Is it necessary to confirm the RNA-seq results by quantitative RT-PCR?

Question

[parviz Heidari]

What is the theory behind RNA-seq confirmation?

Is it necessary to confirm the RNA-seq results by quantitative RT-PCR?

[seta]

RNA-seq results come from computational prediction and you need to experimentally validate your results, of course, all differentially expressed genes may not be confirmed by qRT-PCR.

[ablanchetcohen]

Well, that would depend on how well the computational analysis is done. :)

The computational predictions are highly accurate, when done properly by a competent bioinformatician, who understands the algorithms used, and is able to explain when the results are unreliable.

You are right that it is easy to draw the wrong conclusions when the analysis is done by someone with no understanding of the algorithms, which I see all too often.

Cuffdiff is probably the greatest culprit for skewing RNA-Seq results, especially when the wrong parameters are given. The default parameters for Cuffdiff will inevitably output incorrect results for short RNAs.

Answer 1

[dariober]

In my experience/opinion (I don't have a reference at hand) qPCR and RNA-Seq correlate very well, meaning that confirming gene expression from RNA-Seq with qPCR on the same RNA sample is probably unnecessary (by the way, is there any evidence that qPCR is better than RNA-Seq?)

However, since RNA-Seq experiments are often done on small sample sizes (2-4), it makes sense to apply qPCR on many more samples to refine the findings from RNA-Seq, for one or few genes of interest, obviously.

[jotan]

is there any evidence that qPCR is better than RNA-Seq?

I've often wondered the same thing too.

However, I frequently use qPCR for validation and reproducibility. If you have 3-4 robust RNA-seq experimental replicates, qPCR is probably no longer necessary.

Answer 2

[cyril-cros]

RT-qPCR is often used to confirm experiments, but may not be necessary.

The main difference are that RNASeq is massively parallel but has difficulties with low coverage genes (you lose your statistical power). RT-qPCR will be much more precise for these weakly expressed genes.

Also, you design your primers with RT-qPCR which means it might be easier to study isoforms (just be sure your primers don't overlap, or there will be some competition effects).