why dont we need technical replicate in case of RNA Seq.?

Question

[neeraj69_sbt]

I want to do RNA seq comparison between two situation in case of plant. is it necessary to use technical replicates in this case. RNA seq is new emerging technology to detect allelic isoform expression as well as DGE in different situation.?

[dariober]

I would like to link my answer to Are different colonies from the same cell line valid biological replicates for DESeq 2? about the distinction (or lack thereof) between what are "technical" and "biological" replicates.

[russhh]

You'll need replicates if you want to publish your results.

[WouterDeCoster]

Important to differentiate between biological and technical replicates.

Answer 1

[WouterDeCoster]

Since the technical reproducibility of RNA-seq is rather good technical replicates are not really necessary. This means that if one skilled person uses the same library prep kit on 5 aliquots of the same sample and creates 5 libraries, and sequences on the same sequencer, the result should be 5 times the same data, approximately. There will be some "Poisson variability", but technical variability is limited.

(Note that if a different kit is used results will be different, obviously.)

But the difference between two plants (of the same species) will be bigger, because of biological variability. So in order to accurately quantify a difference between two groups, you need to assess the intra-group variability, the biological variability. To properly characterize the biological variability you need biological replicates.

The very minimum of replicates would be 3 samples per group. But if you can afford it, definitely do more samples.

[Joe]

There's a school of thought currently that even 3 is no where near enough to get the power that RNAseq needs (https://www.ncbi.nlm.nih.gov/pubmed/27022035 ). 12 replicates would be crazy expensive however, so everyone I know does just do 3.

[WouterDeCoster]

Hmmm valid remark and interesting reference. I would have assumed this also depends on how big this biological variability is. If you have inbred littermate mice you probably need less than having very different human subjects. However, in the paper you linked they work on (I assume well-controlled) yeast samples.

[Joe]

Yeah I can imagine that's fairly reasonably to assume. We work on (in theory) clonal bacterial populations in my lab, but its surprising how varied even that can be (as many of my lab mates, to their annoyance, would attest!).

Answer 2

[GouthamAtla]

I don't think "RNA seq is new emerging technology". You don't need technical replicates but you need biological replicates. For allele isoform expression, you need many samples ( as you need to work with genotypes ) with more sequencing depth than that is required for DGE.