Using RNAseq to measure gene expression in a tissue when the gene is NOT differentially expressed

Question

[Rduncan]

There is a lot of interest in tissue-specific and differentially expressed genes and a lot of people are using RNA-seq to determine if genes are up-regulated in one tissue vs. another. However, I am interested in a particular gene that, based on qPCR, is highly expressed in a particular aphid tissue (bacteriocyte), but it is not differentially expressed (it is also highly expressed overall in the body and in other tissues, like head and gut). So I'm not interested in using RNA-seq to measure differential expression of this gene, because I don't necessarily expect to find it. I am interested in using RNA-seq to confirm high expression in bacteriocyte (relative to other genes in the same gene family). Is there a way to do this? I know that RNA-seq does not measure absolute expression, but would it be valid to rank some measure of expression (like TPM) of, say, my gene of interest and other members of the same gene family within bacteriocyte? I would like to know if this gene and its orthologs are, or are not, highly expressed in bacteriocyte tissues relative to other genes in the same gene family. In theory, I would like to be able to quantitatively compare across orthologs/different insects, but I would be happy with independent yes/no answers for each insect.

Answer 1

[Devon Ryan]

This is really a worst-case scenario for RNAseq. I would suggest that you just do qPCR, which will be both faster and vastly cheaper.

The issue with trying to do this with RNAseq is that there are GC and other biases that will affect the quantification of one isoform versus another. This can still be done, but you really need to be very very comfortable with the data and the underlying biases. It's much easier to do absolute quantification with qPCR, which will directly give you an answer.

[Rduncan]

Thanks Devon. Wouldn't relative quantification with qPCR be more appropriate though? If I want to know if the gene is highly expressed in bacteriocyte, it seems like I would need to compare it to something like a housekeeping gene. Then I couldn't quantitatively compare across species, but it is more important for me to know within a species if the gene of interest is highly expressed in bacteiocyte. I initially thought about using qPCR, but it would be useful to know how expression of the gene of interest compares to other genes in the gene family, of which there are quite a few (~40). Hence, the idea of using RNAseq instead.

[Devon Ryan]

As long as you know the efficiency of the probes then sure, relative quantification will work fine. Yeah, with 40 genes qPCR isn't so ideal.

Answer 2

[jotan]

Maybe you could consider a different platform like this:

http://www.nanostring.com/applications/technology

[Rduncan]

That is a very interesting idea. Thanks for sharing!

[unksci]

bDNA sm-FISH would be another approach for measuring absolute number of molecules (without requiring much tweaking across samples).

For yes / no: most ISH or antibodies (or even non-quantitive RT-PCR) should do it, given that it was biologically possible to perform cognate control experiments