Skip to main content

Table 2 Simulation study results for model CUU

From: Two-way learning with one-way supervision for gene expression data

 

G=2,…,10

G=g known

Case

Average G

Most chosen model

Average ARI

Average ARI

CUC, g known=2

    

Low var, good cluster sep

2.5 (0.8)

CUU

0.955 (0.083)

1.0 (0.0)

Mid var, good cluster sep

2.4 (0.8)

CUU

0.955 (0.094)

1.0 (0.0)

High var, good cluster sep

4.0 (1.0)

CUC

0.708 (0.106)

1.0 (0.0)

High var, close clusters

6.4 (1.2)

CUC

0.502 (0.135)

1.0 (0.0)

CUU, g known=2

    

Low var, good cluster sep

2.4 (0.7)

CUU

0.969 (0.072)

1.0 (0.0)

Mid var, good cluster sep

2.5 (0.8)

CUU

0.964 (0.071)

1.0 (0.0)

High var, good cluster sep

4.0 (1.0)

CUC

0.705 (0.103)

1.0 (0.0)

High var, close clusters

6.4 (1.3)

CUC

0.485 (0.138)

1.0 (0.0)

CUC, g known=3

    

Low var, good cluster sep

3.5 (0.7)

CUU

0.981 (0.039)

1.0 (0.0)

Mid var, good cluster sep

3.4 (0.7)

CUU

0.984 (0.033)

1.0 (0.0)

High var, good cluster sep

5.1 (1.0)

CUC

0.864 (0.081)

1.0 (0.0)

High var, close clusters

8.8 (1.1)

CCC

0.601 (0.066)

1.0 (0.0)

CUU, g known=3

    

Low var, good cluster sep

3.5 (0.6)

CUU

0.984 (0.028)

1.0 (0.0)

Mid var, good cluster sep

3.4 (0.7)

CUU

0.975 (0.050)

1.0 (0.0)

High var, good cluster sep

5.0 (1.1)

CUC

0.866 (0.079)

1.0 (0.0)

High var, close clusters

8.8 (1.0)

CUC

0.590 (0.070)

1.0 (0.0)

CUC, g known=4

    

Low var, good cluster sep

4.4 (0.7)

CUU

0.989 (0.254)

1.0 (0.0)

Mid var, good cluster sep

4.3 (0.5)

CUU

0.992 (0.020)

1.0 (0.0)

High var, good cluster sep

6.2 (1.0)

CUC

0.887 (0.048)

1.0 (0.0)

High var, close clusters

9.7 (0.5)

CUC

0.658 (0.045)

1.0 (0.0)

CUU, g known=4

    

Low var, good cluster sep

4.4 (0.9)

CUU

0.989 (0.031)

1.0 (0.0)

Mid var, good cluster sep

4.4 (0.7)

CUU

0.989 (0.024)

1.0 (0.0)

High var, good cluster sep

4.6 (0.8)

CUU

0.970 (0.048)

1.0 (0.0)

High var, close clusters

9.8 (0.5)

CUC

0.653 (0.046)

1.0 (0.0)

  1. Average ARI, most frequently chosen model, and the number of observation clusters selected for the CUU and CUC models using simulated data with low, medium, and high variance (var) with good cluster separation (sep), and high variance with relatively close clusters when fitting G=2,…,10 observation clusters using 100 data sets and 20 random starts. The last column presents the ARI when fixing G=g known, where g known represents the number of observation clusters the data was generated from. Values in brackets represent the respective standard deviation