Question: Gene Co-expression analysis with less number of replicates
gravatar for amoltej
4.0 years ago by
amoltej90 wrote:


I have 3 tissue samples with three replicates each (HiSeq data). I want to do gene network co expression analysis on this data. can anybody suggest me best software to do this analysis. I was reading about WGCNA but they recommend it to have at least 20 samples. is there any equally strong alternative for this?

Thank you for support Amol

ADD COMMENTlink modified 21 months ago by IrK50 • written 4.0 years ago by amoltej90

Hi Amol, Did you find the answer to your question? I also have three replicates and I'd like to do gene coexpression analysis, and I can't find alternative to WGCNA. Thanks

ADD REPLYlink written 21 months ago by IrK50

I think the critical point here is not (only) the number of replicates, but rather the number of different condition / samples you have. Given that the original poster only had 3 different samples, I guess it was impossible in his case to build a gene co-expression network. With so few samples in fact, 1 in ~6 genes will correlate by chance and therefore you wouldn't be able to distinguish biologically relevant correlations from random ones.

So, how many conditions do you have?

ADD REPLYlink modified 21 months ago • written 21 months ago by Martombo2.6k

I have 3KOs vs 3WTs (biological replicates), WTs have to be excluded from this kind of analysis that leaves only three KO samples. So I conclude it's not possible to build coexpression network in my case as well. Very sad!!! Sounded like very interesting kind of analysis to learn.

Martombo, thanks for your promt reply, that helped me to conclude that it wouldn't be possible to conduct this type of analysis on a small size.

ADD REPLYlink written 21 months ago by IrK50

I would neither recommend the construction of a network from such low numbers. Co-expression networks are typically built from correlation matrices (but not necessarily); thus, does it make sense to correlate 3 versus 3? The number of false-positive correlations would be uncomfortably high. For future reference, I have a tutorial here: Network plot from expression data in R using igraph

ADD REPLYlink written 21 months ago by Kevin Blighe54k
Please log in to add an answer.


Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.3.0
Traffic: 1075 users visited in the last hour