trained subtypes

Cho73

Northcott90

Kool62

 

A

B

C

(accu.)

A

B

C

(accu.)

A

B

C

(accu.)

Cho73

       

(98.9%)

   

(96.7%)

A

    

7

0

0

 

9

0

0

 

B

    

0

27

0

 

0

13

0

 

C

    

0

1

55

 

0

2

38

 

Northcott90

   

(97.3%)

       

(100%)

A

8

0

0

     

9

0

0

 

B

0

17

2

     

0

15

0

 

C

0

0

46

     

0

0

38

 

Kool62

   

(100%)

   

(100%)

    

A

8

0

0

 

7

0

0

     

B

0

17

0

 

0

28

0

     

C

0

0

48

 

0

0

55

     
  1. This table shows the numbers of cases in each testing set predicted to be of a subtype trained by the training set (first column). For example, the "2" in the testing set Kool62 indicates that 2 cases were self-trained to be of subtype B, but predicted to be of subtype C by the Cho73 dataset. accu.=accuracy.