RNA-seq experiment validation by qPCR

Question

[Inbalor]

I have analysed data from RNA-seq and I am thinking how I should select genes to validate my RNA-seq experiment by qPCR/nanostring/microarrays. I know that you could use Fold change and P-value threshold, others are using results from GO and pathways analysis... What is your approach to select genes? I am looking forward to hearing from you! Thanks!

Answer 1

[vahapel]

Hi,

Selecting the genes to verify (statistical significance and biological relevance) the transcripts obtained from RNA-Seq or microarray experiment is totally based on the your biological questions.

Of course, you also select some transcripts whose expression are quietly significant according to sequencing experiment. However, sorting the differentially expressed transcripts (DETs) based on the p-value (even corrected p-value) is not good option always for biological relevance, but it should be good for validating RNA-Seq results.

After comprehensive annotation and evaluation (GO, KEGG, InterPro, Pfam, Ka/Ks ration, transcript expansion or depletion, DET analysis, etc...) of the sequencing data, you should reach a conclusion about the topic you have been discussing, then decide the transcript for qPCR experiment. As you mention the qPCR, of course, the selected transcript you did should exhibit a differences (big or little) in its expression.

Answer 2

[cyril-cros]

Vahapel's answer is great. Just let me add a few things.

• If you do RT-qPCR, you will likely want to do relative quantification analysis. You will need a couple reference genes which have the same expression across your samples. Software such as GeNorm / Normfinder is useful.
• You may want to find some gene markers for other experiments: say, compare young/adult samples. Just keep in mind that a sample is usually constituted of quite a few different cell populations and that you get an averaged expression.
• You will need to still have some leftover biological samples for the consistency of your testing.
• The one thing you get with RT-qPCR is a very precise quantification over several orders of magnitude of expression. The problem is you are restricted to maybe a dozen genes per run...
• Validating your experiment is not that necessary I believe; precising an observation is nice on the other hand. Usually, I see articles comparing micro-array and RNASeq data, not RNA Seq and RT-qPCR. Both techniques compare a huge number of genes at the same time...
• Isoforms: if you want to find the precise distribution of an isoform or analyse splicing, well this is possible.

Answer 3

[Antonio R. Franco]

wondering why we give the same or even more value to the RT-PCR than to the direct information provided by the RNA-Seq experiment

Some people think the opposite. Instead using RT-PCR to validate our RNA-Seq analysis, maybe we should be doing the opposite