Table 3.
Summary of two-way ANOVA results
| RS | RSEXON | |
| total # of genes (i.e. RSs or RSEXONs) | 563 | 304 |
| # of genes that showed neither sample- nor platform- related variation | 293 | 170 |
| # of genes that showed sample-related variation | 27 | 13 |
| # of genes that showed platform-related variation | 260 | 132 |
| # of genes that showed both sample- and platform- related variation | 20 | 11 |
Two-way ANOVA was conducted for all microarray platforms that had been analyzed with technical replicates (Affymetrix, Amersham, Mergen, ABI). The genes that have valid measurements after normalization and filtering procedure in all technical replicates were included for the analysis. The analysis was conducted on gene expression measurements matched across platforms based on RefSeq ID (RS) and RefSeq Exon (RSEXON). A gene was said to have sample-related variation if the p-value of the sample effect size was significant (p < 0.001), and similarly for platform-related variation.