Appropriate statistical tools, including normalization, clustering, and identification of differentially expressed genes need to be applied in any microarray experiment.
"Metabolism (01)" and "Transport facilitation (67)" were represented as two major functional classes with differentially expressed genes, reflecting need for uptake and metabolism of two different carbon sources.
Short-read sequencing technology allows identification of differentially expressed genes without the need for prior annotations, and thus is ideal for detecting induced components of the D. virilis immune system that lack homologs in D. melanogaster.
For effective enrichment of ratio differentially expressed genes, the concentration ratio needs to be more than 5-fold.
However, the biochemical function of at least 50 differentially expressed genes remains unknown and needs further investigation.
When testing 20,000 genes with a per-gene alpha value of 0.05 the number of samples needed to detect differentially expressed genes by t-tests with a fold change of 1.2 is 112, whereas only eight samples are needed to statistically have the power to detect a 2-fold difference (http://bioinformatics.mdanderson.org/MicroarraySampleSize/MicroarraySampleSize.aspx).aspx
We estimated that to test 18,500 genes at the 5% significance level and ensure 80% power, 23 samples were needed to detect differentially expressed genes by t-tests with a fold change of at least 1.5 (http://bioinformatics.mdanderson.org/MicroarraySampleSize/).org/MicroarraySampleSize/
Studies using larger numbers of LCLs along with follow-up experiments will be needed to validate differentially expressed genes from this pilot study as true CAC biomarkers; this will become possible in the near future as we deploy RNA-Seq to the entire ClinSeq® cohort.
Leveraging this knowledge to improve the power of genomic tests would provide significant advantages; and FDR weighting methods that combine biological information and knowledge with gene expression measurements to more effectively identify differentially expressed genes represent an area of unmet need.
Validation using both spike-in data and real experimental data proves the method is effective at isolating differentially expressed genes statistically, thereby eliminating the need for arbitrary restrictions such as fold change.
