Table 1.
estimated numbers of differentially expressed genes. Estimated numbers of differentially expressed genes (numberDEG) for between (left part of table) and within (right part of table) condition analyses. For the between condition analyses the numbers of true positives (number TP) and proportions of true positives among those declared differentially expressed (proportion TP among DE) are also given (middle part of table). The upper part of the table shows result for the Choe data set, the lower for that of the Affy data set. Each row gives results for a particular method on the given data set. Shown are, for each data set and method the mean (in bold) and the minimum and maximum values (in parentheses) obtained over all the analyses of pairs of arrays performed. The BGX method has the highest positive predictive value and sensitivity.
| between | within | |||
| number DEG | number TP | proportion TP among DE | number DEG | |
| Choe data set: | ||||
| BGX (P(dg < 0)) | 705 (507,882) | 674 (499,826) | 0.961 (0.937,0.984) | 4 (0,8) |
| BGX (Efron z-stat) | 1161 (786,1448) | 862 (738,949) | 0.757 (0.647,0.939) | 512 (185,830) |
| EBarrays(RMA) | 773 (519,1006) | 570 (448,686) | 0.754 (0.574,0.863) | 13 (0,80) |
| EBarrays(MAS5) | 347 (240,443) | 2 (1,4) | 0.008 (0.002,0.013) | 742 (266,1141) |
| Wilcoxon | 2940 (2827,3019) | 806 (755,840) | 0.274 (0.267,0.278) | 53 (29,242) |
| Affy data set: | ||||
| BGX (P(dg < 0)) | 28 (4,57) | 13 (6,25) | 0.510 (0.333,0.800) | 2 (0,14) |
| BGX (Efron z-stat) | 574 (318,894) | 26 (23,28) | 0.041 (0.029,0.075) | 736 (365, 942) |
| EBarrays(RMA) | 69 (40,100) | 28 (19,35) | 0.423 (0.310,0.542) | 7 (0,22) |
| EBarrays(MAS5) | 2831 (2610,3134) | 12 (5,20) | 0.004 (0.002,0.007) | 2888 (2592,3115) |
| Wilcoxon | 114 (75,167) | 4 (3,6) | 0.031 (0.018,0.057) | 102 (71,186) |