Skip to main content
. 2012 Oct 31;13(3):509–522. doi: 10.1093/biostatistics/kxr033

Table 1.

Simulation study: summary statistics based on 200 replicates

nm # clus ρ Meta-analysis
Intensity approach
iLB iSLB
Enet LB Enet LB
Correlation structure: autoregressive
30 50 0.3 80, 13, 46 70, 10, 44 41, 10, 36 33, 9, 31 32, 15, 20 28, 16, 14
0.7 74, 17, 40 75, 12, 41 29, 14, 20 25, 12, 19 34, 17, 15 24, 17, 12
100 0.3 97, 8, 47 96, 9, 47 31, 5, 40 44, 6, 41 30, 14, 29 29, 14, 29
0.7 93, 10, 43 88, 10, 41 29, 9, 24 32, 10, 20 34, 16, 22 25, 17, 13
100 50 0.3 105, 23, 34 69, 14, 32 36, 23, 21 39, 12, 25 38, 21, 18 27, 22, 10
0.7 79, 25, 19 65, 15, 21 25, 22, 13 37, 17, 14 42, 23, 12 26, 23, 11
100 0.3 122, 19, 40 72, 15, 35 24, 17, 25 41, 12, 20 40, 19, 13 29, 21, 10
0.7 110, 23, 25 71, 15, 23 30, 21, 14 40, 16, 18 48, 24, 18 29, 22, 12
Correlation structure: banded
30 50 0.3 85, 13, 36 70, 9, 38 29, 13, 20 32, 9, 23 38, 17, 19 28, 15, 17
0.7 74, 10, 40 91, 10, 38 35, 13, 27 40, 11, 29 35, 20, 16 29, 21, 15
100 0.3 85, 5, 46 76, 10, 40 40, 11, 23 44, 8, 24 37, 14, 22 25, 18, 20
0.7 65, 7, 45 85, 8, 43 25, 8, 34 44, 10, 31 31, 20, 20 27, 15, 19
100 50 0.3 87, 23, 18 66, 10, 20 32, 20, 12 42, 15, 14 41, 18, 15 29, 19, 9
0.7 98, 26, 26 62, 12, 24 37, 22, 15 40, 13, 17 45, 24, 16 29, 20, 8
100 0.3 124, 25, 24 68, 13, 29 37, 20, 15 43, 16, 16 43, 18, 17 32, 23, 9
0.7 171, 20, 32 72, 18, 28 36, 20, 16 42, 19, 18 51, 21, 19 32, 19, 13
Correlation structure: compound symmetric
30 50 0.3 61, 11, 40 51, 12, 31 34, 13, 23 40, 8, 25 34, 19, 19 31, 18, 20
0.7 69, 12, 27 74, 10, 29 33, 12, 14 33, 11, 14 40, 20, 12 20, 15, 8
100 0.3 84, 7, 43 88, 8, 40 41, 13, 35 44, 10, 36 36, 12, 20 30, 14, 19
0.7 58, 7, 36 70, 8, 38 30, 10, 13 34, 9, 15 30, 16, 18 27, 13, 20
100 50 0.3 101, 26, 22 67, 12, 21 32, 24, 14 41, 10, 17 35, 19, 16 29, 24, 8
0.7 128, 26, 14 62, 13, 19 27, 19, 9 40, 20, 19 43, 24, 23 30, 25, 7
100 0.3 143, 25, 27 75, 12, 30 36, 23, 14 38, 13, 21 38, 17, 15 26, 19, 12
0.7 104, 22, 14 74, 18, 21 28, 18, 10 44, 12, 17 49, 27, 15 26, 22, 9

nm, sample size of each data set; # clus, number of gene clusters; ρ, correlation coefficient parameter; LB, LogitBoost.

In cells with “a, b, c,” a is the number of genes identified, b is the number of true positives, and c is the number of prediction errors per 100 subjects.