Minimum number of replicates for gene co-expression analysis

Question

[4galaxy77]

Hi,

I am looking to see if there is a difference in the number and membership of gene co-expression networks between two different treatments (from RNAseq data). I was going to use WGCNA but I have too few replicates (14/15 compared to the 20 they suggest) for it to work reliably.

Generally is 14/15 replicates too few for any analysis, or are there other programs which would be able to use this number?

Thanks

[GouthamAtla]

They suggest 20 but they say minimum is 15. It also depends on how good are these replicates in terms of variation.

[WouterDeCoster]

Just give it a try with your samples. As Goutham Atla wrote, it depends on how strong the variation is between your replicates. If there rather similar, 15 is probably fine. If they're very different, you might want 50 ;)

Answer 1

[Manoj]

WGCNA is not recommended when sample size is less than 15 WGCNA

You can try getting the correlation coefficient (pearson correlation, etc ) between the genes based on their expression across conditions and construct a network based on a  threshold(say 0.9).