RNA-seq replicates & experiment design

[sebastion]

Hi folks,

I'm pretty new to NGS analysis & I'm coming at it from the biology side of things so it seems Im already at a disadvantage. That being said this forum is incredible & thanks to everyone who keeps it running.

Im still in the design stages of my project(RNA-sequencing) and I was just wondering if someone can give me some advice on the replicate issue.
This is probably a stupid question but im confused. When talking about the number of replicates needed in an RNA seq experiment-Im confused do they have to be from the same individual? To be more specific if I say wanted to compare gene expression patterns in RNA taken from disease-affected Vs unaffected individuals (same tissue source) and I sequenced say 20 disease and 10 control (or unaffected) individuals -do i need 2 biological replicates (so new RNA extractions) for each of my individuals? or would i just take all individuals in each group together and do they count as replicates?

I hope this question makes sense... I'm very weary of making mistakes at this stage...
thanks in advance to anyone who replies!
Seb

[chadn737]

Wouldn't samples collected from the same individual be a technical replicate, while the samples from separate individuals being the biological reps?

If you have the money and can do it, technical replicates would be great, otherwise give priority to the biological replicates as that will be your greatest source of variation.

[sebastion]

Thanks chadn737 -I think this is where Im getting mixed up(as a result of too much time in cell culture) I thought that if I just sequence the same cDNA library again(so the same original RNA extraction from the same individual) that would be technical whereas biological would be going back to that individual extracting more RNA & repeating the prep ?

[pmiguel]

I don't think there is any fixed meaning to the term "technical replicate" other than "not a full biological replicate".

I am not a statistician. But here is my take:
Your best bet is to consult with whomever you plan to have analyze your data. You will want to explain all the steps of your protocols and focus on areas that may create systematic bias. Then price out your cost/replicate and come up with a plan that gives you the most power in the area that interests you.

Or, just forget about the technical replicates and do biological replicates -- they will capture all the variance of the entire process anyway.

When your cost/replicate is low, it can make sense to attempt to isolate the sources of variance in your experiments with exhaustive technical replicates. But for the most part I would think this is meta-data. Information that you could use to plan better experiments in the future.

--
Phillip

[mbblack]

Thanks chadn737 -I think this is where Im getting mixed up(as a result of too much time in cell culture) I thought that if I just sequence the same cDNA library again(so the same original RNA extraction from the same individual) that would be technical whereas biological would be going back to that individual extracting more RNA & repeating the prep ?

A biological replicate would involve extracting RNA from a different individual organism. Anything done with the same individual donor pool of RNA would be a form of technical replicate.

In other words, by repeatedly sequencing the same individual organism (either by repeatedly sequencing the same RAN prep, or repeatedly sampling RNA from the same individual) you are better characterizing that individuals RNA pool, but you are not capturing any additional information about the biological variation in the population of organisms.