Table 1.
An overview of the results for the two spike-in experiments
| (ROC50 scores) | HG_U95A | HG_U133A | ||||
|---|---|---|---|---|---|---|
| Method name | fold chg. single repl. | SAM all | fold chg. all | fold chg. single repl. | SAM all | fold chg. all |
| MAS 5.0 | 0.00 | 0.25 | 0.01 | 0.01 | 0.45 | 0.02 |
| PLIER | 0.00 | 0.49 | 0.01 | 0.04 | 0.61 | 0.10 |
| VSN | 0.62 | 0.42 | 0.77 | 0.65 | 0.32 | 0.70 |
| PDNN | 0.63 | 0.45 | 0.73 | 0.59 | 0.70 | 0.71 |
| RMA_NBG | 0.64 | 0.53 | 0.78 | 0.69 | 0.66 | 0.77 |
| RMA | 0.47 | 0.27 | 0.75 | 0.68 | 0.70 | 0.77 |
| GCRMA | 0.71 | 0.68 | 0.84 | 0.65 | 0.57 | 0.72 |
| RDN | 0.80 | 0.66 (0.76) | 0.89 | 0.75 | 0.77 | 0.80 |
| RDN [qq] | 0.67 | 0.57 | 0.80 | 0.71 | 0.65 | 0.77 |
| RDN [qq, loc] | 0.70 | 0.65 | 0.83 | 0.72 | 0.69 | 0.78 |
| RDN [qq, hyb] | 0.76 | 0.74 | 0.87 | 0.75 | 0.75 | 0.80 |
| RDN [qq, hyb, amp] | 0.76 | 0.73 | 0.87 | 0.75 | 0.75 | 0.80 |
| RDN [qq, hyb, loc] | 0.80 | 0.68 | 0.88 | 0.76 | 0.77 | 0.80 |
See Additional file 1, Table S1 for a description of the methods. Both fold change and the SAM test statistic were used to determine ROC50 scores (see Methods section). The 'single repl.' score is determined by applying the normalization and scoring methods to each of the three replicate array-sets individually and averaging the results, while the 'all' score is obtained by using all replicates simultaneously. The lower 5 rows describe versions of RDN which use a subset of quantile-quantile normalization ('qq'), array location ('loc'), hybridization ('hyb') and amplification ('amp') correction to determine their individual influence.