Differential expression of housekeeping genes

[richhogg]

Hi all, i am currently using the Tophat/Cufflinks/Cuffmerge/Cuffdiff pipeline through Galaxy to perform differential gene expression analysis on several cancer cell lines. While following the pipeline i used Integrated Genome Viewer to visualize my alignments. While using IGV i attempted to check the number of reads associated with GAPDH and a few other housekeeping genes as these should be expressed at similar levels across all different cell lines. Unfortunately there appears to be highly variable expression of some of these housekeeping genes when viewed in IGV and i am unsure of how this may effect downstream analysis with Cuffdiff and whether the data can be used confidently in differential gene expression analysis.
Apologies if this is a silly question but i am quite new to RNA Seq.

[mbblack]

There are no universal housekeeping genes. Housekeeping genes need to be determined on a case by case basis. GAPDH can indeed be differentially expressed under any number of conditions. GAPDH has been shown to be overexpressed in a number of cancers, for example, and may play a non-glycolytic roll in cell death.

When looking at whole transcriptome changes in expression, what you need is expression estimates from a reference transcriptome, not a housekeeping gene or genes. You need a noncancerous analog to your cancer cell lines for comparison as the reference for differential expression. You also need replicate samples BTW.

If you ultimately do wish to verify some genes by qPCR, you could pick potential housekeeping genes empirically from your comparisons of expression in cancer and noncancer analogs - if a gene shows relatively uniform expression across those, it may make for a reasonable housekeeping gene and you will empirically exclude genes differentially expressed in your particular cancer of interest.

[aggp11]

Depending on the overall library depth differences, using IGV to count and compare read counts might not give you accurate results.

Like mbblack mentioned, you'll need some replication to make any claims on gene expression differences. If you have replicates then I would suggest using a tool like edgeR or DESeq2 to estimate differential gene expression. The following link is a very nice tutorial on how to use DESeq2 for the analysis of differential gene expression:

http://www.bioconductor.org/help/workflows/rnaseqGene/

Also, I recently came across this paper (http://www.sciencedirect.com/science...68952513000899), where the authors compare several published rna-seq datasets (different tissues and experimental conditions) and present their findings on differential expression of several housekeeping genes. This could be a good starting point to select appropriate housekeeping genes.