When you plan a study of gene expression with Agilent microarray, you can select 1-color (single channel) or 2-color (dual channel) design. Which should you apply? GSE27183 is a good example to know about the quality of 1-color and 2-color experimental data.
Advantage of 1-color design.
- Cost effective
- Flexible to design an experiment.
Advantage of 2-color design.
- Maybe more sensitive and to detect changes against reference channel, because the normalization of 1-color data is less solid. See the 1-color data in normalized log2 ratio and signal intensities.
Notice that this advantage is valid only when you directly compare to reference channel within a sample. If you compare ratios between samples, it's no more advantageous. If you apply common reference design, this advantage disappear.
And 2-color design can easily get messy. 4 pairs of 8 samples (GSM671801, GSM671802, GSM671805, GSM671806, GSM671821, GSM671822, GSM671845, GSM671846) looks wrongly labeled "Cy3" and "Cy5." The SSA files of the 2-color data is modified to invert dyes.
Some genes are strongly affected by dye (Cy3 or Cy5) like PPM1G, but mostly such effect is ignorable. So dye swap is not likely to be necessary for all samples. Such genes are detectable only with 2 or 3 pairs of dye swapping samples. It cuts the cost to about a half.
Conclusion
If you’re very keen about slight fluctuation against reference sample, direct comparison with 2-color design will be more powerful. 1-color design is preferable otherwise.
Data Source
Gene Expression Omnibus (GEO)