Skip to main content

Table 3 Evaluating sample size estimates from different methods.

From: A mixture model approach to sample size estimation in two-sample comparative microarray experiments

    No correlation With correlation
π 0 pFDR power True JMB (sd) HZW (sd) PC (sd) PMMP (sd) True JMB (sd) HZW (sd) PC (sd) PMMP (sd)
0.7 0.05 0.6 6 6 (0.4) 11 (0.0) 11 (1.6) 6 (0.4) 6 6 (0.5) 11 (0.0) 11 (1.4) 6 (0.5)
0.7 0.05 0.7 8 8 (0.6) 18 (0.2) 14 (2.8) 7 (1.0) 8 8 (1.0) 18 (0.0) 14 (2.3) 7 (0.9)
0.7 0.05 0.8 11 11 (1.5) 39 (0.7) 20 (4.8) 9 (2.5) 11 11 (2.0) 38 (0.5) 19 (4.1) 9 (2.0)
0.7 0.05 0.9 23 24 (3.4) 146 (2.4) 30 (9.5) 13 (6.5) 23 23 (4.5) 145 (1.6) 29 (7.9) 15 (6.8)
0.7 0.01 0.6 9 9 (0.5) 17 (0.5) 16 (2.7) 9 (0.9) 9 9 (0.7) 17 (0.5) 16 (2.4) 8 (0.8)
0.7 0.01 0.7 11 11 (1.1) 28 (0.5) 21 (4.6) 10 (1.7) 11 11 (1.7) 28 (0.5) 21 (3.8) 10 (1.7)
0.7 0.01 0.8 16 16 (2.4) 60 (1.1) 28 (7.9) 13 (4.5) 16 16 (3.3) 60 (0.8) 27 (6.4) 13 (3.4)
0.7 0.01 0.9 34 37 (6.0) 231 (4.2) 42 (14.5) 18 (10.0) 32 36 (7.5) 229 (2.8) 40 (11.9) 22 (11.3)
0.9 0.05 0.6 9 9 (2.1) 16 (0.5) 24 (7.8) 9 (3.9) 8 9 (2.6) 16 (0.5) 23 (6.6) 9 (3.1)
0.9 0.05 0.7 11 11 (3.5) 27 (0.8) 31 (11.4) 11 (7.1) 10 12 (4.1) 27 (0.8) 30 (9.7) 11 (6.6)
0.9 0.05 0.8 16 16 (5.2) 59 (1.7) 41 (16.7) 15 (11.1) 14 16 (5.9) 58 (1.7) 41 (14.4) 15 (11.8)
0.9 0.05 0.9 34 26 (7.6) 227(6.6) 60 (26.8) 21 (17.9) 29 25 (8.5) 225 (6.7) 59 (23.1) 22 (18.8)
0.9 0.01 0.6 11 12 (3.3) 22 (0.7) 33 (11.8) 12 (6.0) 11 12 (4.1) 22 (0.6) 32 (9.9) 12 (4.9)
0.9 0.01 0.7 14 15 (5.3) 38 (1.2) 43 (16.9) 15 (10.4) 13 16 (6.0) 37 (1.2) 42 (14.4) 15 (9.9)
0.9 0.01 0.8 21 21 (7.4) 82 (2.6) 56 (24.3) 20 (15.6) 19 21 (8.4) 81 (2.7) 55 (20.9) 21 (16.8)
0.9 0.01 0.9 46 35 (10.0) 318 (10.0) 79 (37.2) 27 (24.3) 38 33 (11.3) 316 (11.3) 78 (32.0) 29 (25.1)
0.7 0.05 0.6 6 6 (0.2) 6 (0.0) 10 (1.0) 6 (1.1) 6 6 (0.3) 6 (0.0) 10 (1.1) 5 (0.7)
0.7 0.05 0.7 7 8 (0.7) 8 (0.3) 12 (1.7) 8 (2.4) 7 8 (0.7) 8 (0.1) 12 (1.8) 7 (1.6)
0.7 0.05 0.8 9 11 (1.4) 16 (0.4) 16 (2.9) 10 (4.9) 9 11 (1.5) 16 (0.2) 16 (3.2) 10 (3.8)
0.7 0.05 0.9 14 23 (4.1) 56 (1.2) 24 (5.5) 18 (11.3) 15 24 (5.3) 56 (1.0) 25 (6.1) 14 (7.8)
0.7 0.01 0.6 8 8 (0.6) 8 (0.0) 11 (1.7) 8 (2.2) 8 8 (0.7) 8 (0.0) 14 (1.8) 8 (1.0)
0.7 0.01 0.7 10 11 (1.1) 12 0.1) 14 (2.8) 11 (4.3) 10 11 (1.3) 12 (0.1) 18 (2.8) 10 (3.1)
0.7 0.01 0.8 13 16 (2.5) 23 (0.5) 24 (4.6) 15 (8.4) 15 16 (2.5) 23 (0.3) 24 (5.0) 13 (6.4)
0.7 0.01 0.9 26 36 (7.2) 84 (1.8) 35 (8.4) 27 (17.8) 30 37 (8.8) 83 (1.3) 34 (9.4) 20 (12.0)
0.9 0.05 0.6 8 10 (2.0) 7 (0.4) 21 (8.2) 8 (1.6) 8 10 (2.3) 8 (0.4) 24 (8.5) 9 (3.6)
0.9 0.05 0.7 9 13 (3.3) 11 (0.7) 27 (12.1) 10 (3.9) 9 13 (3.4) 12 (0.7) 32 (12.8) 11 (5.7)
0.9 0.05 0.8 12 18 (5.5) 21 (1.2) 37 (19.0) 13 (8.0) 13 19 (5.0) 23 (1.5) 44 (19.7) 14 (8.7)
0.9 0.05 0.9 24 31 (8.5) 76 (4.6) 54 (30.2) 18 (14.5) 25 31 (7.9) 85 5.4) 65 (32.6) 20 (14.5)
0.9 0.01 0.6 11 13 (3.1) 9 (0.5) 29 (12.6) 11 (2.4) 11 13 (3.6) 10 (0.6) 34 (13.2) 12 (5.6)
0.9 0.01 0.7 13 17 (5.0) 16 (0.6) 38 (18.5) 14 (6.2) 14 19 (5.0) 17 (0.8) 45 (19.6) 15 (8.2)
0.9 0.01 0.8 18 25 (8.1) 28 (1.5) 50 (27.2) 17 (11.8) 19 26 (7.1) 31 (1.9) 60 (29.2) 19 (12.3)
0.9 0.01 0.9 51 43 (11.4) 103 (6.2) 72 (42.8) 23 (19.8) 53 43 (10.9) 115 (7.9) 88 (46.3) 26 (19.7)
  1. True and estimated per group sample sizes for simulated data sets having π0 = 0.7 and π0 = 0.9, and for different pFDR and average power cutoffs. The reported sample size estimate is the average of 50 such estimates rounded off to nearest integer. The standard deviation (sd) was based on the corresponding 50 data sets. Estimates made using the method discussed in this paper are termed JMB in the table (from the author names), while estimates made by the methods discussed by Hu et al. [13], Pounds and Cheng [14] and Pawitan et al. [16] are termed HZW, PC and PMMP respectively.