Table 3

The mean square errors (MSE) of estimated gene expression levels (up:down = 9:1) for simulated cDNA microarray data with the R-I plots similar to Figure 3.
Percentage of DEG
 Descriptive Statistics


Method
 Mean
 Minimum
 25% Quantile
 Median
 75% Quantile
 Maximum
1%
 OLS
 0.0370
 0.0116
 0.0258
 0.0323
 0.0410
 2.2092

Huber
 0.0248
 0.0112
 0.0195
 0.0223
 0.0255
 1.7156

Tukey
 0.0238
 0.0109
 0.0191
 0.0217
 0.0247
 1.5648

LOWESS
 0.0375
 0.0119
 0.0251
 0.0321
 0.0419
 2.2162








5%
 OLS
 0.0489
 0.0126
 0.0265
 0.0333
 0.0422
 1.5836

Huber
 0.0324
 0.0110
 0.0198
 0.0228
 0.0263
 1.1765

Tukey
 0.0299
 0.0108
 0.0194
 0.0222
 0.0254
 1.0278

LOWESS
 0.0493
 0.0125
 0.0255
 0.0325
 0.0429
 1.6134








10%
 OLS
 0.0692
 0.0124
 0.0285
 0.0359
 0.0464
 1.6907

Huber
 0.0455
 0.0098
 0.0210
 0.0245
 0.0288
 1.1667

Tukey
 0.0404
 0.0102
 0.0204
 0.0236
 0.0276
 1.0175

LOWESS
 0.0692
 0.0119
 0.0270
 0.0349
 0.0461
 1.6846








20%
 OLS
 0.0961
 0.0137
 0.0324
 0.0428
 0.0570
 1.8614

Huber
 0.0632
 0.0124
 0.0235
 0.0282
 0.0354
 1.2969

Tukey
 0.0547
 0.0127
 0.0225
 0.0266
 0.0329
 1.0525

LOWESS
 0.0950
 0.0154
 0.0325
 0.0431
 0.0580
 1.8834








40%
 OLS
 0.1439
 0.0147
 0.0493
 0.0665
 0.1007
 2.5021

Huber
 0.0960
 0.0134
 0.0330
 0.0446
 0.0673
 1.9988

Tukey
 0.0821
 0.0136
 0.0305
 0.0401
 0.0602
 1.6771

LOWESS
 0.1562
 0.0187
 0.0832
 0.1121
 0.1418
 3.0480








*70%
 OLS
 0.1530
 0.0138
 0.0554
 0.1146
 0.1882
 1.2717

Huber
 0.1040
 0.0121
 0.0366
 0.0791
 0.1267
 0.9153

Tukey
 0.0901
 0.0115
 0.0337
 0.0700
 0.1098
 0.7778

LOWESS
 0.4082
 0.0350
 0.1651
 0.3563
 0.6331
 1.0816

DEG: differentially expressed genes. *: all DEG are up-regulated. OLS: the TW-SLM using the ordinary least squares.

Huber: the robust TW-SLM using Huber's weight function. Tukey: the robust TW-SLM using Tukey's weight function.

Wang et al. BMC Bioinformatics 2005 6:14   doi:10.1186/1471-2105-6-14